What is Hacking's Argument for Entity Realism? · PDF fileWhat is Hacking's Argument for...
Transcript of What is Hacking's Argument for Entity Realism? · PDF fileWhat is Hacking's Argument for...
WhatisHacking'sArgumentforEntityRealism?
(ForthcominginSynthese)
BoazMiller
Abstract According to Ian Hacking’s Entity Realism, unobservable entities that scientists carefully
manipulate to studyotherphenomena are real.AlthoughHackingpresentshis case in an intuitive,
attractive, and persuasive way, his argument remains elusive. I present five possible readings of
Hacking’sargument:ano‐miracleargument,anindispensabilityargument,atranscendentalargument,
aVichianargument,andanon‐argument.IelucidateHacking’sargumentaccordingtoeachreading,
andreviewtheirstrengths,theirweaknesses,andtheircompatibilitywitheachother.
Keywords Scientific Realism; Entity Realism; Inference to the Best Explanation; Transcendental
Arguments;IndispensabilityArguments;Scepticism
1. Introduction
IanHacking’sEntityRealism(ER) isan influentialpositionwithin theScientificRealism(SR)
debate.AccordingtoER,unobservableentitiesthatscientistscarefullymanipulatetostudyother
phenomena, such as electrons that physicists spray froman electron gun, are real. Although
Hackingpresentshiscaseinanintuitive,attractive,andpersuasiveway,hisargumentforER
remains elusive. I suggest five interpretations of Hacking’s argument for ER: a no‐miracle
argument(NMA),anindispensabilityargument,atranscendentalargument,aVichianargument,
andanon‐argument.1Idrawthedistinctionsbetweenthesedifferentkindsofargumentbased
onthedifferentepistemicwarrantstheysupposedlyenjoy.
Withinthispaper,Iamlessinterestedinthequestionofwhichreadingmostaccurately
capturesHacking’sintentions,andmorewiththequestionofwhichreadingbestsupportsER.
ButIacknowledgethatanyanswertothelattershouldbetruetotheformer’sspirit.Thenext
B.Miller()
TheSidneyM.EdelsteinCenterforHistoryandPhilosophyofScience,TechnologyandMedicine,TheHebrewUniversityofJerusalem,9190401,Israel.
TheGraduatePrograminSTS,BarIlanUniversity,RamatGan,Israel,5290002.
e‐mail:[email protected]
1ERisassociatedwithotherscholarsaswell(Cartwright1983;Giere1988).ThispaperisdevotedsolelytoHacking.
‐2‐
sectionreviewsSRandsituatesERwithinit.Thesubsequentsectionsrespectivelydiscussthe
fivedifferent interpretations, and the last sectionexamines theextent towhich thedifferent
argumentscanworktogethermakingarobustcaseforER.
2. ERwithinSR
Roughly, SR states that science aims at discovering the truth about nature, and succeeds in
attaining this aim.Amoreprecise characterizationof SR is as a conjunctionof three theses,
dividedintosixsub‐theses:
Metaphysical:
(M1) Thereexistsamind‐independentworld.
(M2) Thisworldhasauniquestructure.
Semantic:
(S1) Scientific claims, particularly about unobservables, should be taken at
their face value; namely, they should be interpreted as purporting to
literallydescribethefurnitureofamind‐independentworld.
(S2) Whetherscientificclaimsaretrueorfalseisdeterminedbytheactualstate
ofaffairsintheworld,independentlyofourabilitytotestofverifyit;
Epistemic:
(E1) Theaimofscienceisattainingthetruthaboutnature,particularlyabout
theunobservableworld.
(E2) Matureempiricallysuccessfulscientifictheoriesachievethisaim,orcome
close to achieving it; namely, mature empirically successful scientific
theoriesaretrueorapproximatelytrueoftheirintendeddomains.2
Differentscholarsparsethesesub‐thesesslightlydifferently,butthisthreefolddivision
isnowstandard(Chakravartty2013).Someclarificationsandcontrastsareinorder.Someofthe
subthesescontrastSRwithcommitmentsofantirealistpositions.(M1)iscontrastedwiththe
2Mysubsequentmentionsoftruthshouldbeunderstoodasreferringtoapproximatetruthaswell.
‐3‐
idealistdenialoftheexistenceofanexternalworld,andwithKantianperspectivism,whichholds
thatexperiencedrealityisinescapablymediatedbyhumancognition.(M2)rejectsnominalism,
which denies a natural grouping of the world into essences or kinds. (S1) rejects logical
empiricists’ and classic instrumentalists’ view that observable termsaremere shorthand for
describingobservablephenomena.(S2)statesanon‐epistemictheoryoftruth.Itiscontrasted
inter alia with verificationism, which states that the truth value of claims depends on their
verificationmethod.Forsatisfying(S2),realiststypicallyadheretoacorrespondencetheoryof
truth,accordingtowhichclaimsaretrueiftheycorrespondtotheactualstateofaffairsinthe
world,andfalseotherwise.SRiscommittedtosomeformofcorrespondencebetweenscientific
claimsaboutunobservablesandreality,butsomerealistssubscribetoadeflationaryaccountof
truth, and shoulder the burden of correspondence in other places, such as one’s theory of
reference,oratheoryoftruth‐making.3(E1)iscontrastedwithconstructiveempiricists’view
that the aim of science is empirical adequacy, i.e., attaining the truth only about observable
phenomena, and with classic instrumentalists, who posit merely providing useful tools for
predictionandcontrolastheaimofscience.(E2)expressesepistemicoptimismaboutscience
meetingitsaim.Notallscientificrealistsacceptalltheabovesubtheses.Thus,therearemany
possiblepartlyrealistpositions,becausethenumberofpossiblecombinationsthatacceptonly
someofthesesubthesesislarge.
AstrongargumentforSRisNMA.ItsmostfamousformulationisbyPutnam(1975,73):
“Thepositiveargumentforrealismisthatitistheonlyphilosophythatdoesn’tmakethesuccess
of science amiracle”. NMA ties together empirical success, reference of theoretical terms to
unobservableentities,andtruth.NMAisaninferencetothebestexplanation(IBE).Underthe
IBE model, we legitimately infer the truth of our best explanation of some given facts.
Explanatorymerits such as simplicity and scopedeterminewhich explanations are thebest.
Explanatory considerations determine plausibility, so the best explanation is the likeliest
3IthankAnjanChakravarttyandananonymousreviewerforthisclarification.
‐4‐
explanation (Lipton 2004). NMA states that there are only two possible types of putative
explanationsfortheempiricalsuccessofscience:explanationsthatappealtothetruthofscience,
ormiraculousexplanations.Becauseputativeexplanationsthatappealtothetruthofscienceis
deemedasuperiorexplanationtoamiracle,welegitimatelyinfertheirtruth.
AnantirealistargumentagainstNMAisthepessimisticinduction(PI),whichchallenges
therelationsbetweensuccess,reference,andtruth.Laudan(1981)givesalonglistoftheories
from the history of science which were empirically successful yet false and contained non‐
referringterms.OnewaytoconstructLaudan’sPIisasanenumerativeinduction:
(PI0) Sinceallpastempiricallysuccessfulscientific theoriesare false, it is likely that
presentempiricallysuccessfultheoriesarealsofalse.
Anarguablybetterwaytoconstructtheargument(Lehoux2012,207)isasareductioofNMA:
(PI1) IfNMAisright,thenverynearlyallempiricallysuccessfultheorieswillbetrue.
(PI2) Butwehavealonglistoffalsesuccessfultheories.
(PI3) Therefore,NMAisfalse.
A common realist response to PI is selective scepticism, i.e., drawing a principled
distinctionbetweenthetrue,belief‐worthyelementsofempiricallysuccessfultheories,which
tendtobepreservedfromoldtonewtheories,andthefalse,belief‐unworthyelements,which
tendtobediscarded.AsChakravartty(2007,29)putsthis:
As in life generally, so too in science: do not believe everything you are told. Not allaspectsofscientifictheoriesaretobebelieved.Theoriescanbeinterpretedasmakingmanyclaimsaboutthenatureofreality,butatbestonehasgoodgrounds,orepistemicwarrant,forbelievingsomeoftheseclaims.Onlysomeaspectsoftheoriesarelikelytoberetainedasthesciencesmarchon.
Selectivescepticismcomesinmanyflavours.AnotableexampleisKitcher’s(1993,149)
distinctionbetweenatheory’sworkingposits,whichemploytheoreticaltermsthatrefertoreal
‐5‐
unobservableentities,andpresuppositionalposits,whichdonot.Variousversionsofstructural
realismalsotrytodistinguishbetweenthetrueelementsofmaturetheories,andalltherest.
ERenters thepicturehere, asan influential versionof selective scepticism.Hacking’s
(1983,Ch.16)ERmakestwodistinctions:betweentheoriesandentities,andbetweentwokinds
ofunobservableentities:experimentalentitiesandtheoreticalentities.Hackingarguesforrealism
onlyaboutexperimentalentities.Experimentalentitiesareentitiesthatscientistsmanipulateto
studyotherphenomena.Electronsareexperimentalentities,becauseexperimentalphysicists
spraythemtostudyweakneutralcurrents,whicharetheorizedtobecarriedbybosons.Aslogan
HackingcoinsforERis“ifyoucanspraythemthentheyarereal”(1983,23).Inadditiontothe
fact of their existence, Hacking (1983, 265) argues that we can know a modest number of
“home truths” about experimental entities, which consist of thinly theory‐laden low‐level
generalizationsabouttheirwellunderstoodcausalproperties;forexample,theirmass,spin,and
charge. Hacking contrasts these low‐level generalizations or “home truths” with full‐blown
scientifictheoriesaboutunobservables,whicharenotbelief‐worthy.
ERinnotaninnocentdoctrine.ItsignificantlynarrowsthescopeofSR.WhileERleaves
in electrons, photons, protons, andpositrons, it leavesoutneutrino, bosons, andblackholes
(Hacking1989),aswellastheoriesassuch,asopposedtotheentitiesthatpopulatethem.Its
modestyarguablymakesERamoredefensibleposition,but itdissatisfiesmanyrealists,who
wouldlikeSRtohaveawiderscope.
In sum,Hacking’sERdrawsa clear distinctionbetweenobservableandexperimental
entities, aboutwhich realism iswarranted, and everything else. Hacking’s argument for ER,
however,islessclear.Inwhatfollows,Iconsiderfiveinterpretationsofhisargument,starting
withNMA.
Istressthatwithinthispaper,Idistinguishargumenttypesonanepistemicbasis,rather
than structural. I distinguish argument types based on the type of epistemic warrant the
argument confers on its conclusion. I am aware that there are other ways of drawing the
boundariesbetweenarguments.Ialsoacknowledgethatthedistinctionsmaynotbeasclear‐cut
‐6‐
asIpresentthem.Mychoiceofterminologyforclassifyingargumentstypes,however,shouldnot
hindermysubstantivepoint,whichisthatitispossibletointerpretHacking’sargumentinfive
waysthatdiffersignificantlyinthetypeofepistemicwarranttheyenjoy.
3. TheNMAInterpretation
ThefirstinterpretationofHacking’sargumentforERisasNMA.RecallthatundertheIBEmodel,
we legitimately infer the truth of our best explanation of some given facts. Butwhy arewe
entitledtotreatexplanatorymeritsaslikelinessindicators?Therearevariouspossibleanswers
to thisquestion, butwithin this paper, I adopt the view thatour justification fordoing so is
empiricalandcontextrelative,specifically,relativetoourbackgroundknowledge.Onthisview,
IBEdoesnotnameaformalinference,butanabstractpatternwhoseforceandsuccessdepend
onthespecificbackgroundassumptionsinvolved,andwhosespecificformchangeswithcontext.
Explanatorymeritsareempiricalobservations.Forexample,ifacourtinfersadefendant’sguilt
from the fact that her fingerprints were found in the crime scene, a person matching her
descriptionwasseenfleeingthecrimescene,etc.,thisinferencereflectsthecourt’sknowledge
ofhowcrimesareusuallycommitted.Generallyspeaking,weobserveregularities,seethatsome
events are more frequent than others, and develop our explanatory merits based on these
observations(Ben‐Menahem1990;Day&Kincaid1994).
AgainstPI, realistsmayargue thatnotall empirical success is the same.Realistsmay
distinguishweakfromstrongorgenuineempiricalsuccess.Accordingtothisview,truthexplains
empiricalsuccessonlyif“empiricalsuccessisrobust,significantenough(sufficientlynoveland
abundanttobeinneedforexplanation),andtruthisagood[putative]explanation”(Solomon
2001,25).Realistsmayinsistthatwhiletheremaybeothergoodexplanationsofweakempirical
success,truthisthebestandonlyexplanationofgenuinesuccess.
Ifrealistswanttotakethisroute,however,theyfaceatwofoldchallenge.Theyneedto
show (1) that their criteria for genuine empirical success are principled and non‐question
begging,i.e.,notconstructedadhoctofavourcurrentscience;(2)thatwe,unlikepasthistorical
‐7‐
actors,areinaprivilegedepistemicpositiontodistinguishgenuinesuccessfromweaksuccess
(Lehoux2012,206‐220).Hacking’sargumentforERmaybeinterpretedasrisingtothistwofold
challenge,andreconstructedasfollows:
(NMA1) Contemporaryexperimentalphysicists’successinmanipulatingexperimental
entities tostudyotherphenomena ismuchmore impressive thanprevious
empirical successes in the history of science in a way that makes it
qualitativelydifferentfromit.
(NMA2) ERistheonlyphilosophythatdoesn’tmakethissuccessamiracle.
(NMA3) Therefore,ERistrue.
Sometextualevidencesupportsthisinterpretation.Clearly,Hackingthinksthatcurrent
experimentalpracticeisqualitativelydifferentfromthatofthe19thcenturyandbefore,suchthat
itcansupportrealisminawaythatpreviouspracticecouldnot:“Anti‐realismaboutatomswas
very sensible… a century ago. Anti‐realism aboutany sub‐microscopic entitieswas a sound
doctrine those days” (1983, 274; emphasis in the origin). “Assumptions about the minute
structure of matter could not be proven then. The only proof could be indirect… Such
indirectnessneedneverproduceconvictioninthephilosopherinclinedtoinstrumentalismor
someotherbrandofidealism”(1983,173).
Hacking writes that we have “evidence” and “proof” for the reality of experimental
entities.ThemoststraightforwardlogicalinferencestructureinwhichHacking’sclaimsabout
experimentalpracticefunctionasevidenceisIBE.YetHackingseemstodenythathisargument
isanIBE,whichseeminglymakestheNMAinterpretationanon‐starter:
Theargument…isnotthatweinfertherealityofelectronsfromoursuccess.Wedonotmake the instruments and then infer the reality of the electrons, as whenwe test ahypothesis,andthenbelieveitbecauseitpassedthetest.Thatgetsthetime‐orderwrong.Bynowwedesignapparatusrelyingonamodestnumberofhometruthsaboutelectronstoproducesomeotherphenomenonthatwewishtoinvestigate(1983,165).
‐8‐
Onceuponatimethebestreasonforthinkingthatthereareelectionsmighthavebeensuccessinexplanation…Ihavesaidthatabilitytoexplaincarrieslittlewarrantoftruth…Luckilywenolongerhavetopretendtoinferfromexplanatorysuccess(i.e.fromwhatmakes ourminds feel good). Prestcottetal. don’t explainphenomenawith electrons.Theyknowhowtousethem(1983,271‐272).
Nevertheless, I thinktheNMAinterpretationisviable.Wemaydistinguishthetypeof
scientific activity that best supports the reality of electrons (theoretical explanation versus
experimentalmanipulation)fromthetypeofargumentHackingisgivingfortheirreality(IBE
versus another type of argument). The explananda of scientific explanations are specific
phenomenaassociatedwithelectron theories,while theexplanandaof theNMAare “science
itselfanditshistoryofempiricalsuccessasawhole”(Ladyman2014).Clearly,Hackingdenies
thatmereexplanatorysuccessofelectrontheoriessupportstherealityofelectrons.Thisdoes
notmeanthathisownargumentisnotanIBE.Hacking’sargumentmaystillbereadasaNMA.
TheNMAreconstructionconstitutesaprimafaciestrongargumentforSR.Ittakesthe
sting out of PI, bynarrowing the base of the induction in its inductive formulation (PI0), or
denyingpremise(PI1)inthereductioformulation.Additionally,itsconclusionisstrong–itstates
therealityofelectrons,ratherthanmerelythatthebeliefintheirrealityisrational.TheNMA
reconstructionwouldnotconvince,however,thoseantirealistswhorejectthevalidityofIBEin
general.
4. TheIndispensabilityInterpretation
An indispensability argument “purports to establish the truth of some claim based on the
indispensabilityoftheclaiminquestionforcertainpurposes(tobespecifiedbytheparticular
argument)”(Colyvan2014,fn.1).Indispensabilityargumentsareinteraliausedforsupporting
mathematicalPlatonism.Theystatethatreferencetomathematicalentitiesisindispensableto
thepracticeofmathematics,thusweshouldcommittotheirexistence(Colyvan2014).Similar
argumentsexistinmeta‐ethicsfortherealityofmoralnorms(Enoch2011).
Structurally,IBEisanindispensabilityargument,wheretheindispensableclaimsarethe
explanans, and the purpose is explanation. Recall, however, that in this paper, I distinguish
‐9‐
arguments types on epistemic grounds, rather than structural. Epistemically, IBE and
indispensabilityargumentsthatarenotIBEareoftwotypes,becausetheyconferdifferenttypes
ofepistemicwarrantontheirrespectiveconclusions.AsIusethetermhere,thewarrantofIBE
isempirical.Bycontrast,thewarrantofindispensabilityargumentsthatarenotIBEisgrounded
inthecoherenceofpracticalandepistemicrationality;namely,therequirementforconsistency
betweenbeliefandaction.Ifonecannotavoidactingasifpistrue,thenforthesakeofrationality,
oneshouldbelievethatp.Otherwise,one isguiltyof “intellectualdishonesty” (Putnam1975,
347).4
SomethingsHackingwritesseemtobeinlinewithanindispensabilityinterpretationof
hisargumentforER:
Thevastmajorityofexperimentalphysicistsarerealistsaboutsometheoreticalentities,namelytheonestheyuse.Iclaimthattheycannothelpbeingso(1983,262;italicsintheorigin,boldadded).
As a first approximation, then,Hacking’s indispensability argumentmaybe reconstructedas
follows:
(IA1) Forthepurposeofexperimentallystudyingotherphenomenawecannothelp
treatexperimentalentitiesasreal.
(IA2) Thereforewearerationallycompelledtobelievethatexperimentalentities
exist.
Severaldifficultiesarisewiththisreconstruction.ThefirstisraisedbyHackinghimself:
Evenifexperimentersarerealistsaboutentities,itdoesnotfollowthattheyareright.Perhaps it is a matter of psychology: maybe the very skills that make for a greatexperimentergowithacertaincastofmindthatobjectifieswhateverhethinksabout(1983,265).
Hacking’s reply to this objection is that experimenters do not believe in the reality of all
theoretical entities, only experimental entities. Thus he concludes that experimenters’ 4ThisisnottodenythatanIBEmayhaveadoublewarrant.Idiscussthisoptionin§5.
‐10‐
compulsiontobelieveinexperimentalentitiesisnotpsychological(1983,265).Hacking’sreply,
however,seemsunsatisfactory.Wemayhypothesizethattheveryskillsthatmakeforagreat
experimentergowithacastofmindthatobjectifieswhatevershethinkssheusesasatool,and
theobjectionstill stands.Asecondproblemwith this reconstruction is that the term“we” is
ambiguous.Whoisrationallycompelledtobelieveinelectrons–onlyexperimenterswhouse
them,or the rest of us aswell? Itdoesnot follow, for instance, from the fact thatpracticing
religiouspeoplearerationallycompelledtobelieveinGodthateverybodyelsemustalsobelieve
in God. Third, there is a gap between claiming that it is rationally required to believe that
electronsexist(anepistemicclaim)toclaimingthatelectronsexist(ametaphysicalclaim).
Aplausiblereconstructionthataddressestheseproblemsisthis:
(IA3) Forthepurposeofexperimentallystudyingotherphenomena,experimental
physicistscannothelptreatexperimentalentities(butnotothertheoretical
entities)asreal.
(IA4) Therefore,theymustrationallybelievethatexperimentalentitiesexist.
(IA4) Forformingrationalbeliefsaboutwhichphysicalentitiesexists,peoplewho
are not experimental physicists must rationally defer to experimental
physicists’epistemicauthority.
(IA5) Therefore,wemustrationallybelievethatexperimentalentitiesexist.
(IA6) Whatwemustrationallybelieveaboutphysicalentitiesistrue.
(IA7) Therefore,experimentalentitiesexist.
Thisimprovedreconstructionraisesotherworries.Regarding(IA4),it iscontroversial
thatlaypeoplemustrationallydefertoexperts’epistemicauthority.Someargueforit(Hardwig
1985);othersrejectit(Goldman2001).Additionally,can’twesimilarlyarguethatforforming
rationalbeliefsaboutreligiousentitieswemustrationallydefertoclergy’sepistemicauthority?
And why are the relevant experts physicists, rather than philosophers, some of whom are
antirealists?(IA6)alsoseemssuspect.Ifitisderivedfromageneralprincipleaccordingtowhich
‐11‐
whatwearerationallycompelledtobelieveistrue,itseemstoostrong.Iftheclaimappliesonly
tophysicalentities,itseemsadhoc.
Idonotwishtogiveanexhaustivelistofprimafaciedifficultieswiththeargument.My
pointisthatanyplausibleindispensabilityargumentforERwouldseemtoincludecontroversial
premisesthatdonotdirectlybearontheSRdebate,suchasaboutwhenitisrationaltodeferto
expert testimony.Realistsshouldprima facienotcommittoanyparticularstanceaboutsuch
premises.TheprospectsofdevelopingarobustindispensabilityargumentforERthereforeseem
somewhatgrim.
5. TheTranscendentalInterpretation
Transcendentalargumentshaveavenerablehistoryinphilosophy.Theymaybedescribedas
trying to reverse‐engineer reality from appearances. That is, they take as a premise some
obviousfactaboutourmentallifeandproceededtoclaimingthatsomeotherstateofaffairsin
realityisanecessaryconditionforthefirstone.Theyaimatrevealingthenecessaryenabling
conditionsofunquestionableexperiencesortheunquestionablepossessionoremploymentof
somekindofknowledgeorcognitiveability.Theyareoftenusedasrepliestoscepticalclaims
that question the aspects of reality that the transcendental arguments try to establish.
Transcendentalargumentsproceedfromtheirpremisestotheirconclusionbythepreservation
oflogicalornecessarytruths(Bardon2006).
Transcendental arguments, indispensability arguments, and IBEs are sometimes
characterized as belonging to an overlapping family of arguments, since they have a similar
structure.Within thispaper, I distinguishargument typesonepistemicgrounds, rather than
structural, andwant to keep the distinction between them clear. Thus,within this family of
structurallysimilararguments,Icallan“IBE”anargumentwhoseepistemicwarrantisempirical,
an“indispensabilityargument”anargumentwhosewarrantisgroundedintherequirementof
consistencybetweenepistemicandpractical rationality, anda “transcendental argument”an
argumentwhoseepistemicwarrantstemsfromthepreservationoflogicalornecessarytruths
‐12‐
fromitspremisesto itsconclusion.Thepremisesofa transcendentalargumentforSRwould
typicallyincludesomeuncontroversialclaimsaboutobservables,andtheconclusionwouldbe
totheexistenceofsomeunobservables.
IfHackingmakesatranscendentalargument,itmaybetotheeffectthattheexistenceof
electrons and other unobservable experimental entities is a necessary condition for the
possibility of the experimental practice in the physics laboratory.Morrison (1990, 17),who
suggests a transcendental reading of Hacking’s argument for ER, interprets his argument as
follows:
[T]heargumentforrealismcouldperhapsbeseenasonewhichclaimsthatentitiesandtheirpropertiesmustbepresupposedinsuccessfulpractice;itsimplymakesnosensetodoubttheontologicalandepistemologicalclaimsthatrenderthatpracticemeaningful.5
AccordingtoMorrison’stranscendentalreading,Hacking’sargumentmaybetotheeffect
that“thecommitmentinvolvedinmanipulationisonethat,contraantirealistslikevanFraassen,
involvesmorethanempiricaladequacy”(1990,18),Specifically,Isuggestthatitmaybetothe
effect that the preservation of reference of the term “electron” to electrons (or other
unobservableexperimentalentities)duringradicaltheorychangesisanecessaryconditionfor
thepossibilityoftheexperimentalpracticeHackingdescribes.Tounderstandwhythismaybe
so, recall the challenge that Kuhn (1970) presents to SR of explaining how reference of
unobservable terms is preserved during paradigm shifts, inwhich theirmeaning undergoes
radicalchanges.Toanswerthischallenge,HackingdrawsonPutnam’scausaltheoryofmeaning,
accordingtowhichthereferenceofatermisdeterminedbyacausalchaintoaninitialbaptizing
eventofitsuse,ratherthanaspeaker’sunderstandingofitsmeaning.
Putnam’stheoryonlyexplainshowthepreservationofreferenceispossible.Itisupto
Hackingtoshowthatitispreserved.IfwedrawaparalleltoKant’stranscendentalarguments,
5 According to my taxonomy of argument types, Morrison’s interpretation can also be read as anindispensabilityinterpretation
‐13‐
wemayinterpretHackingastryingtoshowthis.AsLipton(2003,216)writes,KuhnandKant
sharemuchincommon,butdifferinthatKuhnis“Kantonwheels”:
TheKantianviewisthatthetruthswecanknowaretruthsabouta‘phenomenal’worldthat is the joint product of the ‘things in themselves’ and the organising, conceptualactivityofthehumanmind.Kuhn,however,isKantonwheels.WhereKantheldthatthehumancontributiontothephenomenalworldisinvariant,Kuhn’sviewisthatitchangesfundamentallyacrossascientificrevolution.
IfHackingindeedemploysKant’smethodologyoftranscendentalarguments,itmaythereforebe
for establishing, contra Kuhn, the invariance of reference to experimental entities despite
paradigmshifts,similarlytotheinvarianceofKant’scategories.
WemayfindsomehintsforatranscendentalreadinginHacking’stext.Hackingwrites
thatearlyKantregardedtheoreticalentitiesaspartofthenoumena,andonlylaterabandoned
thisview,realizingitsinconsistencywithhislargerphilosophy.ButthenHackingasks:
WhatpositionoughtKant tohavetakenabouttheoreticalentities thatreallydosomeworkinscience?Whatwouldhehavedonewhen,inthetwentiethcentury,welearnedhowtomanipulateandevensprayelectronsandpositrons?(1983,100‐101)
HackingspeculatesthatKantwouldnothavechangedhisposition,butHackingmaybeinclined
totakeadifferentstance,anduseKant’sownmethodtoprovetheirexistence.
CriticallyevaluatingatranscendentalargumentforERissomewhatdifficult.Thevalidity
ofatranscendentalargumentcruciallydependsonwhetheritpreserveslogicalandnecessary
truthsfromitspremisestoitsconclusionwithoutsmugglingincontroversialhiddenpremises.
Thus, evaluating the validity of a transcendental argument requires a careful scrutiny of an
elaborateformulationoftheargument.ButifHacking’sargumentistranscendental,heprovides
atbestasuggestivesketch,whichisinsufficientforsuchascrutiny.
Moreover,itisunclearifatranscendentalargumentcanbeprovidedforERingeneral,or
whether a separate argument should be provided for the existence of every unobservable
experimental entity. This is because such an argument must start with an uncontroversial
descriptionofanexperimentalpracticeatthelevelofobservableentitiesalone,andthenshow
‐14‐
that some conclusions about unobservables logically or necessarily follow from it. Since
scientistsmanipulateentitiesinmanywaysandexperimentalpracticesignificantlyvaries,itis
unclear that a generalized description of scientific experimental practice at the level of
observables,whichisrequiredforageneraltranscendentalargumentforER,canbeprovided.
AnotherdifficultywithatranscendentalargumentforERisthatitisunclearwhetherits
conclusion,namely,thattherealityofanunobservableentityisanecessarypresuppositionfor
anexperimentalpracticethatinvolvesitsmanipulation, isrightinlightofapparentempirical
examples to the contrary. Morrison (1990, 9‐13) describes a case in which physicists
manipulatedquarkstostudycharm,whiletheywerestillnotconvincedoftherealityofquarks.
Theconsistencyof thedata theyobtainedservedasa retroactivereason forbelieving in the
existence of both the manipulated and studied entities. For Morrison, this means that
“manipulationcanoccurincontextwheretherearenofirmlyheldbeliefsabouttheentitiesbeing
manipulated” (1990, 9). Even worse, Gelfert (2003) describes in detail a case in which
experimentersmanipulatequasi‐particles,whichtheybelievetobemeretheoreticalconstructs
thatdonotgenuinelyexist.
HackingmightobjectthatMorrison’sexampleisacaseofthe“experimenter’sregress”
(Collins1985),wherethesuccessconditionsforbelievingthatadetectorisworkingproperly
andthatanentityhasinfactbeendetectedarelogicallyinterdependent.Hackingmightobject
that sucha case shouldnotbegiven in supportof SR.Hackingmightargue that the casehe
describesdiffersfromMorrison’squarkcaseinthatthesuccessconditionsforbelievingthatthe
detectionequipmentworksproperlyarelogically independentoftheconditionsforbelieving
thattheexperimentssucceeded.Butthen,onemaywonderwhatissospecialforHackingabout
empiricalsuccess inthephysicslaboratoryasopposedtootherempiricalsuccess.Whatisso
specialaboutsprayingelectronsforstudyingbosons,whichwarrantsabeliefintheirexistence,
asopposedtosprayingelectronsforproducinganimageonanoldCRTtelevision,whichthe
antirealistcandismantlewithherusualargumenttoolkit?
‐15‐
There is, however, a prominent criticism of transcendental arguments, to which a
transcendentalargumentforERshouldnotbesusceptible.Transcendentalargumentsseemingly
establish,atmost,conclusionsaboutwhatbeliefsarerationalorconceptuallynecessary,butthey
fallshortofestablishingconclusionsabouttheexternalworld,asrealistswouldlike.Because
theirpremisesseeminglyconcernonlysubjects’innerworld,transcendentalargumentscannot
warrant conclusions about the external world, since the conclusions should follow only
deductively or necessarily (Stroud 1968). Thus, a transcendental argument for SR may be
criticisedonthegroundsthatinprincipleitisnotabletoestablishconclusionsaboutwhatexists
intheexternalworld,includingunobservables.
As Grundmann & Misselhorn (2003) argue, however, it is possible to overcome this
problembyintroducingintoatranscendentalargumentaminimalpremiseabouttheexternal
worldthatboththerealistandantirealistwouldfindunobjectionable,andwouldfunctionasa
bridgeprinciplethatestablishesarelationbetweensubjectiveexperienceandobjectivereality.
Within theSR debate, this suggestion looks promising, because scientific antirealists arenot
idealists. They deny neither the existence of a mind‐independent world (M1), nor that
knowledgeaboutobservablesispossible(Chakravartty2013).Thus,inprinciple,itispossibleto
introduceintoatranscendentalargumentforERanuncontroversialpremiseaboutobservables
fromwhichsomeconclusionsaboutunobservableswould logicallyornecessarily follow.For
example,anargumentmayintroducesomeuncontroversialclaimsaboutthe“hometruths”that
areassociatedwithelectronsandcanbeexpressed inobservable termsalone,suchasabout
theirbehaviouraschargedparticles,andderive theexistenceofelectrons,quaunobservable
entities,fromthem.
Howdo the three interpretations ofHacking’s argument that I reviewed so farwork
together?DotheyjointlyprovideastrongercaseforERthaneachofthemprovidesalone?The
first three arguments seem mutually consistent. They all accept the realist commitments
enumerated in §2. Because they enjoy warrants of different kinds, they implicitly rely on
different backgroundassumptionsaboutwhatmakes a claim justified.Whiledifferent, these
‐16‐
warrants need not be incompatiblewith each other; for example, a claim can have both an
empiricalwarrantandfollowdeductivelyornecessarilyfromuncontroversialpremisesabout
observables.While,asIhaveargued,theindispensabilityargumentrequiresfillinginmissing
premisesthatconcerntheargument’sownscopeandapplicabilityanddonotdirectlybearon
theSRdebate,thereislittlereasontothinkthatthesepremisesneedtobeintensionwiththe
NMA and transcendental arguments. Therefore, the three arguments may work together
providingamorerobustcaseforERthananyofthemprovidesalone.
Yetthemutualcompatibilityandstrengthofthesethreeargumentsdependsalsoonone's
generalphilosophicalcommitments.Notallscientificrealistswouldbewillingtoendorseallthe
threearguments,oranyofthem.Forexample,Maddy(2007,403‐405)arguesthatthetaskof
thescientificrealistphilosopheriscriticallyanalyzingandexplicatingscientists’owndetection
claims of unobservables, rather than adding an extra layer of philosophical analysis and
argumentation to them.Maddy rejects IBEs, transcendental arguments, and indispensability
argument for SR, since she views them as adding such an extra layer of philosophical
argumentation.Thus,shewouldnotsidewithanyofthethreeargumentsreviewedsofar.Ithink,
however,thatMaddyiswrongtorejectIBE,sincescientists'owndetectionclaimsareoftenmost
charitablyinterpretedandexplicatedasIBEs,wheretheexistenceofaparticularunobservable
entityisthebestexplanationofaparticularobservablephenomenon.6Weshouldalsobearin
mindthat IBE isaverycommonformof inference inmanynon‐philosophicalcontexts(Ben‐
Menahem1990). I therefore believe that at least some IBEs for SR are in linewithMaddy’s
philosophical naturalism. If I am right, then a scientific realist who adheres to Maddy’s
philosophicalmethodologymightaccepttheNMAargumentforER,butrejectthetranscendental
andindispensabilityarguments.
6Infact,itishardtointerpretMaddy’s(2007,406)ownexplicationandanalysisofPerrin’sargumentfortheexistenceofatomsotherthanasanIBE.EvenaccordingtoMaddy'sownaccount,PerrininferredtheexistenceofunobservableatomsfromthefactthattheyproducedanobservablepatternthatwasidenticaltothepatternofBrownianmotionofobservableentities,wheretheproductionofsuchBrownianmotionbyatomshadbeenanovelpredictionofstatisticalthermodynamics.
‐17‐
I have so far suggested three different interpretations of Hacking’s argument for ER,
which enjoy epistemicwarrants of different kinds. I have argued thatwhile different, these
interpretationsneednotnecessarilybeintensionwitheachother.Inthenexttwosections,I
review two additional interpretations – the Vichian interpretation, and the no‐argument
interpretation,andarguethattheystandintensionwiththefirstthreeinterpretationsaswellas
witheachother.
6. TheVichianInterpretation
For reviewing the fourth interpretation, consider the following excerpt from Hacking
(1983,247):
Thebestkindsofevidencefortherealityofapostulatedorinferredentityisthatwecanbegintomeasureitorotherwiseunderstanditscausalpowers.Thebestevidence,inturn,thatwehave thiskindofunderstanding is thatwecansetout, fromscratch, tobuildmachinesthatwillworkfairlyreliably,takingadvantageofthiscausalnexus.
ThisechoesItalianphilosopherVico’sMaker’sTheoryofKnowledge.Vico(1668‐1744)
equates truthwithmaking(“verum ipsum factum”),andarguesthat inasmuchasweproduce
something, we also know its causes and understand its true principles. “The verum‐factum
principledefinesknowledgeorscienceaspre‐eminentlycausal incharacterandas therefore
concernedwiththegenesisorconstructionofaphenomenon”(Zagorin1984,18).7
Boon & Knuuttila (2011) sympathetically read Hacking in this spirit. They interpret
Hacking as claiming that a mere mirroring (isomorphism, similarity, etc.) notion of
representation,whichdominates thephilosophyof science, is emptywithout an intervening
component.Theepistemic valueofmodels lies inourability tomanipulate them forgaining
knowledgeabouttheworldorusethemtomanipulatetheworld.Boon&Knuuttila(2011,87)
write:
7IthankYemimaBen‐MenachemfordrawingmyattentiontotheresemblancebetweenHackingandVico.
‐18‐
…takingacuefromHacking,wesuggestthattherepresentationalidiomdoesnotdescribeadequately our actual theoretical practices, which instead of trying to accuratelyrepresenttheworld,aremoretunedtoouractiveengagementwiththeworldthanwhatiscustomarilysupposed.
WemayreconstructHacking’sVichianargumentasfollows:
(VA1) Experimenters reliably produce phenomena by manipulating the causal
powersofelectrons.
(VA2) Havingsuchcausalmanipulativeabilitiesisasufficientconditionforhaving
knowledgeofelectrons,whereknowledgeentailstruth.
(VA3) Havingknowledgeofelectronsentailshavingknowledgethatelectronsexist.
(VA4) Therefore,experimentersknowthatelectronsexist.
(VA5) Therefore,electronsexist.
ThewarrantofthisargumentisgroundedinwhateverprinciplesunderlieVico’sMaker
TheoryofKnowledge.Thus,(VA2)seemstocarrytheepistemicweightoftheargument.Vico
thinksthathumanscanonlyhaveknowledgeofthehumanworld,becauseonlyGodproduces
natural phenomena. As Carrier (2011, 48‐52) argues, however, we nowadays study non‐
naturallyoccurringphenomenaofourownmaking,andusethemtostudyotherphenomenaas
well.Carrierdoubts,however,thatcausalmanipulativeabilitiesaresufficientforknowledgefor
two reasons. First, sometimes scientists know how to produce certain effects without fully
understanding theunderling causal dynamics at play. Second,manipulative ability is usually
consistentlyachievedundercertainbackgroundconditions,andscientistsdonotalwaysknow
howchangingthemwouldaffecttheoutcome.
TheVichianargumentdoesnotsquarenicelywiththefirstthreeinterpretationsorwith
thestandardrealistcommitments.Vico’sequationoftruthwithmakingviolatesSRcommitment
to a non‐epistemic theory of truth (S2). That is, the Vichian argument is framed within an
anthropocentricepistemologythatequatestruthwithmaking,yetrealistsholdthatthetruth
valuesofclaimsisindependentofthehumanabilitytotestorverifythem.AccordingtoHacking,
‐19‐
hisownERcanbeexpressedintermsofacorrespondencetheoryoftruthorsomethinglikeit
(1988,291).
Moreover, a Vichian framework forms a hierarchy of knowability, which seems
incompatiblewith thehierarchyassumedwithin theSRdebate.Thoughmanyargue that the
lines between observable and unobservable entities are blurry, it is still generally assumed
within the SR debate that observable entities are on more secure epistemic grounds than
unobservable.Thatis,allsidestothedebateacceptthatwehaveknowledgeofobservables,and
thedebateconcernsunobservables.Manipulabilitydrawstheepistemiclinesdifferentlyfrom
observability.Itplacesmanipulatableentities,suchasunobservableelectronsandobservable
car engines, higher on its epistemic hierarchy than non‐manipulatable entities, such as
unobservablebosonsandobservablehurricanes.
IdonotsuggestthataVichianepistemologyandthestandardSRassumptionscannotbe
reconciledwith eachother ifonehasenoughphilosophicalmotivationand free time. Sucha
reconciliation,however,wouldhavetosignificantlyrelaxtherealistcommitments,toanextent
thatitwouldbeunclearifERcouldstill legitimatelybecalled“realism”,bearinginmindthat
becauseERisantirealistabouttheories,itisalreadyamodestrealistposition.Alternatively,such
areconciliationwouldtakethewindoutoftheMaker’sTheoryofKnowledge,becauseitwould
havetorelaxitsanthropocentricfocus,whichmakesitattractivetobeginwith.Thus,arobust
caseforERthatincludesaVichianlineofargumentseemsunpromising.
7. TheNon‐ArgumentInterpretation
ThelastinterpretationofHacking’sargumentisthatitisnotanargumentatall,asHackingdoes
not think that an argument can establish the reality of electrons. Rather, the experimental
practicedirectlyillustratestheirreality.Accordingtothisinterpretation,theepistemicwarrant
forERliesindirectlyperceivingthelaboratorypractice,ratherthananyfeatureofanargument
forit.Accordingtothisreading,therootsofHacking’s(non)argument
‐20‐
lieintheanti‐theoreticaltraditionsofpragmatismandinthecommonsensicalreactiontoBerkeley'sidealismmadefamousbySamuelJohnson.WhenJohnsonmanipulatedastonewithhisfootheatonceaffirmedthatthestonewasmorethanameretheoreticalconstructandthathewaswarrantedinapplyingcommonsensereasoningtoaffirmitsexistence(LyonsandClarke2002,xx‐xxi).
ThisreadingaccordswiththefactthatHackingdoesnotgiveanexplicitargumentforER,
butgivesadetailedvividdescriptionoftheexperimentalpracticeinthephysicslaboratory.It
alsoaccordswithHacking’sdepictionofhisconversationwithaphysicist,inwhichthephysicist
saidphysicistsspraypositrons,andafterwhichHackingbecamearealist(1983,23).
Insuggestingthisinterpretation,IfollowGreco’sinterpretationofMoore’sfamousproof
ofanexternalworld,inwhichMoore(1962,144)gestureswithhishandswhilesaying“hereis
onehand”and “here isanother”,and thendeduces theexistenceofanexternalworld.Greco
(2002)arguesthattheepistemicforceofMoore’sproofdoesnotlieinanyfeatureofMoore’s
argument,but in thedirectperceptionofanundeniable fact, i.e.,Moore’shands.Grecoreads
Moore’sargumentasplayful,whereMoore’sverypoint is thatwedonotknowthatexternal
thingsexistbyprovingthis,butbyperceivingthem.ThisissimilartoBrown’s(1997)claimthat
somepicturesconstituteinandbythemselvesmathematicalproofs,becausetheyallowpeople
todirectlyperceivethetruthofamathematicaltheorem,withoutneedingtotranslatethemto,
oraccompanythemwithanargument.
Thisinterpretationdoesnotrequireareconstruction,butwemayfollowGreco’sMoore
andgivethefollowingreconstruction,whichshouldbereadplayfully:
(NA1) Experimenterssprayelectrons.
(NA2) Therefore,electronsexist.
The non‐argument interpretation returns SR to its common‐sense roots. Before SR
shatteredintocountlesspositionswithhair‐splittingepistemicandonticdistinctionsbetween
them, itwasconsideredacommon‐sensephilosophy,whichtakesourbestscienceat its face
value, and elucidates scientists’ own philosophy. In particular, it was the common‐sense
‐21‐
responsetologicalempiricists’elaboratetakeonscientificclaims,theirtruth,andtheirmeaning.
Yet,onemayquestionthecommonsensicalstatusofknowledgeofunobservableentities,such
aselectrons,asopposedtoobservableentities,suchasMoore’shands.Inresponse,Hackingmay
argue that forexperimentersand thosewhostudy them,knowledgeofelectronsbecomesas
commonsensicalasknowledgeofone’shands.Theprospectsofsuchasuggestionneedfurther
exploring.
Therearedifficulties reconciling thenon‐argumentargument forERwith the rest.As
Morrisonobserves,ifweacceptthetranscendentalinterpretation,“wemustdismissHacking's
claimthatmanipulationoffersdirectproofoftherealityofentities”(1990,18).Thesamewould
applytoalltherestoftheinterpretationstoo.Accordingtothenon‐argumentinterpretation,the
rest of the arguments commit a category mistake: they try to prove with a philosophical
argument what can only be proven with a direct unmediated physical demonstration. The
non‐argumentinterpretation,then,nullifiesanyotherphilosophicalargumentsforER.Itthus
cannotbeintegratedwithinarobustcaseforER.
8. Conclusion
WhileHacking’scatchyslogan“ifyoucanspraythemthentheyarereal”seems intuitiveand
persuasive, and Hacking’s argument for Entity Realism seems simple and straightforward,
appearancescanbemisleading.IdistinguishedfivepossiblereadingsofHacking’sargumentfor
EntityRealismbasedontheepistemicrationalethatunderliesthem:(1)ano‐miraclesargument,
whose epistemic warrant is empirical; (2) an indispensability argument, whose warrant is
groundedintherequirementofconsistencybetweenepistemicandpracticalrationality;(3)a
transcendentalargument,whosewarrantisgroundedinthepreservationoflogicalornecessary
truthsfromasetofuncontroversialpremisestoitsconclusion;(4)aVichianargument,whose
warrant isgrounded inanepistemology in thespiritofVico’sMaker’sTheoryofKnowledge,
whichconnectstruthwithmaking;(5)anon‐argument,wheretheepistemicwarrantdoesnot
lieinanyfeatureofaphilosophicalargumentforEntityRealism,butinthedirectunmediated
‐22‐
demonstrationpoweroftheexperimentalpracticeinthephysicslaboratory.Theversionofthe
argumentascientificrealistwouldliketosidewith,ifany,dependsondeepissuesattheheart
ofthedebate,suchasthenatureoftruth,representation,andknowledge.
Acknowledgements
I thank Anjan Chakravartty, Jacob Stegenga, Yemima Ben‐Menachem, and Daniel Steel forhelpfulcommentsanddiscussions.ThispaperwaspresentedattheIsraelSocietyforHistoryandPhilosophyofScienceAnnualMeeting,Jerusalem,2013,andatthePSABiennialMeeting,Chicago,2014.Ithanktheaudiencemembersforhelpfulcomments.IamgratefultothestudentsatmyseminarinphilosophyofscienceattheBarIlanUniversityGraduatePrograminScience,Technology,andSocietyforrealizingthatgettingatthebottomofHacking’sargumentistricky.IthanktheDanDavidFoundation,theCohnInstitutefortheHistoryandPhilosophyofScienceandIdeas,TelAvivUniversity,andtheSidneyM.EdelsteinCenterforHistoryandPhilosophyofScience, Technology and Medicine, the Hebrew University of Jerusalem, for postdoctoralfellowships.
References
Bardon,Adrian.2006.“TranscendentalArguments.”InTheInternetEncyclopediaofPhilosophy,ed.JamesFieser&BradleyDowden.http://www.iep.utm.edu/trans‐ar/
Ben‐Menahem,Yemima.1990.“InferencetotheBestExplanation”.Erkenntnis33:319‐334.
Boon,Mieke&TarjaKnuuttila.2011.“BreakingUpwiththeEpochalBreak:TheCaseofEngineeringSciences.”InScienceTransformed?DebatingClaimsofanEpochalBreak,eds.AlfredNordmann,HansRadder&GregorSchiemann,86‐79.Pittsburgh,PA:UniversityofPittsburghPress.
Brown,JamesR.1997.“ProofsandPictures.”BritishJournalforthePhilosophyofScience48(2):161‐180.
Carrier,Martin.2011.“‘KnowledgeIsPower,’orHowtoCapturetheRelationshipbetweenScienceandTechnoscience.”InScienceTransformed?DebatingClaimsofanEpochalBreak,eds.AlfredNordmann,HansRadder,&GregorSchiemann,43‐53.Pittsburgh,PA:UniversityofPittsburghPress.
Cartwright,Nancy.1983.HowtheLawsofPhysicsLie.Oxford:ClarendonPress.
Chakravartty,Anjan.2007.AMetaphysicsforScientificRealism:KnowingtheUnobservable.Cambridge:CambridgeUP.
Chakravartty,Anjan.2013.“ScientificRealism.”InTheStanfordEncyclopediaofPhilosophy,ed.EdwardN.Zalta.http://plato.stanford.edu/entries/scientific‐realism/
‐23‐
Collins,Harry.1985.ChangingOrder:ReplicationandInductioninScientificPractice.Chicago:UniversityofChicagoPress.
Colyvan,Mark.204.“IndispensabilityArgumentsinthePhilosophyofMathematics.”InTheStanfordEncyclopediaofPhilosophy,ed.EdwardN.Zalta.http://plato.stanford.edu/entries/mathphil‐indis/
Day,Timothy&HaroldKincaid.(1994).“PuttingInferencetotheBestExplanationinitsPlace.”Synthese98(2):271‐295.
Enoch,David.2011.TakingMoralitySeriously:ADefenseofRobustRealism.Oxford:OUP.
Gelfert,Axel.2003.“ManipulativeSuccessandtheUnreal”.InternationalStudiesinthePhilosophyofScience17(3):245‐263.
Giere,Ronald.1988.ExplainingScience:ACognitiveApproach.Chicago:UniversityofChicagoPress.
Goldman,AlvinI.2001.“Experts:WhichOnesShouldYouTrust?”PhilosophyandPhenomenologicalResearch63(1):85‐111.
Greco,John.2002.“HowtoReidMoore”.ThePhilosophicalQuarterly52(209):544‐563.
Grundmann,Thomas&CatrinMisselhorn.2003.“TranscendentalArgumentsandRealism.”InStrawsonandKant,ed.Hans‐JohannGlock,205‐218.Oxford:OxfordUP.
Hacking,Ian.1983.RepresentingandIntervening:IntroductoryTopicsinthePhilosophyofNaturalScience.Cambridge:CambridgeUP.
Hacking,Ian.1988."TheParticipantIrrealistAtLargeintheLaboratory".BritishJournalforthePhilosophyofScience39(3):277‐294.
Hacking,Ian.1989.“ExtragalacticReality:TheCaseofGravitationalLensing.”PhilosophyofScience56(4):555‐581.
Hardwig,John.1985.“EpistemicDependence.”TheJournalofPhilosophy82(7):335‐349.
Kitcher,Philip.1993.TheAdvancementofScience:SciencewithoutLegend,ObjectivitywithoutIllusions.Oxford:OxfordUP.
Kuhn,ThomasS.1970.TheStructureofScientificRevolutions,2nded.Chicago:UniversityofChicagoPress.
Ladyman,James.“StructuralRealism.”InTheStanfordEncyclopediaofPhilosophy,ed.EdwardN.Zalta.http://plato.stanford.edu/entries/structural‐realism/
Lehoux,Daryn.2012.WhatDidtheRomansKnow?AnInquiryintoScienceandWorldmaking.Chicago:UniversityofChicagoPress.
‐24‐
Lipton,Peter.2003.“KantonWheels.”SocialEpistemology17(2&3):215–219.
Lipton,Peter.2004.InferencetotheBestExplanation,2nded.London:Routledge.
Lyons,TimothyD.&SteveClarke.2002.“Introduction:ScientificRealismandCommonsense.”InScientificRealismandCommonsense,eds.SteveClarke&TimothyD.Lyons,ix‐xxiii.Dordrecht:Kluwer.
Maddy,Penelope.2007.SecondPhilosophy:ANaturalisticMethod.Oxford:OUP.
Moore,GeorgeE.1962.“ProofofanExternalWorld”.InPhilosophicalPapers,126‐148.NewYork:CollierBooks.
Morrison,Margaret.1990.“Theory,InterventionandRealism.”Synthese82(1):1‐22.
Putnam,Hilary.1975.Mathematics,MatterandMethod.Cambridge:CambridgeUP.
Solomon,Miriam.2001.SocialEmpiricism.Cambridge,MA:MITPress.
Stroud,Barry.1968.“TranscendentalArguments.”TheJournalofPhilosophy65(9):241‐256.
Zagorin,Perez.1984.“Vico’sTheoryofKnowledge:ACritique.”ThePhilosophicalQuarterly34(134):15‐30.