LOGICAL POSITIVISMEditedbyA.J.AYER TheFree Press,NewYork
-iiiCopyright1959byTheFreePress,aCorporation
PrintedintheUnitedStatesofAmerica
Allrightsreserved.Nopartofthisbookmaybereproducedor
utilizedinanyformorbyanymeans,electronicormechanical,
photocopying,recording,orbyanyinformationstorageand
retrievalsystem,withoutpermissioninwritingfromthePublisher.
Collier-MacmillanCanada,Ltd.,Toronto,Ontario
LibraryofCongressCatalogCardNumber:58-6467 SecondprintingJuly1966
-iv-
PrefaceLogicalPositivismisthesecondinaseriesofbookswhichwillmake
availabletothegeneralpublicsomeofthemostinterestingworkof
philosophersofverydiverseviewpoints.Eachvolumewilldealwith
oneor,insomecases,withtwo,philosophical"schools"or
"movements."Itisfortunatethatphilosophersarerarelyunitedby
thekindofcommonpurposewhichinspirespoliticalorreligious
"movements."Nevertheless,itisfrequentlyhelpfultoconsiderthe
workofdifferentwritersaccordingtothesimilaritiesintheiraim
andcontentandthisisthepolicywhichhasbeenadoptedin
designingtheLibraryofPhilosophicalMovements.
Thisvolumepresents,forthefirsttimeinEnglish,manyofthe
mostinfluentialpapersbyleadingmembersoftheViennaCircle.
Theseandotherarticlescontainauthoritativeexpositionsofthe
doctrinesmostcommonlyassociatedwithlogicalpositivism.
However,forreasonsexplainedbyProfessorAyerinhis
introduction,severalpieceswhichcannotberegardedas
expositionsordefensesoflogicalpositivismhavealsobeen
included.Thescopeofthebibliography,too,isbroaderthanthe
titleofthebookwouldsuggest.Novolumedealingwithotherforms
ofanalyticphilosophyiscontemplatedinthisseriesanditwas therefore
thoughtdesirabletolistthemostimportantbooksand
articlesdealingwithalltypesofanalyticphilosophyandnotonly
withlogicalpositivism.
Iwishtoexpressmygratitudetoallthetranslatorswhogenerously
contributedtheirlabor,toMarvinZimmerman,JamesBayley,Irving
Saltzmann,andanumberofmystudentsforhelpingtocompilethe
bibliography,andtoLeonSatinoff,MaxwellGrober,andJos
Huerta-Jourdaforpreparingtheindex.Specialthanksaredueto
ProfessorsCarnapandHempelforsupplyingnoteswhichindicate
theirpresentpositionontheissuesdealtwithintheirpapers.
ProfessorCarnapwasalsokindenoughtoassistinthetranslation
ofhisownarticles. PAULEDWARDS -v[Thispageintentionallyleftblank.]
-vi-
Contents
PREFACE EDITOR'SINTRODUCTION
LogicalAtomism
v 3 31
Philosophy,MetaphysicsandMeaningMoritzSchlick"THETURNINGPOINTIN
PHILOSOPHY" RudolfCarnap"THEELIMINATION OF
METAPHYSICSTHROUGHLOGICAL ANALYSISOFLANGUAGE"
MoritzSchlick"POSITIVISMANDREALISM" CarlG.Hempel"THEEMPIRICIST
CRITERION OFMEANING" RudolfCarnap"THEOLDANDTHENEW LOGIC"
HansHahn"LOGIC,MATHEMATICSAND KNOWLEDGEOFNATURE" -vii-
BertrandRussell"LOGICALATOMISM"
53 60 82 108
LogicandMathematics
133 147
KnowledgeandTruth
RudolfCarnap"PSYCHOLOGYINPHYSICAL LANGUAGE"
OttoNeurath"PROTOCOLSENTENCES" MoritzSchlick"THEFOUNDATIONOF
KNOWLEDGE"
165 199 209
EthicsandSociology
A.J.Ayer"VERIFICATION AND EXPERIENCE"
MoritzSchlick"WHATISTHEAIMOF ETHICS?"
C.L.Stevenson"THEEMOTIVEMEANINGOF ETHICALTERMS"
OttoNeurath"SOCIOLOGYAND PHYSICALISM"
228
247 264 282 321 327 345 381 447
AnalyticalPhilosophy
FrankD.Ramsey"PHILOSOPHY" GilbertRyle "PHILOSOPHICALARGUMENTS"
FriedrichWaismann"HOWISEE PHILOSOPHY"
BIBLIOGRAPHYOFLOGICALPOSITIVISM INDEX -viii-
LOGICALPOSITIVISM-1[Thispageintentionallyleftblank.] -2-
Editor'sIntroduction
I.HISTORYOFTHELOGICALPOSITIVIST
MOVEMENTTHETERM"LogicalPositivism"wascoinedsomethirtyyearsagoto
characterize thestandpointofagroupofphilosophers,scientists
andmathematicianswhogavethemselvesthenameoftheVienna
Circle.Sincethattimeitsreferencehasbeenextendedtocover
otherformsofanalyticalphilosophysothatdisciplesofBertrand
Russell,G.E.MooreorLudwigWittgensteinatCambridge,or
membersofthecontemporaryOxfordmovementoflinguistic
analysismayalsofindthemselvesdescribedaslogicalpositivists.
Thiswiderusageisespeciallyfavoredbythosewhoarehostileto
thewholemoderndevelopmentofphilosophyasananalytical
ratherthanaspeculativeenquiry.Theywishtotaralltheir
adversarieswithasinglebrush.Thisisirritatingtotheanalysts
themselveswhoarerathermoresensitivetotheirdifferencesthey
wouldpreferthattheappellationof"logicalpositivist"bereserved
forthosewhosharethespecialoutlookoftheViennaCircle.In
compilingthisanthology,Ihavenotbeenquitesostrict.Ihave
drawnmainlyonthewritingsofthemembersoftheViennaCircle,
orofthosewhostandclosesttothem,butIhavealsoincluded
severalpieceswhichfalloutsidethisrange.Theyareall,insome
sense,analyticalbutthescopeofwhatIregardasanalytical
philosophyiswide.Itallowsforseriousdisagreement,notonlyover
technicalniceties,butonmajorpointsofdoctrine,includingthe
methodandpurposeofanalysisitself.
TheViennaCirclecameintobeingintheearly1920'swhenMoritz
Schlick,aroundwhomitcentered,arrivedfromKieltobecome
professorofphilosophyattheUniversityofVienna.Onthe
philosophicalsideitsleadingmembers,besidesSchlickhimself,
wereRudolfCarnap,OttoNeurath,HerbertFeigl,Friedrich
Waismann,EdgarZilselandVictorKraftonthescientificand
mathematicalside,PhilippFrank,KarlMenger,KurtGdelandHans
Hahn.Atthebeginning,itwasmoreofaclubthananorganized
movement.Findingthattheyhadacommoninterestin,anda
-3similarapproachto,acertainsetofproblems,itsmembersmet
regularlytodiscussthem.Thesemeetingscontinuedthroughoutthe
lifeoftheCirclebuttheycametobesupplementedbyother
activitieswhichtransformedtheclubintosomethingmorenearly
resemblingapoliticalparty.Thisprocessbeganin1929withthe
publicationofamanifestoentitled"Wissenschaftliche
Weltauffassung,DerWienerKreis"--The ViennaCircleItsScientific
Outlook-whichgaveabriefaccountofthephilosophicalpositionof
thegroupandareviewoftheproblemsinthephilosophyof
mathematicsandofthephysicalandsocialsciencesthattheywere
chieflyconcernedtosolve.Thispamphlet,whichwaswrittenby
Carnap,NeurathandHahn,isalsoofinterestasshowinghowthe Circle
situateditselfinthehistoryofphilosophy.Afterclaimingthat theywere
developingaViennesetraditionwhichhadfloweredat
theendofthenineteenthcenturyintheworkofsuchmenasthe
physicistsErnstMachandLudwigBoltzmann,and,inspiteofhis
theologicalinterests,thephilosopherFranzBrentano,theauthors
setoutalistofthosewhomtheyregardedastheirmain
precursors.AsempiricistsandpositiviststheynamedHume,the
philosophersoftheenlightenment,Comte,Mill,Avenariusand
Machasphilosophersofscience,Helmholtz,Riemann,Mach,
Poincar,Enriques,Duhem,BoltzmannandEinsteinaspureand
appliedlogicians,Leibniz,Peano,Frege,Schrder,Russell,
WhiteheadandWittgensteinasaxiomatists,Pasch,Peano,Vailati,
PieriandHilbertandasmoralistsandsociologistsofapositivistic
temper,Epicurus,Hume,Bentham,Mill,Comte,Spencer,
Feuerbach,Marx,Mller-Lyer,Popper-LynkeusandtheelderCarl
Menger.Thislistissurprisinglycomprehensive,butitmustbe
rememberedthatinmostcasesitisonlyaquestionofaspecial
aspectoftheauthor'sworks.ThusLeibnizisincludedforhislogic,
notforhismetaphysicsKarlMarxisincludedneitherforhislogic
norhismetaphysicsbutforhisscientificapproachtohistory.Ifwe
excludecontemporariesfromthelist,thosewhostandclosestto
theViennaCircleintheirgeneraloutlookareHumeandMach.Itis
indeedremarkable howmuchofthedoctrinethatisnowthoughtto
beespeciallycharacteristicoflogicalpositivismwasalreadystated,
oratleastforeshadowed,byHume.
Amongcontemporaries,Einstein,Russell,andWittgensteinare
singledoutbytheauthorsofthepamphletfortheirkinshiptothe
ViennaCircleandtheextentoftheirinfluenceuponit.Wittgenstein,
indeed,stoodtotheViennaCircleinaspecialrelation.Havingbeen
apupilofRussell'satCambridge beforethefirstworldwarhe
returnedtoViennaandwastherewhenhisLogischPhilosophischeAbhandlung
-4Abhandlungwaspublishedin1921.Thisfamousbook,whichis
betterknownasTractatusLogico-Philosophicus,thetitlegiventoits
Englishtranslation,hadanenormouseffectuponthepositivist
movement,bothinViennaandelsewhere.Itwouldnotbequite
correcttosaythattheViennaCircledrewitsinspirationfromit.
Schlickhimself,inhisbookonthetheoryofknowledge,Allgemeine
Erkenntnislehre ,ofwhichthefirsteditionappearedin1918,had
independentlyarrivedatasimilarconceptionofphilosophyand
thereisahintofmysticismintheTractatuswhichsomemembers
oftheCircle,especiallyNeurath,founddisquietingbutasawhole
theyacceptedit,anditstoodoutasthemostpowerfuland
exciting,thoughnotindeedthemostlucid,expositionoftheirpoint
ofview.WittgensteindidnotofficiallyadheretotheCirclebuthe
maintainedclosepersonalrelationsatleastwithSchlickand
Waismannwhomhecontinuedtoinfluenceevenafterhisdeparture
forCambridgein1929.InCambridge,wherehetaughtuntil1947,
fouryearsbeforehisdeath,heexercisedanalmostdespoticsway
overhispupils,andthoughhepublishednothingduringtheseyears
exceptoneshortarticlehisinfluencewasstrongly,ifinmostcases
indirectly,feltbyalmostalltheyoungergenerationofBritish
philosophers.Hehimselfmodifiedtherigorsofhisearlypositivism
toanextentthatcanbemeasuredbycomparingtheTractatuswith
hisposthumouslypublishedPhilosophicalInvestigationsanditisto
hisinfluence,combinedwiththatofMoore,thatonemaylargely
attributethepreoccupationofcontemporaryBritishphilosophers
withtheeverydayusesoflanguage,andtheirtendencytodealwith
philosophicalquestionsinanunsystematic,illustrativeway,in
contrasttothemorerigorouswould-be scientificmethodwhichwas
favoredbytheViennaCircle.Thisisonereasonwhytheyarenot
happytobedescribedasLogicalPositivists.ButIshallhavemore
tosayaboutthesealternative conceptionsofanalysislateron.
Itwasin1929alsothattheViennaCircleorganizeditsfirst
internationalcongress.ItwasheldatPragueandwasfollowedat
intervalsthroughoutthethirtiesbyfurthercongressesat
Knigsberg,Copenhagen,Prague,ParisandCambridge.These
meetingsfurtheredtheambitionoftheCircletodevelopLogical
Positivismasaninternationalmovement.Ithadformedanearly
alliancewiththeso-calledBerlinschoolofwhichHansReichenbach,
RichardvonMises,KurtGrellingandatalaterdateCarlHempel
weretheleadingmembers.Thecongresseshelpedittomake
contactalsowithScandinavianphilosopherssuchasEinoKaila,Arne
Naess,ke Petzll,JoergenJoergensen,andtheUppsalaschoolof
empiricists,withtheDutch
-5grouparoundthephilosopherMannourywhopursuedwhatthey
calledthestudyofSignifics,withtheMnstergroupoflogicians
underHeinrichScholtz,withAmericansympathizerssuchasNagel,
CharlesMorrisandQuine,andwithBritishanalystsofvarious
shadesofopinion,suchasSusanStebbing,GilbertRyle,R.B.
Braithwaite,JohnWisdomandmyself.ThebrilliantCambridge
philosopherF.P.Ramseywasmarkedasanadherent,buthedied
in1930attheearlyageof26.Analliancewasalsoformedwiththe
veryimportantPolishgroupsofphilosophersandlogicians,ofwhom
Lucasiewicz,Lesnievsky,Chwistek,Kotarbinski,Ajduciewiczand
Tarskiwere perhapsthemostprominent.TheinfluenceofTarski's
work,particularlyonCarnap,wasnoticeablystrong.
ThemissionaryspiritoftheCirclefoundafurtheroutletinits
publications.In1930ittookoverajournalcalledAnnalender
Philosophie,renameditErkenntnisandmadeit,underthe
editorshipofCarnapandReichenbach,theprincipalorganofthe
positivistmovement.Inthefollowingyearstherealsoappeareda
seriesofmonographswiththecollectivetitleof
Einheitswissenschaft--UnifiedScience--andaseriesofbooks,under
thegeneraleditorshipofSchlickandPhilippFrank,withthe
collectivetitleofSchriftenzurWissenschaftlicheWeltauffassung.
Schlickhimselfcontributedtoitabookonethics,ofwhichthefirst
chapterisincludedinthisvolume,andFrankabookonthelawof
causalityanditslimitations.Amongtheothervolumestoappearin itwere
animportantbookbyCarnaponthelogicalsyntaxof
language,towhichIshallhaveoccasiontoreferagain,abookon
sociologybyNeurathwithsomewhatMarxistleanings,andKarl
Popper'sfamousLogikderForschungwhichwasdevotedtothe
philosophyofscience.Popperwasnotinfactamemberofthe Circle
andwouldatnotimehavewishedtobeclassedasa
positivist,buttheaffinitiesbetweenhimandthepositivistswhom
hecriticizedappearmorestrikingthanthedivergencies.Inany
casethemembersoftheCircledidnotinallpointsagreeamong themselves.
Thoughthelogicalpositivistmovementgatheredstrength
throughoutthethirties,theViennaCircleitselfwasintheprocessof
dissolution.By1933,whenIattendeditsmeetings,Carnapand
FrankhadacceptedchairsattheUniversityofPragueandthe discussionswere
chieflycarriedonbySchlick,Neurath,Waismann
andHahn.ButHahndiedin1934andtwoyearslaterSchlickwas
murdered,attheageof54,byadementedstudentwhoshothim
ashewasenteringtheUniversity.Thehostiletoneoftheobituaries
whichwere devotedtoSchlickinthegovernmentalpress,implying
almostthatlogical
-6positivistsdeservedtobemurderedbytheirpupils,foreshadowed
thetroubleswhichwere soontofallupontheCircle.Exceptfor
Neurath,whohadparticipatedintherevolutionarySpartacist
GovernmentinMunichattheendofthefirstworldwar,its
membershadnotbeenconspicuouslyactiveinpolitics,buttheir
criticalandscientifictempermadethemsuspecttotheright-wing
clericalgovernmentsofDolfussandSchuschniggandstillmoreso
totheNazis.Themajorityofthemwereforcedintoexile.The
adventofNazismwasfatalalsototheBerlinschool,andthePolish groupswere
disruptedbythewar.Neurath,whohadtakenrefuge
inHolland,madeavaliantattempttokeepthemovementgoing.
ThetitleofErkenntniswaschangedtoTheJournalofUnified
ScienceanditsplaceofpublicationtotheHague.Arrangements
weremadeforthepublicationbytheUniversityofChicago,where
Carnapwasestablished,ofaseriesofbrochuresambitiously
entitledtheInternationalEncyclopediaofUnifiedScience.Further
congresseswere planned.Butwiththeoutbreakofwarand
Neurath'sdeathinEnglandsomeyearslater,themovementlostits
cohesion.Mostofthevolumeswhichweredesignedtoconstitute
theEncyclopediahaveinfactappeared,buttheJournalofUnified
Scienceverysoonceasedpublicationandhasnotbeenrevived.
BesidesCarnap,Feigl,Gdel,Frank,HempelandTarskiarestillat
universitiesintheUnitedStates,andWaismannandPopperat
universitiesinEngland.ScholtzhasremainedatMnsterand
KotarbinskiandAjduciewiczinPolandandVictorKraftresumedhis
chairofphilosophyattheUniversityofVienna.But,howevermuch
influencethesephilosophersmayexertindividually,theydonot
constituteaschool.Inthissense,thelogicalpositivistmovement
hasbeenbrokenup.
Neverthelessitstraditionhasbeencontinued,especiallyinEngland,
ScandinaviaandtheUnitedStates.InScandinavia,Kailahasbeen
joinedatHelsinkibyVonWright,apupilofWittgenstein'swho
succeededhimforatimeasprofessorofphilosophyatCambridge,
theUppsalaschoolstillflourishes,underthedirectionofHedenius,
SegerstedtandMarc-Wogau,withsupportfromthelogician
WedberginStockholm,andArne NaessinOslopursueshis
sociologicalresearchesintothecurrentusesoflanguage.Petzll
continuedtoteachatLunduntilhisdeathin1957andJoergensenis
stillteachinginCopenhagen,thoughhispositivismhasbeen
modifiedbyaninjectionofMarxism.IntheUnitedStatesanumber
ofphilosopherslikeQuine,NagelandNelsonGoodmanconduct
logicalanalysisinasystematicscientificspiritthatisprobablycloser
totheoriginalidealoftheViennaCirclethananythingthatisnow
tobemetwithelse-7-
where.InthisconnectionNelsonGoodman'sbookTheStructure of
Appearance (1951)andQuine'scollectionofessaysFromaLogical
PointofView(1953)areespeciallynotable.Theiractiveinterestin
symboliclogicbringsQuineandGoodmanalsointorelationwith
Tarski,Gdel,Churchandothermembersoftheimportant
contemporarygroupofAmericanlogicians.Thesameoutlookis
maintainedbyCarnapandhispupils,notablyBarHillel,whoisnow
teachingattheUniversityofJerusalem,andbyFeiglandHempel.
OtherphilosophersintheUnitedStatessuchasNormanMalcolm,
MaxBlack,MorrisLazerowitzandC.L.Stevensonowemoretothe
influenceofG.E.MooreorthelaterWittgenstein,andconsequently
displayanapproachtophilosophicalquestionswhichiscloserto
thatofthecontemporaryBritishschools.
InspiteoftheexampleofBertrandRussell,thereisnotnow
amongBritishphilosophersthesameinterestinformallogic,or
beliefintheutilityofsymbolictechniquesforclarifyingphilosophical
issues,asistobefoundintheUnitedStates.Neitheristherethe
samedesiretoconnectphilosophywithscience.MyownLanguage,
TruthandLogic,ofwhichthefirsteditionappearedin1936,did
somethingtopopularizewhatmaybecalledtheclassicalpositionof
theViennaCirclebutsincethewartheprevailingtendencyin
Englandhasbeentoreplacethisuncompromisingpositivismwithits
blanketrejectionofmetaphysics,itsrespectforscientificmethod,
itsassumptionthatinsofarasphilosophicalproblemsaregenuine
atalltheycanbedefinitelysolvedbylogicalanalysis,byan
approachtophilosophywhichisempiricalinthepoliticalsense,the
senseinwhichBurke wasachampionofempiricism.
Generalizationsaredistrusted,particularexamplesaremultiplied
andcarefullydissected.Anattemptismadetoilluminateevery
facetofaproblemratherthantohammerorcarveoutasolution,
commonsensereignsasaconstitutional,ifnotanabsolute,
monarch,philosophicaltheoriesareputtothetouchstoneofthe
wayinwhichwordsareactuallyused.Themetaphysicianistreated
nolongerasacriminalbutasapatient:theremaybegood
reasonswhyhesaysthestrangethingsthathedoes.This
therapeutictechnique,asithasbeencalled,iswelldisplayedinthe
workofJohnWisdom,nowaprofessoratCambridge,whose
collectedvolumesofarticles,OtherMindsandPhilosophyand
Psycho-Analysis,appearedin1952and1953.Amorerobustform
oftherapyispractisedbyGilbertRyle,professorofmetaphysicsat
Oxford,whoseConceptofMind(1949),withitsattackonthe
Cartesianmythof"theghostinthemachine,"hashadaverygreat
-8influence.RyleshareswithWisdomataste,andagift,foranalogy
andmetaphor,andafondnessforpilingupexamples,butheisless
afraidofageneralization,lesstolerantofdeparturesfromordinary
usage,moredirectinhismethodthananypresent-day
Wittgensteinian,andmorereadytoassumethataphilosophical
problemhasacorrectsolution.Whatisnowsometimescalledthe
Oxfordschool,whichtakesitstonefromJ.L.Austinmorethan
fromRyle,carriesitsinterestintheordinaryuseoflanguagetoa
pointwhereitmaybethoughtthatphilosophicalanalysishasgiven
waytothestudyofphilology.Butthistendencyisnotall-prevailing.
TheworkofsuchphilosophersasStuartHampshire,P.F.Strawson
andDavidPearsshowsthatevenwithintheframeworkofthe
Oxfordmannerthereisstillroomforafairlywidelatitudeof
approach.Thechargeofscholasticismwhichisbroughtagainst
"Oxfordphilosophy"isnotentirelybaselessbutitisnotatruly
warrantedindictment.
Atthepresenttime,thephilosophicalworldiscuriouslydivided.If
positivismbetakeninitswidestsense,thesenseinwhichit
embracesallshadesofanalytical,linguistic,orradicallyempirical
philosophy,itisdominantinEnglandandinScandinavia,and
commandsconsiderableallegianceinHollandandBelgium,in
AustraliaandintheUnitedStates.Elsewhere,itmakeshardlyany
showingatall.Theoretically,itisnotinallrespectsatoddswith
Marxism:thetwoatleasthavecertainenemiesincommon:butit
cannotflourishunderCommunistregimes,sinceLenin'sMaterialism
andEmpirioCriticism,anattackonMachandhisfollowerswhich
appearedin1905,declaresittobeaformofbourgeoisidealism.In
othercountriesagain,onefindsphilosopherssubscribingtoneoThomismortoneo-Kantianismortoneo-Hegelianismorto
ExistentialismorwhateverformofGermanmetaphysicsmaybein
fashion.TheascendancyofGermanyoverFranceinthisrespectis
especiallyremarkable.Conversely,inEnglish-speakingcountries
therehasbeenthroughoutthepresentcenturyanalmostcomplete
disregardofthecurrentextravaganciesofGermanspeculative
thought.Suchnationaldivisionsareindeedregrettable.Theydonot
occurtoanythinglikethesameextentinotherbranchesof
learning.Itisespeciallycharacteristicofphilosophersthattheytend
todisagreenotmerelyaboutthesolutionofcertainproblemsbut
abouttheverynatureoftheirsubjectandthemethodsbywhichit
istobepursued.Like othersbeforethem,theViennaCircle
believedthatthiscouldandshouldberemedied.Theythoughtthat
theyhadsucceeded,whereKanthadfailed,infindingaway"toset
philosophyuponthesurepathofascience."Thisend
-9hasnotbeenattained:itmay,indeed,beunattainable.Allthe
same,therecanbeprogressinphilosophyandinonewayand
anotherthepositivistmovementisachievingit.
II.THEATTACKONMETAPHYSICS"Whenwerunoverlibraries,persuadedoftheseprinciples,what
havocmustwemake?Ifwetakeinourhandanyvolumeof
divinityorschoolmetaphysics,forinstanceletusask,Doesit
containanyabstractreasoningconcerningquantityornumber?No.
Doesitcontainanyexperimentalreasoningconcerningmatterof
factandexistence?No.Commititthentotheflames:foritcan
containnothingbutsophistryandillusion."Thisquotationistaken
fromDavidHume'sEnquiryConcerningHumanUnderstanding.Itis
anexcellentstatementofthepositivist'sposition.Inthecaseofthe
logicalpositivists,theepithet"logical"wasaddedbecausethey
wishedtoannexthediscoveriesofmodernlogictheybelieved,in
particular,thatthelogicalsymbolismwhichhadbeendevelopedby
Frege,PeanoandRussellwouldbeserviceabletothem.Buttheir
generaloutlookwasverymuchthesameasHume's.Like him,they
dividedsignificantpropositionsintotwoclassesformalpropositions,
likethoseoflogicorpuremathematics,whichtheyheldtobe
tautological,inasensethatIshallpresentlyexplain,andfactual
propositions,ofwhichitwasrequiredthattheyshouldbe
empiricallyverifiable.Theseclassesweresupposedtobe
exhaustive:sothatifasentencesucceededneitherinexpressing
somethingthatwasformallytrueorfalsenorinexpressing
somethingthatcouldbeempiricallytested,theviewtakenwasthat
itdidnotexpressanypropositionatall.Itmighthaveemotive
meaningbutitwasliterallynonsensical.Agreatdealof
philosophicaltalkwasheldtofallintothiscategory:talkaboutthe
absolute,ortranscendententities,orsubstance,orthedestinyof
man.Suchutterancesweresaidtobemetaphysicalandthe
conclusiondrawnwasthatifphilosophywastoconstituteagenuine
branchofknowledgeitmustemancipateitselffrommetaphysics.
TheViennesepositivistsdidnotgosofarastosaythatall
metaphysicalworksdeservedtobecommittedtotheflames:they
allowed,somewhatperfunctorily,thatsuchwritingmighthave
poeticmeritoreventhatitmightexpressanexcitingorinteresting
attitudetolife.Theirpointwasthatevensoitdidnotstate
anythingthatwaseithertrueorfalseandconsequentlythatitcould
contributenothingtotheincreaseofknowledge.Metaphysical
utteranceswere condemnednotforbeingemotive,whichcould
hardlybeconsideredasobjectionableinitself,butforpretendingto
becognitive,formasqueradingassomething -10thattheywere
not.Attacksonmetaphysicsoccurfairlyfrequentlyin
thehistoryofphilosophy.IhavequotedHumeandImightalso
havequotedKantwhomaintainedthatthehumanunderstanding
lostitselfincontradictionswhenitventuredbeyondtheboundsof
possibleexperience.Theoriginalityofthelogicalpositivistslayin
theirmakingtheimpossibilityofmetaphysicsdependnotuponthe
natureofwhatcouldbeknownbutuponthenatureofwhatcould
besaid.Theirchargeagainstthemetaphysicianwasthathebreaks
theruleswhichanyutterancemustsatisfyifitistobeliterally
significant. Attheoutset,theirformulationoftheseruleswaslinkedwitha
conceptionoflanguagewhichWittgenstein,whoinheriteditfrom
Russell,madefullyexplicitinhisTractatus.Theunderlying
assumptionisthattherearestatementswhichareelementaryin
thesensethat,iftheyaretrue,theycorrespondtoabsolutely
simplefacts.Itmaybethatthelanguagewhichweactuallyuse
doesnotcontainthemeansofexpressingthesestatements:the
statementswhichitcanserve toexpressmaynoneofthembe
entirelyelementarybutthesemorecomplexstatementsmuststill
restuponafoundationofelementarystatements,evenifthe
foundationbehidden.Theyaresignificantonlyinsofarastheysay
whatwouldbesaidbyaffirmingcertainelementarystatementsand
denyingcertainothers,thatis,onlyinsofarastheygiveatrueor
falsepictureoftheultimate"atomic"facts.Theycan,therefore,be
representedasbeingconstructedoutofelementarystatementsby
thelogicaloperationsofconjunctionandnegation,insuchaway
thattheirtruthorfalsehoodisentirelydependentonthetruthor
falsehoodoftheelementarystatementsinquestion.Thus,
assumingpandqtobeelementarystatements,the"molecular"
statement"porq"istakentobeequivalentto"not(not-pandnotq)"andthismeansthatitisfalseifbothpandqarefalse,but
trueinthethreeremainingcases,namelythatinwhichpandqare
bothtrue,thatinwhichpistrueandqfalse,andthatinwhichpis
falseandqtrue.Ingeneral,givennelementarystatements,where
nisanyfinitenumber,thereare2npossibledistributionsoftruth
andfalsehoodamongthem:andthemeaningofthemorecomplex
statementswhichcanbeconstructedoutofthemisconstitutedby
theselectionoftruthdistributionswithwhichtheyagreeor disagree.
Asarule,itwillbefoundthatastatementagreeswithsometruth
distributionsanddisagreeswithothers:amongthepossiblestates
ofaffairswithwhichitisconcerned,somewouldmakeittrue,and
otherswouldmakeitfalse.Thereare,however,twoextreme
casesthatinwhichastatementagreeswitheverytruthdistribution and
-11thatinwhichitagreeswithnone.Intheformercaseitistruein
anycircumstanceswhatsoever,andinthelattercasefalse.
AccordingtoWittgenstein,thesetwoextremesarethoseof
tautologyandcontradiction.Onthisview,allthetruthsoflogicare
tautologiesandifRussellandWhiteheadsucceededintheir
attempttoshowthatmathematicsisreducibletologic,soarethe
truthsofmathematics.Wittgensteinhimselfdidnotallowthat
mathematicalstatementswere tautologieshesaidthattheywere
identities:butapartfromtechnicalconsiderations,thiscomesto
muchthesamething.Thepointisthatneithersayanythingabout
theworld.Theonlywayinwhichtheycanaddtoourknowledge is
byenablingustoderiveonestatementfromanother:thatis,by
bringingouttheimplicationsofwhat,inasense,weknowalready.
Tautologiessaynothingbecauseoftheirexcessivemodesty:since
theyagreewitheverypossiblestateofaffairs,theymakenoclaim
uponthefacts.Thus,Iobtainsomeinformation,whethertrueor
false,aboutthehabitsoflionsifIamtoldthattheyare
carnivorous,andequallyifIamtoldthattheyarenotbuttotell
methattheyareeithercarnivorousornotistotellmenothing
aboutthematall.Similarly,contradictionssaynothingbecauseof
theirexcessive cantankerousnesstodisagreewitheverypossible
stateofaffairsisagaintobedisqualifiedfromgivingany
information.Ilearnnothing,notevenanythingfalse,aboutthe
habitsoflionsifIamtoldthattheyareandarenotcarnivorous.
Onthisinterpretation,tautologiesandcontradictionsare
degeneratecasesoffactualstatements.Metaphysicalassertions,on
theotherhand,aremeaninglessbecausetheybearnorelationto
fact.Theyarenotconstructedoutofelementarystatementsinany wayatall.
SinceWittgensteindidnotsaywhathetookhiselementary
statementstobe,hedidnotmakeitquiteclearatwhatpointone
isdeemedtoenterintometaphysics.Itwouldseem,however,that
anyattempttocharacterize realityasawhole,anysuchassertion
asthattheUniverse isspiritual,orthateverythinghappensforthe
bestinthebestofallpossibleworlds,mustforhimbe
metaphysicalforsuchassertionsdonotdiscriminatebetween
possiblestatesofaffairswithintheworld--nomatterwhathappens,
itistobecharacterizedasspiritual,orregardedashappeningfor
thebest--fromwhichitfollowsthattheyarenotfactual.Neitherdo
theyseemtobeconstructedoutoffactualstatementsintheway
thattautologiesare.Andeveniftheyweretheywouldstillsay nothing.
WhatevermayhavebeenWittgenstein'sownview,hisfollowers
tookitforgrantedthattheelementarystatementswhichyielded this
-12criterionofmeaningwere reportsofobservations.Asweshallsee
lateron,theysooncametodisagreeaboutthecharacterofthese
reports.Therewasadisputeoverthequestionwhethertheywere
infallible,andwhethertheyreferredtotheprivatesensationsofthe
speaker,ortopublicphysicalevents.Butitwasagreedthat,inone
formoranother,theyprovidedthetouchstonebyreference to
whichallotherstatementswere empiricallyverified.Andsince,
accordingtoWittgenstein'stheory,theyalonefurnishedthese
statementswiththeirfactualcontent,theywerealsoresponsiblefor
theirmeaning.Thisviewwasthensummedupinthefamous
sloganthatthemeaningofapropositionisitsmethodof verification.
Theassumptionbehindthissloganwasthateverythingthatcould
besaidatallcouldbeexpressedintermsofelementary
statements.Allstatementsofahigherorder,includingthemost
abstractscientifichypotheses,wereintheendnothingmorethan
shorthanddescriptionsofobservableevents.Butthisassumption
wasverydifficulttosustain.Itwasparticularlyvulnerablewhenthe
elementarystatementsweretakentoberecordsofthesubject's
immediateexperiences:forwhileithassometimesbeen
maintainedthatstatementsaboutphysicalobjectscanbefaithfully
translatedintostatementsaboutsense-data,nosuchtranslation
haseverbeenachieved:thereare,indeed,goodgroundsfor
supposingthatitisnotfeasible.Moreoverthischoiceofabasis
raisedthequestionofsolipsismtheproblemofmakingthe
transitionfromthesubject'sprivate experiencestotheexperiences
ofothersandtothepublicworld.Carnap,indeed,inhisDer
logischeAufbauderWelt(1928)madeavaliantattemptto
reconstructourwholeapparatusofempiricalconceptsona
solipsisticfoundation,takingashisstarting-pointthesingle
undefinednotionofrememberedsimilarity:buthelater
acknowledgedthatthisenterprisedidnotsucceed.Thepositionwas
easierforthosewhotreatedelementarystatementsasdescriptions
ofphysicalevents,thoughtheirrighttodothisremainedin
question:theyatleastwerenottroubledbytheproblemof
solipsismorbythedifficultyofreducingphysicalobjectstosensedata.Butotherdifficultiesremained.Themostseriousofall,
perhaps,waspresentedbythecaseofuniversalstatementsoflaw.
Forwhilethetruthofsuchastatementmaybeconfirmedbythe
accumulationoffavorable instances,itisnotformallyentailedby
themthepossibilitythatafurtherinstancewillrefuteitmust
alwaysremainopen:andthismeansthatstatementsofthissort
arenotconclusivelyverifiable.Ontheotherhand,theycanbe
conclusivelyfalsifiedinthesensethatanegativeinstanceformally
contradictsthem.ForthisreasonKarlPoppersuggestedinhis
-13LogikderForschungthatwhatshouldberequiredofafactual
statementwasjustthatitbecapableintheoryofbeingfalsified.
Andhearguedthatapartfromthelogicalsuperiorityofthis
criterionitwasmoreinaccordwithscientificpracticeforscientists
setuphypotheseswhichtheytestbylookingforcounter-examples:
whenacounter-example isdiscoveredthehypothesisisrejectedor
modifiedotherwise itisretained.ButPopper'scriterionhas
demeritsofitsown.Forinstance,ashehimselfrecognizes,itallows
onetodenyanindefiniteexistentialstatementbutnottoaffirmit.
Onecansaythattherearenoabominablesnowmen,forthiscould
befalsifiedbyfindingthem,butonecannotsaythatthereare
abominablesnowmen,forthiscouldnotbefalsifiedthefactthat
onehadfailedtofindanywouldnotproveconclusivelythatnone
existed.Whatcouldbedisprovedwouldbethatanyofthemexisted
ataparticularplaceandtime,anditisonlyifthisfurther
specificationisgiventhatthestatementbecomeslegitimate: otherwise
itistobecountedasmetaphysical.Butthisistobring
thefrontiersofmetaphysicsratherclose.
Becauseofthisandotherdifficultiestheviewwhichcametoprevail
amongthelogicalpositivistswasthatthedemandsthata
statementbeconclusivelyverifiable,orthatitbeconclusively
falsifiable,werebothtoostringentascriteriaofmeaning.They
choseinsteadtobesatisfiedwithaweakercriterionbywhichitwas
requiredonlythatastatementbecapableofbeinginsomedegree
confirmedordisconfirmedbyobservationifitwerenotitselfan
elementarystatement,ithadtobesuchthatelementary
statementscouldsupportit,buttheydidnotneedtoentailitorto
entailitsnegation.Unfortunately,thisnotionof"support"or
"confirmation"hasneveryetbeenadequatelyformalized.Various
attemptshavebeenmadetogive"theverificationprinciple,"inthis
weakerform,athoroughlypreciseexpression,buttheresultshave
notbeenaltogethersatisfactory.However,theemploymentofthe
principledidnotwaituponitsproperformulationitsgeneral
purportwasheldtobesufficientlyclear.Ihavealreadygiven
examplesofthekindofphilosophicaltalkthatitservedto
eliminate:butitsdestructivenesswasnotconfinedtowhatone
mightcallthegrosserformsofmetaphysics.Asemployedbythe
Viennesepositivists,itmadeshortworkofmostoftheperennial
problemsofphilosophy.Thus,thequestionsatissuebetween
monistsandpluralists,orbetweenrealistsandidealists,were
accountednolessspuriousthanquestionsaboutthelimitationsof
Being,oratranscendentworldofvalues.Forwhatempiricaltest
couldpossiblygotodecidewhethertheworldisone
-14ormany,orwhetherthethingsthatweperceive doordonotexist
outsidesomeone'smind?Itischaracteristicofsuchrival
philosophicalthesesasrealismandidealismthateachisconsistent
withalltheappearances,whatevertheircontentmayhappentobe.
But,forthepositivist,itisjustthisthatcondemnsthem.
Anobviousobjectiontotheverificationprinciple,whichthe
positivists'opponentswere quicktoseizeon,isthatitisnotitself
verifiable.Isupposethatitmightbetakenasanempirical
hypothesisaboutthewayinwhichpeopleactuallyusetheword
"meaning,"butinthatcaseitwouldappeartobefalseforitisnot
contrarytoordinaryusagetosaythatmetaphysicalutterancesare
meaningful.Neitherdiditssponsorsputtheprincipleforwardasthe
resultofanysuchempiricalinvestigation.Butthenwhatstatusdid
theythinkithad?Mightitnotitselfbemetaphysical?Surprisingly,
Wittgensteinaccededtothischarge."Mypropositions,"hesaidat
theendoftheTractatus,"areelucidatoryinthisway:hewho
understandsmefinallyrecognizesthemassenseless,whenhehas
climbedoutthroughthem,onthem,overthem.(Hemustsoto
speakthrowawaytheladder,afterhehasclimbeduponit.)He
mustsurmountthesepropositionsthenheseestheworldrightly."
Butthisisavainattempttohaveitbothways.Nodoubtsome
piecesofnonsensearemoresuggestivethanothers,butthisdoes
notgivethemanylogicalforce.Iftheverificationprinciplereallyis
nonsensical,itstatesnothingandifoneholdsthatitstates
nothing,thenonecannotalsomaintainthatwhatitstatesistrue.
TheViennaCircletendedtoignorethisdifficulty:butitseemsto
mefairlyclearthatwhattheywereinfactdoingwastoadoptthe
verificationprincipleasaconvention.Theywerepropoundinga
definitionofmeaningwhichaccordedwithcommonusageinthe
sensethatitsetouttheconditionsthatareinfactsatisfiedby
statementswhichareregardedasempiricallyinformative.Their
treatmentofaprioristatementswasalsointendedtoprovidean
accountofthewayinwhichsuchstatementsactuallyfunction.To
thisextenttheirworkwasdescriptiveitbecameprescriptive with
thesuggestionthatonlystatementsofthesetwokindsshouldbe
regardedaseithertrueorfalse,andthatonlystatementswhich
werecapableofbeingeithertrueorfalseshouldberegardedas
literallymeaningful.
Butwhyshouldthisprescriptionbeaccepted?Themostthathas
beenprovedisthatmetaphysicalstatementsdonotfallintothe
samecategoryasthelawsoflogic,orasscientifichypotheses,or
ashistoricalnarratives,orjudgmentsofperception,oranyother
commonsensedescriptionsofthe"natural"world.Surelyitdoes not
-15followthattheyareneithertruenorfalse,stilllessthattheyare
nonsensical? No,itdoesnotfollow.Orrather,itdoesnotfollowunlessone
makesitfollow.Thequestioniswhetheronethinksthedifference
betweenmetaphysicalandcommonsenseorscientificstatements
tobesufficientlysharpforittobeusefultounderlineitinthisway.
Thedefectofthisprocedure isthatittendstomakeoneblindto
theinterestthatmetaphysicalquestionscanhave.Itsmeritisthat
itremovesthetemptationtolookuponthemetaphysicianasasort
ofscientificoverlord.Neitheristhisatrivialmatter.Ithasfartoo
oftenbeenassumedthatthemetaphysicianwasdoingthesame
workasthescientist,onlydoingitmoreprofoundlythathewas
uncoveringadeeperlayeroffacts.Itisthereforeimportantto
emphasizethatheisnotinthissensedescribinganyfactsatall.
Butthenwhatishedoing?Whatisthepointofsaying,like
McTaggart,thattimeisunrealor,likeBerkeley,thatphysical
objectsareideasinthemindofGodor,likeHeidegger,thatthe
"nothingnihilatesitself"?Itshouldnotbeassumedthatthereisa
generalanswertothisquestion,thatmetaphysiciansarealways
doingthesamesortofthing.Onemustbegininanycaseby
lookingatthecontextinwhichsuchpronouncementsoccur.
Heidegger'sremarkisapieceofverbiage,butitcontributesinits
fashiontohisdevelopmentofthethemethatitisamatterfor
wonderthattheworldexists.Whyisthereanythingatall,heasks,
andnotrathernothing?Thisisindeedthekindofquestionthat
peopleexpectphilosopherstoput:ithasanairofprofundityabout
it.Thetroubleisthatitdoesnotadmitofanyanswer.Ontheface
ofit,McTaggart'scontentionthattimeisunrealseemshardlymore
sensible.Iftakenliterally,asimplyingthatnothingeverhappens,it
isgrotesquelyfalse.Andifitisnottobetakenliterally,whatdoesit
mean?TheansweristobefoundbylookingatMcTaggart's
arguments.Heshowshimselftheretobeperplexedbytheideaof
thepassageoftimehetriestoprove thatthenotionofanevent's
beingsuccessivelyfuture,present,andpastinvolvesavicious
infiniteregress.Theproofisinvalid,butwecanlearnsomething
fromit.Indefendingouruseoftemporalexpressionsagainst
McTaggart'sargumentswemayreachaclearerunderstandingofall
thatitimplies.Berkeley,again,wasconcernedtodiscoverwhat
couldbemeantbysayingthatphysicalobjectsexist:heconvinced
himselfbyplausibleargumentsthatwhenwespeakofphysical
objectswecanbereferringonlytocollectionsof"sensible
qualities,"theexistenceofwhichconsistsintheirbeingperceived
andhethenbroughtinGodasthepermanentsen-16soriumwhichwasneededtokeepthingsinbeing.Hisarguments
canbewithstoodbuttheydoraiseimportantphilosophical
problemsaboutthemeaningandjustificationofthestatements
thatwemakeaboutthe"externalworld."
TheViennesepositivistswerechieflyinterestedintheformaland
thenaturalsciences.Theydidnotidentifyphilosophywithscience,
buttheybelievedthatitoughttocontributeinitsownwaytothe
advanceofscientificknowledge.Theythereforecondemned
metaphysicsbecauseitfailedtomeetthiscondition.Thelogical
analystsofto-dayaremoreindulgent.Theytooareopposedto
metaphysicsinsofarasitismerelyrhapsodical:eveninthesphere
ofethicstheywishtodissociatephilosophyfrompreaching.But
theyallowthatthemetaphysicianmaysometimesbeseeingthe
worldinafreshandinterestingwayhemayhavegoodreasonfor
beingdissatisfiedwithourordinaryconcepts,orforproposingto revise
them.Inmanycasesnodoubtheisthevictimofalogical
errorbutsucherrorsmaybeinstructive.Ifphilosophicalproblems
arise,asWittgensteinthought,becauseweareledastraybycertain
featuresofourlanguage,themetaphysician,byhisvery
extravagancies,mayalsocontributetotheirdissolution.
III.LANGUAGEANDFACTWiththeireliminationofmetaphysics,theViennesepositivists
hopedthattheyhadalsoputthetheoryofknowledge behindthem,
butinthistheywere deceived.Thefirstsourceoftroublewasthe
notionofelementarystatements.Boththeircharacterandstatus
becameamatterofdispute.
Attheoutset,asIhavesaid,theprevailingviewwasthatthese
statementsreferredtothesubject'sintrospectibleorsensory
experiences.Thisviewwasadoptedbecauseitseemedtofollow
fromtheequationofthemeaningofastatementwiththemethod
ofitsverification.Forinthelastresortitisonlythroughsomeone's
havingsomeexperience thatanystatementisactuallyverified.In
mostcases,theverificationwouldconsistintheperceptionofsome
physicalobjectbutitwasheld,followingRussellandultimately
Berkeley,thatperceivingphysicalobjectswastobeanalyzedin
termsofhavingsensations,orasRussellputit,ofsensingsensedata.Thoughphysicalobjectsmightbepubliclyaccessible,sensedatawere
takentobeprivate.Therecouldbenoquestionofour
literallysharingoneanother'ssense-data,anymorethanwecan
literallyshareoneanother'sthoughtsorimagesorfeelings.The
resultwasthatthetruth
-17ofanelementarystatementcouldbedirectlycheckedonlybythe
persontowhoseexperienceitreferred.Andnotonlywashis
judgmentsovereigninthemostfavorablecase,itwasheldtobe
infallible.Onecanindeedbemistakenabouttheexperiencesthat
oneisgoingtohaveinthefuture,orevenaboutthosethatone
hashadinthepastitisnotmaintainedthatourmemoriescannot
deceiveus:butifonesetsoutmerelytorecordanexperience that
oneisactuallyhaving,then,onthisview,thereisnopossibilityof
error.Sinceonecanlie,one'sstatementmaybefalsebutone
cannotbeindoubtormistakenaboutitstruth.Ifitisfalseone
knowsittobeso.Awayinwhichthispointissometimesputisby
sayingthatstatementsofthiskindare"incorrigible."
Thisconceptionofelementarystatementswasexposedtoattackon
variousgrounds.Thereweresometowhomitseemedthatno
empiricalstatementcouldbeincorrigible,inthesenserequired. Theywere
therefore inclinedtomaintaineitherthatonecouldbe
mistakenaboutthecharacterofone'spresentexperience,sothat
thestatementswhichpurportedtorecorditwere falliblelikethe
rest,orthatthese"directrecordsofexperience"werenotgenuine
statements,sincetheypurchasedtheirsecurityattheexpenseof
sacrificingalldescriptivecontent.Butthemostseriousdifficultylay
intheprivacyoftheobjectstowhichtheelementarystatements
weresupposedtorefer.Ifeachoneofusisboundtointerpretany
statementasbeingultimatelyadescriptionofhisownprivate
experiences,itishardtoseehowwecanevercommunicateatall.
Eventospeakof"eachoneofus"istobegaquestionforitwould
seemthatonthisviewthesuppositionthatotherpeopleexistcan
havenomeaningformeunlessIconstrueitasahypothesisabout
myownobservationsofthem,thatis,aboutthecourseofmyown
actualorpossibleexperiences.ItwasmaintainedbyCarnapand
othersthatthesolipsismwhichseemedtobeinvolvedinthis
positionwasonlymethodologicalbutthiswaslittlemorethanan
avowalofthepurityoftheirintentions.Itdidnothingtomitigate
theobjectionstotheirtheory.
Atfirst,itwasthoughtthatthedifficultyaboutcommunicationcould
bemetbydrawingadistinctionbetweenthecontentofexperiences
andtheirstructure.Content,itwasmaintained,was
incommunicable.Sinceotherpeoplecannotsensemysense-data,
orsharemythoughtsorfeelings,theycannotverifythestatements
thatImakeaboutthemneithercanIverifythecorresponding
statementsthattheymakeabouttheirexperiences.AndifIcannot
verifythem,Icannotunderstandthemeither.Tothisextentwe
inhabitentirelydifferentworlds.Whatcanbeverified,however,is
thattheseworlds -18-
haveasimilarstructure.Ihavenomeansoftellingthatthefeeling
whichanotherpersonrecordswhenhesaysthatheisinpainisat
alllikethefeelingthatIcallpain:Ihavenomeansoftellingthat
thecolorswhichheidentifiesbytheuseofcertainwordslookatall
thesametohimasthecolorsforwhichIusethesewordslookto
me.ButatleastIcanobservethatweapplythewordsonthe
sameoccasions,thathisclassificationofobjectsaccordingtotheir
colorcoincideswithmineIcanobservethatwhenhesaysheisin
painhedisplayswhatIregardastheappropriatesigns.Andthisis
allthatisrequiredforcommunication.Itdoesnotmattertome
whatmyneighbour'sexperiencesactuallyare:forallthatIcan
everknowtheyareutterlydifferentfrommine.Whatmattersis
thatthestructure ofourrespectiveworldsissufficientlyalikefor
metobeabletorelyontheinformationthathegivesme.Anditis
inthissenseonlythatwehaveacommonlanguagewehave,asit
were,thesamecanvaswhicheachofuspaintsinhisownprivate
fashion.Itfollowsthatiftherearepropositions,likethe
propositionsofscience,whichhaveaninter-subjective meaning,
theymustbeinterpretedasdescriptionsofstructure.
AsIhavealreadyremarked,thefundamentalobjectiontothisview
isthatitinconsistentlyputsthe"private worlds"ofotherpeopleon
alevelwithone'sownitresultsinacurious,andindeed
contradictory,theoryofmultiplesolipsism.But,apartfromthis,the
distinctionwhichittriestomakebetweencontentandstructure
doesnotseemtobetenable.Forwhatwouldbeanexampleofa
statementwhichreferredonlytostructure?Thereisanechohere
ofLocke's"primaryqualities."Butstatementswhichrefertothe
"geometrical"propertiesofobjects,to"figure,extension,number
andmotion"havetobeinterpretedintermsofcontent,justas
muchasstatementswhichrefertocolorsandsounds.IfIhaveno
meansofknowingthatmyneighbormeansthesameasIdobyhis
useofcolor-words,Ihaveequallynomeansofknowingthathe
meansthesamebyhisuseofwordswhichrefertospatialrelations
ortonumericalquantities.IcannottelleventhatwhatItaketobe
thesamewordreallyisthesameforhim.AllthatIamleftwithis
theapparentharmonyofourbehavior.Moreoveritseemsthatthe
attempttodrawadistinctionwithintheboundariesofdescriptive
languagebetweenwhatcanandcannotbecommunicatedmustbe
self-defeating.ItleadstotheabsurditytowhichRamseydraws
attentioninhisshortpaperon"Philosophy,"whichisincludedinthis
volume:"thepositionofthechildinthefollowingdialogue:'Say
breakfast.''Can't.''Whatcan'tyousay?''Can'tsaybreakfast.'"
-19Becauseofsuchdifficulties,Neurath,andsubsequentlyCarnap,
rejectedthiswholeconceptionofelementarystatements.They
arguedthatifelementarystatementswere toserveasthebasisfor
theintersubjective statementsofscience,theymustthemselvesbe
intersubjective.Theymustrefer,nottoprivateincommunicable
experiences,buttopublicphysicalevents.Moregenerally,
statementswhichostensiblyrefertoexperiences,orto"mental"
statesorprocessesofanykind,whetherone'sownoranybody
else's,mustallbeequivalentto"physicalstatements":foritisonly
inthiswaythattheycanbepubliclyintelligible.Thisisthethesisof
physicalism.Ishallnotdwelluponithere,asIhaveinsertedan
articlebyCarnap,"PsychologyinPhysicalLanguage",whichsetsit
outatlength. Theviewthattheywere
includedin"thephysicallanguage"took
awayfromelementary,or,asNeurathandCarnapcalledthem,
"protocol,"statementstheirprivilegedposition.Theywereno
longerthoughttobeincorrigible.Theirtruth,likethatofanyother
physicalstatements,wasalwaysopentoquestion.But,morethan
this,theylosteventheirjudicialstatus.Ifaprotocol-statement
conflictedwithastatementofahigherorder,suchasascientific
hypothesis,oneorotherofthemwouldhavetobeabandoned,but
itneednotnecessarilybethescientifichypothesis:incertain
circumstancesitmightbemoreconvenienttorejecttheprotocolstatementinstead.
Ascanbeseenfromhispaperonthefoundationofknowledge
("berdasFundamentderErkenntnis")Schlickfoundthisconclusion
unacceptable.Hearguedthattotreatthereportsofobservation,
whichwaswhatprotocol-statementsweresupposedtobe,inthis
cavalierfashion,wastoputscientifichypotheses,andindeedall
would-beempiricalstatements,outsidethecontroloffact.Neurath
andCarnap,however,werenotimpressedbythisargument.They
haddecidedbythistimethatitwasmetaphysicaltotalkof
comparingstatementswithfacts.Forwhatcouldthis"comparison"
beifnotalogicalrelation?Andtheonlythingtowhichastatement
couldstandinanylogicalrelationwasanotherstatement.
Consequently,theywere ledtoadoptacoherencetheoryoftruth.
Theirversionofthecoherencetheorywasinsomewaysless
objectionablethanthatwhichtheHegelianidealistshadmade
familiar.Evenso,forthereasonswhichIsetoutinmypaperon
VerificationandExperience,itseemstomequiteuntenable.
CarnaphimselfabandoneditafterhehadbeenconvincedbyTarski
oftherespectabilityofsemanticsforsemanticsprovidesuswith
themeansofreferringtotherelationshipbetweensentencesand
whattheyareusedtosignify.Itprovides,asTarskishowed,an
adequatereformu-20lationofthecorrespondencetheoryoftruth.Ontheotherhand
Carnaphasnot,sofarasIknow,abandonedthethesisof
physicalism.But,ifhedoesstillholdit,Ithinkthatheismistaken.
Itnowseemscleartomethatstatementsabouttheexperiencesof
otherscannotbelogicallyequivalenttostatementsabouttheir
overtbehaviorwhiletomaintainthatthestatementswhichone
makesaboutone'sownexperiencesareequivalenttostatements
aboutthepubliclyobservable conditionofone'sbodyis,asRamsey
putit,tofeignanaesthesia.Nevertheless,thedifficultieswhichthis
thesiswasdesignedtomeetremain.Neitherisiteasytoseehow
elsetheycanbeavoided.Isuggest,however,thatmuchofthe
troublemayarisefromtheacceptanceoftwofalseassumptions,
thefirstbeingthatforalanguagetobepublicitmustreferto
publicobjects,andthesecondthatinmakinganempirical
statementoneisalwaysreferringtoone'sownexperiences.Istill
thinkthatempiricalstatementsmustrefertoexperiences,inthe
sensethattheymustbeverifiable butthereferenceneednotbe
totheexperiencesofanyoneperson,asopposedtoanyother.But
Iacknowledgethatthisattemptto"neutralize"theverification
principlemeetswithconsiderabledifficultiesofitsown.
IV.ETHICSOneoftheattractions,especiallyforNeurath,ofthethesisof
physicalismwasthatitsupportedthedoctrineoftheUnityof
Science.Inoneaspect,thiswaslessofadoctrinethanaprogram
itwasdesiredthatscientistsofdifferentdisciplinesshould
collaboratemorecloselywitheachotherandwithphilosophersthan
theyusuallydo:butitwasalsomaintainedthattheywere,or
shouldbe,speakingacommonlanguage,thatthevocabularyof
thesciencesshouldbeunified.Thus,theViennaCirclerejectedthe
view,whichmanystillhold,thatthereisaradicaldistinction
betweenthenaturalandthesocialsciences.Thescaleanddiversity
ofthephenomenawithwhichthesocialsciencesdealtmadethem
lesssuccessfulinestablishingscientificlaws,butthiswasadifficulty
ofpractice,notofprinciple:theytoowere concernedintheend
withphysicalevents.
Eventhosewhodidnotacceptthethesisofphysicalismagreedthat
therewasnoessentialdifferenceinaimormethodbetweenthe
variousbranchesofscience.Inthesocialsciences,nolessthanin
thenatural,anattemptwasmadetoformulatehypotheseswhich
couldbetestedbyobservation.ThusSchlick,whoincludedethics
amongthesocialsciences,deniedthatitsresultsdependedupon
theuseofanyspecialfacultyofmoralintuition.Thequestions which
-21ariseinethicsare,inhisopinion,questionsoffactwhypeoplehold
theprinciplesthattheydo,whatitisthattheydesire,andhowtheir
desirescanbefulfilled.Inshort,hisgeneralpositionisverysimilar
tothatoftheUtilitarians.Ithasmuchthesamemeritsandmuch
thesamedefects.
TheViennaCircleasawholewasnotverygreatlyinterestedin
ethicsbutitdidnotdisputeSchlick'sviewthatifethicalstatements
weretobebroughtintothescientificfold,theymustbehandledin
thewaythatheproposed.Theonlyquestionwaswhetherthey
belongedwithinthefold,whethertheywere statementsoffactat
all.Carnap,forexample,maintainedthattheywerenothesaid thattheywere
disguisedimperatives.Hedidnotdevelopthis
suggestion,butithassincebeengivensubstancebyR.M.Harein
hisbookonTheLanguageofMorals(1952).Thisimperative theory
ofethicsmayberegardedasaversionoftheso-calledEmotive
Theorywhich,mainlythroughtheworkofEnglishandAmerican
philosophers,hascometobemostcloselyassociatedwithlogical
positivism.Thesalientpointisthatethicalstatementsarenot
descriptiveofnaturalfacts,stilllessofanallegednon-naturalworld
ofvalues:theyarenotdescriptiveofanythingatall.Theproblemis
thentodeterminehowtheydofunction.InC.L.Stevenson'sbook
EthicsandLanguage(1944),where theemotivetheorywasfirst
workedoutindetail,itwasarguedthatethicalstatementsserved
thedualpurposeofexpressingtheirauthor'sapproval,or
disapproval,ofwhateverwasinquestionandrecommendingothers
tosharehisattitude.Helaidparticularemphasisuponthe persuasive
useofethicalterms.Hisviewshavenotpassedwithout
criticismevenfromthosewhosharehisgeneralstandpointbutthe
alternative accountsofethicswhichthesecriticshaveputforward
belong,asitwere,tothesamefamily.
Indiscussionsoflogicalpositivism,thistheoryofethicsisaptto receive
adisproportionatemeasureofattention,consideringthatit
standsontheperipheryofthesystem.Onereasonforthisisthatit
hasbeenthought,quitewrongly,thatitwasanonslaughtupon
morals.Ithasevenbeenasserted,withoutashadowofempirical
evidence,thatitsadvocateswerecorruptersofyouth.Infact,the
theoryonlyexplorestheconsequencesofasoundandrespectable
pointoflogicwhichwasalreadymadebyHumethatnormative
statementsarenotderivablefromdescriptivestatements,or,as
Humeputsit,that"ought"doesnotfollowfrom"is."Tosaythat
moraljudgmentsarenotfact-statingisnottosaythattheyare
unimportant,oreventhattherecannotbeargumentsintheir
favor.Buttheseargumentsdonotworkinthewaythatlogicalor
scientificargumentsdo.Itisnotasiftheintuitionistshad
discoveredgroundsformoral
-22judgmentswhichtheemotiviststriedtotakeaway.Onthe
contrary,asMr.Strawsonshowsinhispaperon"Ethical
Intuitionism,"theintuitioniststhemselvesdonotsupplyany
foundationformoraljudgments.Itisthereforeonlyonpersonal
groundsthattheycanbeentitledtoputthemselvesforwardasthe
guardiansofvirtue.
V.PHILOSOPHICALANALYSISSomeofthedissatisfactionthatisarousedbytheemotivetheoryof
ethics,andindeedbylogicalpositivismingeneral,maybedueto
thefactthatpeoplearestillinclinedtolooktophilosophyfor
guidanceastothewaytheyoughttolive.Whenthisfunctionis
deniedtoit,andwhenitisdeniedeventhepossibilityof
penetratingtheveilofappearanceandexploringthehiddendepths
ofreality,theyfeelthatitisbeingtrivialized.Ifthistime-honored
programisnonsensical,whatremains?AsRamseysays,
"philosophymustbeofsomeuse,andwemusttakeitseriously."
Butwhatfunctiondothepositivistsleaveittoperform?
FromthepointofviewofWittgenstein'sTractatus,itsfunction
wouldappeartobepurelynegative,thoughnotforthatreason
unimportant."Therightmethodofphilosophy,"saidWittgenstein,
"wouldbethis.Tosaynothingexceptwhatcanbesaid,i.e.the
propositionsofnaturalscience,i.e.somethingthathasnothingto
dowithphilosophy:andthenalways,whensomeonewishedtosay
somethingmetaphysical,todemonstratetohimthathehadgiven
nomeaningtocertainsignsinhispropositions.Thismethodwould
beunsatisfyingtotheother--he wouldnothavethefeelingthatwe
wereteachinghimphilosophy--butitwouldbetheonlystrictly
correctmethod."Thisratherdepressingviewofthephilosopher's
dutywasnotstrictlymaintainedbyWittgensteinhimself.The
PhilosophicalInvestigationscontainsagreatdealmorethana
seriesofproofsthatpeoplehavefailedtoattachanymeaningto
certainsignsintheirpropositions.Neverthelessitstillgivesthe
impressionthattophilosophizeistogetintoamuddle,ortorescue
oneselforothersfromone.Philosophyis"abattleagainstthe
bewitchmentofourintelligencebymeansoflanguage.""Whatis
youraiminphilosophy?Toshowtheflythewayoutofthefly
bottle."Allthesame,itismeritoriousoftheflytobethere.Itisthe
criticalintelligencesthatgetthemselvesbewitched.
TheTractatusleftnoroomforphilosophicalpropositions.Thewhole
fieldofsignificantdiscoursewascoveredbyformalstatementson
theonehandandempiricalstatementsontheother.There
remainednothingforphilosophytobeabout.Itwasforthisreason
-23thatWittgenstein,andalsoSchlick,maintainedthatphilosophywas
notadoctrinebutanactivity.Theresultofphilosophizing,said
Schlick,wouldnotbetoaccumulateastockofphilosophical
propositions,buttomakeotherpropositionsclear.
Buttomakepropositionsclearitmustbepossibletotalkabout
them.AsRussellpointsoutinhisintroductiontotheTractatus,
Wittgensteinappearednottoallowforthis,ortoallowforitonlyto
alimitedextent.Heimpliedthatanattempttodescribethe structure
oflanguage,asopposedtoexhibitingitinuse,must
resultinnonsense.Butthoughthisconclusionmayhavebeen
formallyacceptedbySchlick,itwasinpracticedisregardedbythe
ViennaCircle.Thus,Carnap,inhisDerLogischeAufbauderWelt,
explicitlysethimselftodescribethestructureoflanguageby
devisingwhathecalleda"Konstitution-System,"inwhichthe
varioustypesoflinguisticexpressions,orconcepts,wereassigned
theirproperplacesinadeductivehierarchy.Ifhehadbeen
questionedaboutthestatusofhisownpropositions,Isupposethat
hewouldhavesaidthattheywereanalyticconsisting,astheydid,
ofdefinitionsandtheirlogicalconsequences,theywouldbelongto
therealmofformaltruths.Howeverthismaybe,hecertainly
believedthatthesepropositionsweresignificantandhecarriedthe
ViennaCirclewithhiminholdingthattheywerethesortof
propositionsthataphilosophershouldbeexpectedtoputforward.
Theattempttobringphilosophywithinthedomainoflogicwas
carriedfurtherbyCarnapinhisbookontheLogicalSyntaxof
Language."Philosophy,"hesaysintheforewordtothisbook,"isto
bereplacedbythelogicofscience--thatistosaybythelogical
analysisoftheconceptsandsentencesofthesciences,forthelogic
ofscienceisnothingotherthanthelogicalsyntaxofthelanguage
ofscience."Thoughhespeakshereofthelanguageofscience,he
doesnotholdthatthereneedbeonlyone.Alternative
languagesystemsmaybedevised,andthechoicebetweenthemis
amatterofconvenience:thisisanimportantdeparturefromthe
positionofWittgenstein'sTractatus.AccordingtoCarnap,a
languageischaracterizedbyitsformation-rules,whichspecifywhat
sequencesofsignsaretocountaspropersentencesofthe
language,andbyitstransformation-rules,whichlaydownthe
conditionsunderwhichsentencesarevalidlyderivable fromone
another.Itmightbethoughtthatifthelanguagewastohaveany
empiricalapplicationitmustalsocontainmeaning-rulesruleswhich
wouldcorrelate itsexpressionswithobservablestatesofaffairs:but
Carnap,inthisformaliststageofhisphilosophy,thoughtthathe
coulddispensewiththem.Hebelieved,quitemistakenly,that
statementsofverbalequiva-24lencescoulddotheworknotonlyofsemanticstatementsbuteven
ofostensivedefinitions.
ItisinthisbookthatCarnapmakeshisfamousdistinctionbetween
thematerialandtheformalmodesofspeech.Hedistinguishes
threekindsofsentences:"object-sentences,"suchas"5isaprime
number"or"Babylonwasabigtown,""pseudo-objectsentences," suchas"Five
isnotathing,butanumber""Babylonwastreatedof
inyesterday'slecture,"and"syntacticalsentences"suchas"'Five'is
notathing-word,butanumberword.""Theword'Babylon'
occurredinyesterday'slecture."Thepseudo-objectsentencesare
saidtobe"quasi-syntactical,"becausetheyaresyntactical
sentencesmasqueradingasobject-sentences.Theyare"quasisyntacticalsentencesofthematerialmodeofspeech."Translation
fromthematerialintotheformalmodereplacesthembytheir
syntacticalequivalents.Toputitlesstechnically,whenonespeaks
intheformalmodeoneisovertlyspeakingaboutwordswhenone
speaksinthematerialmodeoneisspeakingaboutwordswhile
seemingtospeakaboutthings.Thisdistinctiondoesnotofcourse
applytoobject-sentences.Carnapwasnotmaintaining,assome
criticshavesupposed,thatalldiscourseisaboutwords.Whathedid
appeartooverlook,however,wastheexistenceofafurther
category,thatofpseudo-syntacticalsentencessentenceswhich
wereaboutthingsbutseemedtobeaboutwords.Asaresult,he
wasapttofallintotheerroroftreatingthesesentencesasifthey
weresyntactical.
Itiswiththeoppositeerrorthathereproachedmostother
philosophers.Hemaintainedthatphilosophicalstatementswere
syntactical,butthattheyhadbeentreatedasiftheywere
objectstatements,becauseofthefashionforexpressingtheminthe
materialmodeofspeech.Thus,totakeaselectionofhisexamples,
hearguedthat"Theworldisthetotalityoffacts,notofthings,"the
firstpropositionofWittgenstein'sTractatus,wasequivalentto "Science
isasystemofsentences,notofnames":"This
circumstanceislogicallynecessary...logicallyimpossible...
logicallypossible"became"Thissentenceisanalytic...
contradictory...notcontradictory":Kronecker'sepigram"God
createdthenaturalnumberseverythingelseinmathematicsisthe
workofman"wasawayofsaying"Thenatural-numbersymbols
areprimitivesymbolsothernumericalexpressionsareintroduced
bydefinition.""Theonlyprimitive dataarerelationsbetween
experiences"wasequivalentto"Onlytwo-ormore-termed
predicateswhoseargumentsbelongtothegenusofexperienceexpressionsoccurasdescriptive
primitive symbols":"Timeisinfinite
inbothdirections"to"Everypositiveornegativerealnumber
expressioncanbeusedasatime-co-ordinate."Eventhe
-25questionofdeterminismwassaidto"concernasyntactical
differenceinthesystemofnaturallaws."Inthiswayrival
philosophicaltheses,iftheymadeanysenseatall,were
representedasalternative proposalsaboutthewayone'slanguage
shouldbeformed.Theywerenottrueorfalse,butonlymoreor lessconvenient.
IthinkthatCarnap'sdistinctionbetweenthematerialandformal
modeswasfruitful,inthatitcalledattentiontothefactthatmany
philosophicalstatementsaredisguisedstatementsaboutlanguage.
Wherehewentwrongforthemostpartwasinsupposingthatthey
weresyntactical.Forwhattheyareconcernedwithisnottheform
ororderofwords,buttheiruse.Thisdoesnotcomeoutin
Carnap'sexamplesbecauseheillicitlysmugglessemanticsinto
syntax.Thus,"experience-expressions"isnotasyntacticalterm.
Whatmakesanexpressionan"experience-expression"isnotits
havinganyparticularformbutitsbeingusedtorefertoan
experience.Butthenthequestionwhatistocountasanexperience
becomesimportant.Neitherisittobesettledbyanarbitrary decision.
Inhismorerecentworks,Carnaphasrecognizedthelegitimacyof
semantics,andindeeddevotedconsiderableattentionbothtothe
developmentofsemantictheoryandtobuildingupsemantic
systems.Aninterestingeffectofthishasbeenamarkedrelaxation
ofhisphilosophicalausterity.Havingacquiredtherighttospeakof
thereference ofwordstothings,hehasallowedalmostanytypeof
wordtodenoteitsspecialsortofobject,thusrecreatingthe
baroqueuniversewhichRussellhadlaboredtodepopulate.His
defenseofthisapparentextravagance istobefoundinhispaper
on"Empiricism,SemanticsandOntology,"where hedistinguishes
between"internal"questionswhicharisewithinagivenconceptual
frameworkand"external"questionswhichconcernthestatusand
legitimacyoftheframeworkitself.Hehimselfhasalwaysbeen
chieflyinterestedintheexternalquestions:hehasthoughtithis
businessasaphilosophertodeviselinguisticsystemsandelaborate
conceptsthatwillbeusefultothescientist.Andnooneshoulddeny
thatthisisaseriousandlegitimateactivity.Whereheiswrong,I
think,isinassumingthattheexternalquestionspresentnoserious
problem:thatnothingmoreisatissuethanachoiceoflinguistic forms.
Itisthisdisregardofquestionsaboutthestatusofhislinguistic
frameworksthatseparatesCarnapfromtheAmericanphilosophers,
likeQuineandGoodman,whoresemblehimintheirsystematic
approachtophilosophyandintheirpreferenceforformal
techniques.Thesephilosophersareinterestedinwhattheycall
ontology,thatis,inthequestionhowfarone'schoiceoflanguage
commitsonetosayingthatcertainthingsexist."Tobe,"says Quine,"istobe
-26thevalueofavariable":andthismeansthattheextentofwhat
Russellcalledthe"furniture"oftheworlddependsupontherange
ofpredicatesthatareneededtodescribeit.BothQuineand
Goodmanwishthisfurnituretobeashardandspareaspossible. They"renounce
abstractentities"notjustbecausetheywishto
exercisetheirlogicalingenuityinseeinghowwelltheycando
withoutthem,butbecausetheycannotbringthemselvestobelieve
thattheyexist.Inthesamespirit,Goodmanforgoesmakingany
useofthenotionofpossible,asopposedtoactual,things,orofthe
distinctionbetweencausalandaccidentalconnections,orofthat
betweenanalyticandsyntheticstatements."Youmay,"hesays,
"decrysomeofthesescruplesandprotestthattherearemore
thingsinheavenandearththanaredreamtofinmyphilosophy.I
amconcerned,rather,thatthereshouldnotbemorethingsdreamt
ofinmyphilosophythanthereareinheavenorearth."Itisnot
clear,however,eitherinhiscaseorinQuine's,onwhatthis
demandforstringenteconomyisbased.Quine,indeed,allowsin
theendthatthequestionofwhatthereismustbesettledon
pragmaticgrounds.AndsoherejoinsCarnapbuthispragmatism
ismuchlessserene.
Aninterestincategories,whichisanotherwayofapproachingthe
problemofwhatthereis,ischaracteristicalsooftheBritish
philosopherswhohavebeeninfluencedbythelaterworkof
Wittgenstein.But,forthemostpart,theyareconcernednotso
muchwithtryingtoeliminatecertaintypesofentity,orto"reduce"
onetoanother,aswithbringingouttheresemblancesand
differencesinthefunctioningofthestatementswhichostensibly
refertothem.AtechniquewhichWittgensteinhimselfusesforthis
purposeisthatofdevisingwhathecallslanguagegames.Theidea
isthatbystudyingdistortedorsimplifiedmodelsofouractual
languagewecanobtainaclearerinsightintothewayitreally
works.Thisisonewayofprotectingusagainsttheerror,intowhich
wesoeasilyfall,ofassumingthatsomethingmustbethecase,
insteadoflookingandseeingwhatactuallyisthecase."Where our
languagesuggestsabodyandthereisnone,there,weshouldlike
tosay,isaspirit."Butthisistoforsakedescriptionforbogus
explanation.Veryoftenthementalprocesseswhichweareledto
postulatejustdonotoccur.Forinstance,"itisnomoreessentialto
theunderstandingofapropositionthatweshouldimagineanything
inconnectionwithitthanthatweshouldmakeasketchfromit."
SuchremarksforeshadowRyle'sattackuponthemythof"theghost
inthemachine."AndmuchasWittgensteindislikedCarnap's
methods,thereisanechoofphysicalisminhisdictumthat"an
'innerprocess'standsinneedofoutwardcriteria."
-27IsupposethatWittgensteinismainlyresponsiblefortheprevalent
interestinthequestionhowwordsareordinarilyused,though
accounthasalsotobetakenoftheinfluenceofG.E.Moore.It
doesnotseemtome,however,thatMoorehaseverbeenso
greatlyconcernedwithordinaryusageassuch.Hehasbeen
concernedwithupholdingthe"commonsenseview"oftheworld
andwithanalyzingthepropositionswhichexemplifyitbuthehas
notinsistedthatwelimitourselvestoordinaryusageincarryingout
thisanalysis.Whenhedoesappealtoordinaryusageitismainlyas
aweaponfordealingwithotherphilosophers.Heshowsthatiftheir
wordsaretakenliterally,theyareusingthemtomakestatements
whicharemanifestlyfalse.Itremainspossiblethattheyaresaying
somethingquitedifferentfromwhattheywouldappeartobe
saying,butthenthediscoveryoftheirmeaningpresentsaproblem.
Iftheyarenotusingwordsinanyordinarysense,thesensein
whichtheyareusingthemhastobemadeclear.
Tomymind,themainachievementofthe"ordinary-language
school"hasbeentheirexaminationanddissectionofthe
"unscientific"usesoflanguage.AgoodexampleofthisisJ.L.
Austin'sdescriptionofwhathecallsperformative statements:
statementslike"Iknow..."or"Ipromise..."ofwhichthe
purposeisnottoassertafactbuttocommitthespeakerincertain
waysortooffersomesortofguarantee.Towhatimaginative
lengthsthisgreaterflexibilityintheapproachtolanguagecanlead
isindicatedbyDr.Waismann'spaperwhichconcludesthisvolume.
Itshowsthatthecurrentconceptionofphilosophicalanalysishas
spreadfarbeyondRamsey'sideaofphilosophyassimplyissuingin
definitions.ButRamseywasrightinsayingthatitis"allpartofthe
vitalworkofclarifyingandorganizingourthought."
VI.IncompilingthisanthologyIhavetriedtoillustratethehistorical
developmentoflogicalpositivism,therangeofitsinterestsandthe
mainpointsofcontroversy.Lackofspacehasobligedmetopass
overmanypiecesthatIshouldhavelikedtoinclude.Inparticular,I
amsorrytohavehadnoroomforQuine'spaperon"Truthby
Convention,"inwhichthepositivists'accountofaprioristatements
iseffectivelycriticized,orforCarnap'sinfluentialarticleson
"TestabilityandMeaning."Itisespeciallytoberegrettedthatthe
volumecontainsnothingofWittgenstein.ButneithertheTractatus
LogicoPhilosophicusnorthePhilosophicalInvestigations,foralltheir
episodiccharacter,isaworktowhichonecandojusticeby
selectingpassages.Theyhavetobereadasawhole. -28-
LogicalAtomism-29[Thispageintentionallyleftblank.] -30-
1 LogicalAtomismBYBERTRANDRUSSELL
THEPHILOSOPHYwhichIadvocateisgenerallyregardedasa
speciesofrealism,andaccusedofinconsistencybecauseofthe
elementsinitwhichseemcontrarytothatdoctrine.Formypart,I
donotregardtheissuebetweenrealistsandtheiropponentsasa
fundamentaloneIcouldaltermyviewonthisissuewithout
changingmymindastoanyofthedoctrinesuponwhichIwishto
laystress.Iholdthatlogiciswhatisfundamentalinphilosophy,
andthatschoolsshouldbecharacterizedratherbytheirlogicthan
bytheirmetaphysic.Myownlogicisatomic,anditisthisaspect
uponwhichIshouldwishtolaystress.ThereforeIpreferto
describemyphilosophyas"logicalatomism,"ratherthanas
"realism,"whetherwithorwithoutsomeprefixedadjective.
Afewwordsastohistoricaldevelopmentmaybeusefulbywayof
preface.Icametophilosophythroughmathematics,orrather
throughthewishtofindsomereasontobelieveinthetruthof
mathematics.Fromearlyyouth,Ihadanardentdesiretobelieve
thattherecanbesuchathingasknowledge,combinedwithagreat
difficultyinacceptingmuchthatpassesasknowledge.Itseemed
clearthatthebestchanceoffindingindubitabletruthwouldbein
puremathematics,yetsomeofEuclid'saxiomswere obviously
doubtful,andtheinfinitesimalcalculus,asIwastaughtit,wasa
massofsophisms,whichIcouldnotbringmyselftoregardas
anythingelse.Isawnoreasontodoubtthetruthofarithmetic,but
Ididnotthenknowthatarithmeticcanbemadetoembraceall
traditionalpuremathematics.AttheageofeighteenIreadMill's
Logic,butwasprofoundlydissatisfiedwithhisreasonsforaccepting
arithmeticandgeometry.IhadnotreadHume,butitseemedto
methatpureempiricism(whichIwasdisposedtoaccept)must
leadtoscepThisessaywasRussell'scontributiontoContemporary
BritishPhilosophy,firstseries(ed.J.H.Muirhead),abook
publishedin1924.Itisherereprintedbythekind
permissionoftheauthorandGeorgeAllenandUnwin Ltd.,London.
-31ticismratherthantoMill'ssupportofreceivedscientificdoctrines.At
CambridgeIreadKantandHegel,aswellasMr.Bradley'sLogic,
whichinfluencedmeprofoundly.ForsomeyearsIwasadiscipleof
Mr.Bradley,butabout1898Ichangedmyviews,largelyasaresult
ofargumentswithG.E.Moore.Icouldnolongerbelievethat
knowingmakesanydifferencetowhatisknown.AlsoIfound
myselfdriventopluralism.Analysisofmathematicalpropositions
persuadedmethattheycouldnotbeexplainedasevenpartial
truthsunlessoneadmittedpluralismandtherealityofrelations.An
accidentledmeatthistimetostudyLeibniz,andIcametothe
conclusion(subsequentlyconfirmedbyCouturat'smasterly
researches)thatmanyofhismostcharacteristicopinionsweredue
tothepurelylogicaldoctrinethateverypropositionhasasubject
andapredicate.ThisdoctrineisonewhichLeibnizshareswith
Spinoza,Hegel,andMr.Bradleyitseemedtomethat,ifitis
rejected,thewholefoundationforthemetaphysicsofallthese
philosophersisshattered.Ithereforereturnedtotheproblem
whichhadoriginallyledmetophilosophy,namely,thefoundations
ofmathematics,applyingtoitanewlogicderivedlargelyfrom
PeanoandFrege,whichproved(atleast,soIbelieve)farmore
fruitfulthanthatoftraditionalphilosophy.
Inthefirstplace,Ifoundthatmanyofthestockphilosophical
argumentsaboutmathematics(derivedinthemainfromKant)had
beenrenderedinvalidbytheprogressofmathematicsinthe
meanwhile.Non-Euclideangeometryhadunderminedtheargument
ofthetranscendentalaesthetic.Weierstrasshadshownthatthe
differentialandintegralcalculusdonotrequiretheconceptionof
theinfinitesimal,andthat,therefore,allthathadbeensaidby
philosophersonsuchsubjectsasthecontinuityofspaceandtime
andmotionmustberegardedassheererror.Cantorfreedthe
conceptionofinfinitenumberfromcontradiction,andthusdisposed
ofKant'santinomiesaswellasmanyofHegel's.FinallyFrege
showedindetailhowarithmeticcanbededucedfrompurelogic,
withouttheneedofanyfreshideasoraxioms,thusdisproving
Kant'sassertionthat"7+5=12"issynthetic--atleastinthe
obviousinterpretationofthatdictum.Asalltheseresultswere
obtained,notbyanyheroicmethod,butbypatientdetailed
reasoning,Ibegantothinkitprobablethatphilosophyhaderredin
adoptingheroicremediesforintellectualdifficulties,andthat
solutionswere tobefoundmerelybygreatercareandaccuracy.
ThisviewIhavecometoholdmoreandmorestronglyastime
wenton,andithasledmetodoubtwhetherphilosophy,asastudy
distinctfromscienceand
-32possessedofamethodofitsown,isanythingmorethanan
unfortunatelegacyfromtheology.
Frege'sworkwasnotfinal,inthefirstplacebecauseitappliedonly
toarithmetic,nottootherbranchesofmathematicsinthesecond
placebecausehispremisesdidnotexcludecertaincontradictionsto
whichallpastsystemsofformallogicturnedouttobeliable.Dr.
WhiteheadandIincollaborationtriedtoremedythesetwodefects,
inPrincipiaMathematica,which,however,stillfallsshortoffinality
insomefundamentalpoints(notablytheaxiomofreducibility).But
inspiteofitsshortcomingsIthinkthatnoonewhoreadsthisbook
willdisputeitsmaincontention,namely,thatfromcertainideasand
axiomsofformallogic,bythehelpofthelogicofrelations,allpure
mathematicscanbededuced,withoutanynewundefinedideaor
unprovedpropositions.Thetechnicalmethodsofmathematical
logic,asdevelopedinthisbook,seemtomeverypowerful,and
capableofprovidinganewinstrumentforthediscussionofmany
problemsthathavehithertoremainedsubjecttophilosophic
vagueness.Dr.Whitehead'sConceptofNatureandPrinciplesof
NaturalKnowledgemayserveasanillustrationofwhatImean.
Whenpuremathematicsisorganizedasadeductivesystem-i.e.as
thesetofallthosepropositionsthatcanbededucedfroman
assignedsetofpremises--itbecomesobviousthat,ifweareto
believeinthetruthofpuremathematics,itcannotbesolely
becausewebelieveinthetruthofthesetofpremises.Someofthe
premisesaremuchlessobviousthansomeoftheirconsequences,
andarebelievedchieflybecauseoftheirconsequences.Thiswillbe
foundtobealwaysthecasewhenascienceisarrangedasa
deductivesystem.Itisnotthelogicallysimplestpropositionsofthe
systemthatarethemostobvious,orthatprovidethechiefpartof
ourreasonsforbelievinginthesystem.Withtheempiricalsciences
thisisevident.Electro-dynamics,forexample,canbeconcentrated
intoMaxwell'sequations,buttheseequationsarebelievedbecause
oftheobservedtruthofcertainoftheirlogicalconsequences.
Exactlythesamethinghappensinthepurerealmoflogicthe
logicallyfirstprinciplesoflogic--atleastsomeofthem--are tobe
believed,notontheirownaccount,butonaccountoftheir
consequences.Theepistemologicalquestion:"WhyshouldIbelieve
thissetofpropositions?"isquitedifferentfromthelogicalquestion:
"Whatisthesmallestandlogicallysimplestgroupofpropositions
fromwhichthissetofpropositionscanbededuced?"Ourreasons
forbelievinglogicandpuremathematicsare,inpart,onlyinductive and
-33-
probable,inspiteofthefactthat,intheirlogicalorder,the
propositionsoflogicandpuremathematicsfollowfromthe
premisesoflogicbypurededuction.Ithinkthispointimportant,
sinceerrorsareliabletoarisefromassimilatingthelogicaltothe
epistemologicalorder,andalso,conversely,fromassimilatingthe
epistemologicaltothelogicalorder.Theonlywayinwhichworkon
mathematicallogicthrowslightonthetruthorfalsehoodof
mathematicsisbydisprovingthesupposedantinomies.Thisshows
thatmathematicsmaybetrue.Buttoshowthatmathematicsis
truewouldrequireothermethodsandotherconsiderations.
OneveryimportantheuristicmaximwhichDr.WhiteheadandI
found,byexperience,tobeapplicableinmathematicallogic,and
havesinceappliedinvariousotherfields,isaformofOckham's
razor.Whensomesetofsupposedentitieshasneatlogical
properties,itturnsout,inagreatmanyinstances,thatthe
supposedentitiescanbereplacedbypurelylogicalstructures
composedofentitieswhichhavenotsuchneatproperties.Inthat
case,ininterpretingabodyofpropositionshithertobelievedtobe
aboutthesupposedentities,wecansubstitutethelogicalstructures
withoutalteringanyofthedetailofthebodyofpropositionsin
question.Thisisaneconomy,becauseentitieswithneatlogical
propertiesarealwaysinferred,andifthepropositionsinwhichthey
occurcanbeinterpretedwithoutmakingthisinference,theground
fortheinferencefails,andourbodyofpropositionsissecured
againsttheneedofadoubtfulstep.Theprinciplemaybestatedin
theform:"Whereverpossible,substituteconstructionsoutofknown
entitiesforinferencestounknownentities."
Theusesofthisprincipleareveryvarious,butarenotintelligiblein
detailtothosewhodonotknowmathematicallogic.Thefirst
instanceIcameacrosswaswhatIhavecalled"theprincipleof
abstraction,"or"theprinciplewhichdispenseswithabstraction."1
Thisprincipleisapplicableinthecaseofanysymmetricaland transitive
relation,suchasequality.Weareapttoinferthatsuch
relationsarisefrompossessionofsomecommonquality.Thismay
ormaynotbetrueprobablyitistrueinsomecasesandnotin
others.Butalltheformalpurposesofacommonqualitycanbe
servedbymembershipofthegroupoftermshavingthesaid
relationtoagiventerm.Takemagnitude,forexample.Letus
supposethatwehaveagroupofrods,allequallylong.Itiseasyto
supposethatthereisacertainquality,calledtheirlength,which
theyallshare.Butallpropositionsinwhichthissupposedquality
occurswill ____________________ 1OurKnowledge ofthe
ExternalWorld,p.42. -34retaintheirtruth-value
unchangedif,insteadof"lengthoftherod
x"wetake"membershipofthegroupofallthoserodswhichareas
longasx."Invariousspecialcases--e.g.thedefinitionofreal
numbers--asimplerconstructionispossible.
AveryimportantexampleoftheprincipleisFrege'sdefinitionofthe
cardinalnumberofagivensetoftermsastheclassofallsetsthat
are"similar"tothegivenset--where twosetsare"similar"when
thereisaone-onerelationwhosedomainistheonesetandwhose converse
domainistheother.Thusacardinalnumberistheclass
ofallthoseclasseswhicharesimilartoagivenclass.Thisdefinition
leavesunchangedthetruth-valuesofallpropositionsinwhich
cardinalnumbersoccur,andavoidstheinferencetoasetofentities
called"cardinalnumbers,"whichwereneverneededexceptforthe
purposeofmakingarithmeticintelligible,andarenownolonger
neededforthatpurpose.
Perhapsevenmoreimportantisthefactthatclassesthemselves
canbedispensedwithbysimilarmethods.Mathematicsisfullof
propositionswhichseemtorequirethataclassoranaggregate
shouldbeinsomesenseasingleentity--e.g.theproposition"the
numberofcombinationsofnthingsanynumberatatimeis2n."
Since2nisalwaysgreaterthann,thispropositionleadsto
difficultiesifclassesareadmittedbecausethenumberofclassesof
entitiesintheuniverseisgreaterthanthenumberofentitiesinthe
universe,whichwouldbeoddifclasseswere someamongentities.
Fortunatelyallthepropositionsinwhichclassesappeartobe
mentionedcanbeinterpretedwithoutsupposingthatthereare
classes.Thisisperhapsthemostimportantofalltheapplicationsof
ourprinciple.(See PrincipiaMathematica,*20.)
AnotherimportantexampleconcernswhatIcall"definite
descriptions,"i.e.suchphrasesas"theevenprime,""thepresent
KingofEngland,""thepresentKingofFrance."Therehasalways
beenadifficultyininterpretingsuchpropositionsas"thepresent
KingofFrance doesnotexist."Thedifficultyarosethrough
supposingthat"thepresentKingofFrance"isthesubjectofthis
proposition,whichmadeitnecessarytosupposethathesubsists
althoughhedoesnotexist.Butitisdifficulttoattributeeven
subsistenceto"theroundsquare"or"theevenprimegreaterthan
2."Infact,"theroundsquaredoesnotsubsist"isjustastrueas
"thepresentKingofFrancedoesnotexist."Thusthedistinction
betweenexistenceandsubsistencedoesnothelpus.Thefactis
that,whenthewords"theso-and-so"occurinaproposition,there
isnocorrespondingsingleconstituentoftheproposition,andwhen
thepropositionis
-35fullyanalyzedthewords"theso-and-so"havedisappeared.An
importantconsequenceofthetheoryofdescriptionsisthatitis
meaninglesstosay"Aexists"unless"A"is(orstandsfor)aphrase
oftheform"theso-and-so."Iftheso-and-soexists,andxisthe
so-and-so,tosay"xexists"isnonsense.Existence,inthesensein
whichitisascribedtosingleentities,isthusremovedaltogether
fromthelistoffundamentals.Theontologicalargumentandmost
ofitsrefutationsarefoundtodependuponbadgrammar.(See
PrincipiaMathematica,*14.)
Therearemanyotherexamplesofthesubstitutionofconstructions
forinferencesinpuremathematics,forexample,series,ordinal
numbers,andrealnumbers.ButIwillpassontotheexamplesin physics.
Pointsandinstantsareobviousexamples:Dr.Whiteheadhas
shownhowtoconstructthemoutofsetsofeventsallofwhichhave
afiniteextentandafiniteduration.Inrelativitytheory,itisnot
pointsorinstantsthatweprimarilyneed,butevent-particles,which
correspondtowhat,inolderlanguage,mightbedescribedasa
pointataninstant,oraninstantaneouspoint.(Informerdays,a
pointofspaceenduredthroughoutalltime,andaninstantoftime
pervadedallspace.Nowtheunitthatmathematicalphysicswants
hasneitherspatialnortemporalextension.)Event-particlesare
constructedbyjustthesamelogicalprocessbywhichpointsand instantswere
constructed.Insuchconstructions,however,weare
onadifferentplanefromthatofconstructionsinpure
mathematics.Thepossibilityofconstructinganevent-particle
dependsupontheexistenceofsetsofeventswithcertain
propertieswhethertherequiredeventsexistcanonlybeknown
empirically,ifatall.Thereisthereforenoapriorireasontoexpect
continuity(inthemathematicalsense),ortofeelconfidentthat
event-particlescanbeconstructed.Ifthequantumtheoryshould
seemtodemandadiscretespace-time,ourlogicisjustasreadyto
meetitsrequirementsastomeetthoseoftraditionalphysics,which
demandscontinuity.Thequestionispurelyempirical,andourlogic
is(asitoughttobe)equallyadaptedtoeitheralternative.
Similarconsiderationsapplytoaparticleofmatter,ortoapieceof
matteroffinitesize.Matter,traditionally,hastwoofthose"neat"
propertieswhicharethemarkofalogicalconstructionfirst,that
twopiecesofmattercannotbeatthesameplaceatthesame
timesecondly,thatonepieceofmattercannotbeintwoplacesat
thesametime.Experienceinthesubstitutionofconstructionsfor
inferencesmakesonesuspiciousofanythingsotidyandexact.One
-36cannothelpfeelingthatimpenetrabilityisnotanempiricalfact,
derivedfromobservationofbilliard-balls,butissomethinglogically
necessary.Thisfeelingiswhollyjustified,butitcouldnotbesoif
matterwere notalogicalconstruction.Animmensenumberof
occurrencescoexistinanylittleregionofspace-timewhenweare
speakingofwhatisnotlogicalconstruction,wefindnosuch
propertyasimpenetrability,but,onthecontrary,endless
overlappingoftheeventsinapartofspace-time,howeversmall.
Thereasonthatmatterisimpenetrableisbecauseourdefinitions
makeitso.Speakingroughly,andmerelysoastogiveanotionof
howthishappens,wemaysaythatapieceofmatterisallthat
happensinacertaintrackinspace-time,andthatweconstructthe
trackscalledbitsofmatterinsuchawaythattheydonotintersect.
Matterisimpenetrablebecauseitiseasiertostatethelawsof
physicsifwemakeourconstructionssoastosecure
impenetrability.Impenetrabilityisalogicallynecessaryresultof
definition,thoughthefactthatsuchadefinitionisconvenientis
empirical.Bitsofmatterarenotamongthebricksoutofwhichthe
worldisbuilt.Thebricksareevents,andbitsofmatterareportions
ofthestructure towhichwefinditconvenienttogiveseparate attention.
Inthephilosophyofmentaloccurrencestherearealso
opportunitiesfortheapplicationofourprincipleofconstructions
versusinferences.Thesubject,andtherelationofacognitionto
whatisknown,bothhavethatschematicqualitythatarousesour
suspicions.Itisclearthatthesubject,ifitistobepreservedatall,
mustbepreservedasaconstruction,notasaninferredentitythe
onlyquestioniswhetherthesubjectissufficientlyusefultobe
worthconstructing.Therelationofacognitiontowhatisknown,
again,cannotbeastraightforwardsingleultimate,asIatonetime
believedittobe.AlthoughIdonotagreewithpragmatism,Ithink
WilliamJameswasrightindrawingattentiontothecomplexityof
"knowing."Itisimpossibleinageneralsummary,suchasthe
present,tosetoutthereasonsforthisview.Butwhoeverhas
acquiescedinourprinciplewillagreethathereisprimafacieacase
forapplyingit.MostofmyAnalysisofMindconsistsofapplications
ofthisprinciple.Butaspsychologyisscientificallymuchless
perfectedthanphysics,theopportunitiesforapplyingtheprinciple
arenotsogood.Theprincipledepends,foritsuse,uponthe
existenceofsomefairlyreliablebodyofpropositions,whichareto
beinterpretedbythelogicianinsuchawayastopreserve their
truthwhileminimizingtheelementofinferencetounobserved
entities.Theprinciplethereforepresupposesamoderately
advancedscience,intheabsence
-37ofwhichthelogiciandoesnotknowwhatheoughttoconstruct.
Untilrecently,itwouldhaveseemednecessarytoconstruct
geometricalpointsnowitisevent-particlesthatarewanted.In
viewofsuchachangeinanadvancedsubjectlikephysics,itisclear
thatconstructionsinpsychologymustbepurelyprovisional.
Ihavebeenspeakinghithertoofwhatitisnotnecessarytoassume
aspartoftheultimateconstituentsoftheworld.Butlogical
constructions,likeallotherconstructions,require materials,anditis
timetoturntothepositivequestion,astowhatthesematerialsare
tobe.Thisquestion,however,requiresasapreliminarya
discussionoflogicandlanguageandtheirrelationtowhattheytry
torepresent. Theinfluenceoflanguageonphilosophyhas,Ibelieve,been
profoundandalmostunrecognized.Ifwearenottobemisledby
thisinfluence,itisnecessarytobecomeconsciousofit,andtoask
ourselvesdeliberatelyhowfaritislegitimate.Thesubjectpredicate
logic,withthesubstance-attribute metaphysic,areacaseinpoint.
Itisdoubtfulwhethereitherwouldhavebeeninventedbypeople
speakinganon-Aryanlanguagecertainlytheydonotseemtohave
ariseninChina,exceptinconnectionwithBuddhism,whichbrought
Indianphilosophywithit.Again,itisnatural,totakeadifferent
kindofinstance,tosupposethatapropernamewhichcanbeused
significantlystandsforasingleentitywesupposethatthereisa
certainmoreorlesspersistentbeingcalled"Socrates,"becausethe
samenameisappliedtoaseriesofoccurrenceswhichweareled
toregardasappearancesofthisonebeing.Aslanguagegrows
moreabstract,anewsetofentitiescomeintophilosophy,namely,
thoserepresentedbyabstractwords--the universals.Idonotwish
tomaintainthattherearenouniversals,butcertainlythereare
manyabstractwordswhichdonotstandforsingleuniversals--e.g.
triangularityandrationality.Intheserespectslanguagemisleadsus
bothbyitsvocabularyandbyitssyntax.Wemustbeonourguard
inbothrespectsifourlogicisnottoleadtoafalsemetaphysic.
Syntaxandvocabularyhavehaddifferentkindsofeffectson
philosophy.Vocabularyhasmostinfluenceoncommonsense.It
mightbeurged,conversely,thatcommonsenseproducesour
vocabulary.Thisisonlypartiallytrue.Awordisappliedatfirstto
thingswhicharemoreorlesssimilar,withoutanyreflectionasto
whethertheyhaveanypointofidentity.Butwhenonceusagehas
fixedtheobjectstowhichthewordistobeapplied,commonsense
isinfluencedbytheexistenceoftheword,andtendstosup-38posethatonewordmuststandforoneobject,whichwillbea
universalinthecaseofanadjectiveoranabstractword.Thusthe
influenceofvocabularyistowardsakindofplatonicpluralismof
thingsandideas. Theinfluenceofsyntax,inthecaseoftheIndo-European
languages,isquitedifferent.Almostanypropositioncanbeputinto
aforminwhichithasasubjectandapredicate,unitedbya
copula.Itisnaturaltoinferthateveryfacthasacorresponding
form,andconsistsinthepossessionofaqualitybyasubstance.
Thisleads,ofcourse,tomonism,sincethefactthattherewere
severalsubstances(ifitwere afact)wouldnothavetherequisite
form.Philosophers,asarule,believethemselvesfreefromthis
sortofinfluenceoflinguisticforms,butmostofthemseemtome
tobemistakeninthisbelief.Inthinkingaboutabstractmatters,the
factthatthewordsforabstractionsarenomoreabstractthan
ordinarywordsalwaysmakesiteasiertothinkaboutthewords
thanaboutwhattheystandfor,anditisalmostimpossibletoresist
consistentlythetemptationtothinkaboutthewords.
Thosewhodonotsuccumbtothesubject-predicatelogicareaptto
getonlyonestepfurther,andadmitrelationsoftwoterms,suchas
before-and-after,greater-and-less,right-and-left.Languagelends
itselftothisextensionofthesubject-predicatelogic,sincewesay
'"AprecedesB,""AexceedsB,"andsoon.Itiseasytoprovethat
thefactexpressedbyapropositionofthissortcannotconsistofthe
possessionofaqualitybyasubstance,orofthepossessionoftwo
ormorequalitiesbytwoormoresubstances.(See Principlesof
Mathematics,214.)Theextensionofthesubject-predicatelogicis
therefore rightsofarasitgoes,butobviouslyafurtherextension
canbeprovednecessarybyexactlysimilararguments.Howfaritis
necessarytogouptheseriesofthreeterm,four-term,five-term..
.relationsIdonotknow.Butitiscertainlynecessarytogobeyond
two-termrelations.Inprojectivegeometry,forexample,theorder
ofpointsonalineorofplanesthroughalinerequiresafour-term relation.
Averyunfortunateeffectofthepeculiaritiesoflanguageisin
connectionwithadjectivesandrelations.Allwordsareofthesame
logicaltypeawordisaclassofseries,ofnoisesorshapes
accordingasitisheardorread.Butthemeaningsofwordsareof
variousdifferenttypesanattribute(expressedbyanadjective)is
ofadifferenttypefromtheobjectstowhichitcanbe(whether
trulyorfalsely)attributedarelation(expressedperhapsbya
preposition,perhapsbyatransitiveverb,perhapsinsomeother way)is
-39ofadifferenttypefromthetermsbetweenwhichitholdsordoes
nothold.Thedefinitionofalogicaltypeisasfollows:AandBare
ofthesamelogicaltypeif,andonlyif,givenanyfactofwhichAis
aconstituent,thereisacorrespondingfactwhichhasBasa
constituent,whicheitherresultsbysubstitutingBforA,oristhe
negationofwhatsoresults.Totakeanillustration,Socratesand Aristotle
areofthesametype,because"Socrateswasa philosopher"and"Aristotle
wasaphilosopher"arebothfacts
SocratesandCaligulaareofthesametype,because"Socrateswas
aphilosopher"and"Caligulawasnotaphilosopher"arebothfacts.
Toloveandtokillareofthesametype,because"Platoloved
Socrates"and"PlatodidnotkillSocrates"arebothfacts.Itfollows
formallyfromthedefinitionthat,whentwowordshavemeanings
ofdifferenttypes,therelationsofthewordstowhattheymeanare
ofdifferenttypesthatistosay,thereisnotonerelationof
meaningbetweenwordsandwhattheystandfor,butasmany
relationsofmeaning,eachofadifferentlogicaltype,asthereare
logicaltypesamongtheobjectsforwhichtherearewords.Thisfact
isaverypotentsourceoferrorandconfusioninphilosophy.In
particular,ithasmadeitextraordinarilydifficulttoexpressinwords
anytheoryofrelationswhichislogicallycapableofbeingtrue,
becauselanguagecannotpreserve thedifferenceoftypebetween
arelationanditsterms.Mostoftheargumentsforandagainstthe
realityofrelationshavebeenvitiatedthroughthissourceof confusion.
Atthispoint,Iproposetodigressforamoment,andtosay,as
shortlyasIcan,whatIbelieveaboutrelations.Myownviewson
thesubjectofrelationsinthepastwere lessclearthanIthought
them,butwere bynomeanstheviewswhichmycriticssupposed
themtobe.Owingtolackofclearnessinmyownthoughts,Iwas
unabletoconveymymeaning.Thesubjectofrelationsisdifficult,
andIamfarfromclaimingtobenowclearaboutit.ButIthink
certainpointsarecleartome.AtthetimewhenIwroteThe
PrinciplesofMathematics,Ihadnotyetseenthenecessityoflogical
types.Thedoctrineoftypesprofoundlyaffectslogic,andIthink
showswhat,exactly,isthevalidelementintheargumentsofthose
whooppose"external"relations.Butsofarfromstrengtheningtheir
mainposition,thedoctrineoftypesleads,onthecontrary,toa
morecompleteandradicalatomismthananythatIconceivedtobe
possibletwentyyearsago.Thequestionofrelationsisoneofthe
mostimportantthatariseinphilosophy,asmostotherissuesturn
onit:monismandpluralismthequestionwhetheranythingis
whollytrueexceptthewholeoftruth,orwhollyrealexceptthe
-40wholeofrealityidealismandrealism,insomeoftheirforms
perhapstheveryexistenceofphilosophyasasubjectdistinctfrom
scienceandpossessingamethodofitsown.Itwillservetomake
mymeaningclearifItakeapassageinMr.Bradley'sEssayson
TruthandReality,notforcontroversialpurposes,butbecauseit
raisesexactlytheissuesthatoughttoberaised.ButfirstofallIwill
trytostatemyownview,withoutargument.2
Certaincontradictions--ofwhichthesimplestandoldestistheone
aboutEpimenidestheCretan,whosaidthatallCretanswereliars,
whichmaybereducedtothemanwhosays"Iamlying"-convinced
me,afterfiveyearsdevotedmainlytothisonequestion,thatno
solutionistechnicallypossiblewithoutthedoctrineoftypes.Inits
technicalform,thisdoctrinestatesmerelythatawordorsymbol
mayformpartofasignificantproposition,andinthissensehave
meaning,withoutbeingalwaysabletobesubstitutedforanother
wordorsymbolinthesameorsomeotherpropositionwithout
producingnonsense.Statedinthisway,thedoctrinemayseemlike
atruism."BrutuskilledCaesar"issignificant,but"Killedkilled
Caesar"isnonsense,sothatwecannotreplace"Brutus"by"killed,"
althoughbothwordshavemeaning.Thisisplaincommonsense,
butunfortunatelyalmostallphilosophyconsistsinanattemptto
forgetit.Thefollowingwords,forexample,bytheirverynature,
sinagainstit:attribute,relation,complex,fact,truth,falsehood,
not,liar,omniscience.Togiveameaningtothesewords,wehave
tomakeadetourbywayofwordsorsymbolsandthedifferent
waysinwhichtheymaymeanandeventhen,weusuallyarrive,
notatonemeaning,butataninfiniteseriesofdifferentmeanings.
Words,aswesaw,areallofthesamelogicaltypethereforewhen
themeaningsoftwowordsareofdifferenttypes,therelationsof
thetwowordstowhattheystandforarealsoofdifferenttypes.
Attribute-wordsandrelation-wordsareofthesametype,therefore
wecansaysignificantly"attribute-wordsandrelationwordshave
differentuses."Butwecannotsaysignificantly"attributesarenot
relations."Byourdefinitionoftypes,sincerelationsarerelations,
theformofwords"attributesarerelations"mustbenotfalse,but
meaningless,andtheformofwords"attributesarenotrelations,"
similarly,mustbenottrue,butmeaningless.Nevertheless,the
statement"attribute-wordsarenotrelation-words"issignificantand
true. ____________________ 2Iam
muchindebtedtomyfriendWittgensteininthismatter.See
hisTractatusLogico-Philosophicus,KeganPaul,1922.Idonot
acceptallhisdoctrines,butmydebttohimwillbeobviousto
thosewhoreadhisbook.
-41Wecannowtacklethequestionofinternalandexternalrelations,
rememberingthattheusualformulations,onbothsides,are
inconsistentwiththedoctrineoftypes.Iwillbeginwithattemptsto
statethedoctrineofexternalrelations.Itisuselesstosay"terms
areindependentoftheirrelations,"because"independent"isa
wordwhichmeansnothing.Twoeventsmaybesaidtobecausally
independentwhennocausalchainleadsfromonetotheotherthis
happens,inthespecialtheoryofrelativity,whentheseparation
betweentheeventsisspace-like.Obviouslythissenseof
"independent"isirrelevant.If,whenwesay"termsare
independentoftheirrelations,"wemean"twotermswhichhavea
givenrelationwouldbethesameiftheydidnothaveit,"thatis
obviouslyfalsefor,beingwhattheyare,theyhavetherelation,
andtherefore whateverdoesnothavetherelationisdifferent.If
wemean--asopponentsofexternalrelationssupposeustomean-thattherelationisathirdtermwhichcomesbetweenthe