are there indefeasible epistemic rules? · section 3 explains the indefeasibility view, and section...

19
volume 19, no. 3 january 2019 Are There Indefeasible Epistemic Rules? Darren Bradley University of Leeds © 2019 Darren Bradley This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License. <www.philosophersimprint.org/ 019003/> 1. Introduction What if your peers tell you that you should disregard your perceptions? Worse, what if your peers tell you to disregard the testimony of your peers? How should we respond if we get evidence that seems to un- dermine our epistemic rules? Several philosophers have argued that epistemic rules are indefeasible. I will argue that all epistemic rules are defeasible. The result is a kind of epistemic particularism, according to which there are no simple rules connecting descriptive and normative facts. I will argue that this type of particularism is more plausible in epistemology than in ethics. The result is an unwieldy and possibly in- finitely long epistemic rule — an Uber-rule. I will argue that the Uber- rule applies to all agents, but is still defeasible — one may get mislead- ing evidence against it and rationally lower one’s credence in it. Section 2 explains the problem of undermining and three possi- ble responses. Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6 contains the core of the positive proposal, arguing that the problem can be solved by understanding epistemic rules as hedged. Section 7 develops the proposal using a generalization of the concept of admissible evidence. Section 8 extends and defends the resulting position — where the only unhedged rule is a single Uber-rule. Section 9 compares our position to Quinean holism. Section 10 concludes. 2. The Problem of Undermining and Three Responses In this section I will explain the problem of undermining, and the re- sponse that epistemic rules are indefeasible. In the next section I will argue that this response is ad hoc. Consider some epistemic rules that have been discussed in the re- cent literature:

Upload: others

Post on 21-Mar-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

volume19,no.3 january2019

Are There Indefeasible

Epistemic Rules?

Darren BradleyUniversity of Leeds

© 2019 DarrenBradleyThis work is licensed under a Creative Commons

Attribution-NonCommercial-NoDerivatives 3.0 License. <www.philosophersimprint.org/019003/>

1. Introduction

Whatifyourpeerstellyouthatyoushoulddisregardyourperceptions?Worse,whatifyourpeerstellyoutodisregardthetestimonyofyourpeers?Howshouldwerespondifwegetevidencethatseemstoun-dermineourepistemicrules?Severalphilosophershavearguedthatepistemicrulesareindefeasible.Iwillarguethatallepistemicrulesaredefeasible.Theresultisakindofepistemicparticularism,accordingtowhichtherearenosimplerulesconnectingdescriptiveandnormativefacts. Iwillarguethatthistypeofparticularismismoreplausibleinepistemologythaninethics.Theresultisanunwieldyandpossiblyin-finitelylongepistemicrule—anUber-rule.IwillarguethattheUber-ruleappliestoallagents,butisstilldefeasible—onemaygetmislead-ingevidenceagainstitandrationallylowerone’scredenceinit.

Section 2 explains the problem of undermining and three possi-bleresponses.Section3explainstheindefeasibilityview,andsection4 argues against it. Section 5 argues against contributory rules, andSection6containsthecoreofthepositiveproposal,arguingthattheproblemcanbesolvedbyunderstandingepistemicrulesashedged.Section7developstheproposalusingageneralizationoftheconceptofadmissibleevidence.Section8extendsanddefendstheresultingposition—wheretheonlyunhedgedruleisasingleUber-rule.Section9comparesourpositiontoQuineanholism.Section10concludes.

2. The Problem of Undermining and Three Responses

InthissectionIwillexplaintheproblemofundermining,andthere-sponsethatepistemicrulesareindefeasible.InthenextsectionIwillarguethatthisresponseisadhoc.

Considersomeepistemicrulesthathavebeendiscussedinthere-centliterature:

Page 2: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –2– vol.19,no.3(january2019)

Third,thenormativeconceptsintheconsequentshouldbeunder-stoodintheepistemicsense,asopposedtothepragmaticoranyothersense.Forexample,ifyouareoffered$1mtobelieveinGod,thereisasenseinwhichyououghttobelieveinGod.Iamnotconcernedwiththispragmaticsenseof‘should’.

Fourth, thenormative concepts in the consequent shouldbeun-derstoodasrelativetotheagent’ssituation.Thereisasenseof‘ought’inwhichyououghttobelievepiffpistrue.5Thereisalsoasenseof‘ought’inwhichyououghttobelievepiffpisjustifiedinyoursitua-tion.ItisthislattersensewhichIwilluse.6

Fifth,Itake‘defeasible’tomean‘canberationallydoubted’.Sixth,Iwillcallruleslikethese“simple”todistinguishthemfrom

thecomplicatedUber-ruleIwilldefendinsections6and7.Seventh, the antecedents contain specific positive claims about

agents’states,butdonotcontaintotalityfactse.g. ‘theagenthasnootherperceptions/testimony’.

Ourproblemisthatitiseasytogeneratecaseswherenarrow-scoperulesconflictorunderminethemselves.Foranexampleinethics,sup-poseacrazedmurdererasksyouwheretheirtargetis.Theprinciplenottolieconflictswiththeprincipletopreventothersfromcomingtoharm.7Foranexampleinepistemology,supposeatablelooksredbutyouaretoldthatitisblue.PerceptionconflictswithTestimony.

Theepistemologyliteraturehasfocussedoncaseswherearuleisself-undermining.Startwithacaseofpeerdisagreement:

Supposethatyouandafriendindependentlyevaluateafactualclaim,basedonthesamerelevantevidenceandarguments.Youbecomeconfidentthattheclaimistrue.Butthenyoufindoutthatyourfriend—whosejudgmentyourespect—hasbecomejustasconfidentthattheclaimisfalse.Shouldthatnewsatallreduceyourconfidencein

5. SeeFassio(20182.b.)fordiscussionofthetruthnormofbelief.

6. Hedden2012p.344.

7. ThisisaproblemforKantianethics.SeeTimmerman(2013).

TestimonyIfanagent’ssituation1includestestimonythatx,thentheagentisrationallyrequiredtobelievethatx.(CompareElga[2007],Titelbaum[2015].)

Perception If an agent’s situation includes a perceptionthatx,thentheagentisrationallyrequiredtobelievethatx. (Compare Chisholm [1966], Huemer [2000], Pryor[2000],Boghossian[2008].)2

Credence-chance linkIfanagent’ssituationincludesfullbe-liefthatthechanceofxisy%,thentheagentisrationallyrequiredtohavecredenceinxofy%.(CompareLewis’s[1980]PrincipalPrinciple.)3

Let me make some clarificatory comments about these rules. First,these rules have descriptive (bywhich Iwillmean non-normative)conceptsintheantecedentandnormativeconceptsintheconsequent,makingthemnarrow-scoperules.4

Second,thesupervenienceofthenormativeonthenon-normativeentailsthattherewillbesomesuchnarrow-scoperules.Onecantakethemtobegroundingprinciples.

1. Theterm‘situation’isusedbymostauthorsinthisliterature.Titelbaumoffersadefinitionof‘situation’:“Iwillassumeonlythatwhateverthetruetheoryofrationalityis,itmayspecifycertainaspectsofanagent’scircumstancesasrelevanttodeterminingwhichoverallstatesarerationallypermittedtoher.Takentogether,theserelevantaspectscomprisewhatI’llcalltheagent’s‘situ-ation.’”Itakeitthisiscompatiblewithallauthors’usages.

2. It is controversialwhether this rule requires a different treatment, as howthingslookisnotapsychologicalattitude.Ithinksimilarruleswillstillholdfornon-attitudinalpsychologicalstates,butitwon’tmatterformyarguments.

3. Lewis formulated thePrincipalPrinciple in termsof conditional epistemicprobabilitiese.g.Cr(A|KnownchanceofA isxandE)=x,and Idonotintend todiverge fromthisversion. Ihaveusedaconditional toshowthesimilaritytotheotherrules.Moreonthisinsection7.

4. Narrow-scoperuleshavetheform[ifpthenyououghttobelieveq].Awide-scope rule would have the form ‘you ought to believe [if p then q]’. SeeBroome(1999).

Page 3: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –3– vol.19,no.3(january2019)

without believing Testimony. So Elga’s argument against TestimonyimplicitlyassumesthatrationalagentsareneverAkratic.

Although I’ve no interest in defending anti-Akrasia, I thinkAnti-Akrasiaisplausible,andthatwecannotblockElga’sargumentbysim-plyallowingAkrasia.9SoIwillgrantAnti-AkrasiaandblockElga’sar-gumentforadifferentreason.

Returning to themain thread, let’s distinguish three ways of re-sponding to the self-undermining problem. One might ignore col-leaguesandmaintainahighcredenceinTestimony.10Whenitcomestoyourbeliefinbeingconciliatory,youshouldstubbornlyignoreyourcolleagues.Conciliatorinessisindefeasible.Theresultisthatrational-ity haswhat Titelbaum calls “fixed points”—these are propositionsexpressing the rules of rationality, and they are indefeasible. Theseruleshavebuilt-inrestrictionstoensuretheycannotbeundermined,soTitelbaumcallsthem“restrictedrules”.

Asecondresponse,whichIwilldefend,istheviewthatyoushould bemovedbyyourcolleagues.Testimonymightstartoffwithahighprior,butyoucangetevidenceagainstitanddecreaseyourcredence,justlikeanyotherbelief.(Andifyourationallydisbelievethatsomeruleisarequirementofrationality,thenitdoesn’tapplytoyou.)Iwilldevelopthisviewusingthedistinctionbetweenhedged and unhedged

9. SeeHorowitz (2014) for anumberof powerful arguments againstAkrasia.Elga(2010)doesn’tmentionAkrasia;Titelbaum(2015)discussesAkrasiaindetail,butdoesn’tengagewithElga’sargument.Hereisanobjectiontomyview:“Yourpositionisthattherearenosimpleunhedgedrules;asAnti-Akra-siaisasimpleunhedgedrule,howcanyouassumeAnti-Akrasia?”Response:OntheviewIwilldevelop,ifAkrasiaiseverrational,itwillberationalonlyinhighlyunusualcases,e.g.whereyourationallybelievethatanexperttellsyou: ‘pandyoushouldnotbelievep’.This isastrangesituation(compareWorsnip2018p.24),soweshouldexpecttobeinastrangebeliefstate.NosuchsituationarisesinthecaseElgadiscusses,soIdon’tthinkallowingAkra-siaisagoodwaytoblockElga’sargument.(NotethattherulesIdiscussarenarrow-scopeepistemicrules,whileAnti-Akrasiaisawide-scopecoherencerequirement[Worsnip2018].)Thankstoarefereeforpressingthispoint.

10. Elga(2010)andTitelbaum(2015)defendthistypeofapproach.Elgawrites:“[O]neshouldbemovedbydisagreementaboutsomesubjectmatters,butnotaboutdisagreementitself…”(p.184).

thedisputedclaim?Conciliatoryviewsondisagreementanswer“yes.”(Elga2010p.175)

Elga(2007) initiallydefendedaconciliatoryview.Forsimplicity,wecanconsiderthemostextremeconciliatoryview:

Testimony Ifanagent’s situation includes testimony thatx,thentheagentisrationallyrequiredtobelievethatx.

ButElgafoundthatmanyofhiscolleaguesdisagreedwithhim.Elgawrites:

Suppose that youhave a conciliatory viewondisagree-ment, but you find out that your respected [colleague]disagrees.Hehasarrivedatacompetingview(aboutdis-agreement),andtellsyouallaboutit.Ifyourconciliatoryviewiscorrect,youshouldchangeyourview.Youshouldbe pulled part way toward thinking that your friend isright.Inotherwords,yourviewondisagreementrequiresyoutogiveupyourviewondisagreement.(p.179)

SoTestimony looksunstable: ifElga receives testimonyagainstTes-timony, andappliesTestimony,hemustgiveupTestimony.8Concil-iatory views sometimes call for their own rejection.This is the self-underminingproblem.

There is an important implicit premise in Elga’s argument weshouldmakeclear–Anti-Akrasia.

Anti-Akrasia:Itisneverrationaltobelieve[xandIshouldnotbelievex]

WithoutAnti-Akrasiaarationalagentcouldbelievetestimony(i.e.be-lievex ifsomeonetellsyoux)andalsobelievethatoneshouldnotbelievetestimony.Theformeristhelower-levelbeliefx;thelatterisahigher-levelbeliefaboutbeliefs.TheagentwouldfollowTestimony

8. SeeElga2010p.181–2andTitelbaum2015p.271.

Page 4: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –4– vol.19,no.3(january2019)

Hedged Unhedged

Absolute Me,Holton?Väyrynen?

Sections6,7and8

Elga,Lasonen-Aarnio,

Titelbaum.

Sections3and4

Contributory Endofsection6 Ross,Christensen?

Section5

Iargueagainstcontributory rules insection5. Iwilldefendhedgedrulesinsections6and7.InthenexttwosectionsIwillexplainandthenargueagainstrestrictedrules.

3. Restricted Rules

Elga (2010) defends the first option—restriction—arguing that weneedtomakeamodificationtoTestimonytomake it immunefromdefeat. Titelbaum (2015) builds on this and suggests that epistemicruleshavethefollowingform:13

Restricted Testimony: Ifanagent’ssituationincludestesti-mony that x, the agent is rationally required tobelievethatx—unless14xcontradicts[Testimony15].

IfxcontradictsTestimony,thenignorexandcontinuebe-lievingothertestimony.16

13. Titelbaum doesn’t actually defend Restricted Testimony. His point is thatepistemicrulesmusthavethisform,whatevertheyturnouttobe.

14. ‘Unless’means‘ifnot’.Roughly,ifthereisnocontradiction,thenbelievetes-timony.Noticethissaysnothingaboutwhattodoifthereisacontradiction.That’swhyweneedthenextline.

15. Theoriginal textsays “this rule”. Iassume ‘this rule’ refers toTestimony. Itwon’tmattermuchif itreferstoRestrictedTestimony.Iarguethatholdingsuchrulestobeindefeasibleisadhoc,whatevertheirexactcontentis.

16. I’ve added “If x contradicts Testimony, then ignore x and continue believ-ing other testimony.” Titelbaum isn’t explicit about how to respond if x

rules.11Hedged ruleshave ceteris paribus clauses, stating situationswhere the rule fails to apply;one such situation iswhereyou ratio-nallydisbelievethatsomeruleisarequirementofrationality.Iwillar-guethatrulesofrationalityarehedged—onecanalwaysgetevidenceagainstthembeingrequirementsofrationality.Theresultisthatthereareno“fixedpoints”,norulesthatrationalagentsshouldalwaysbe-lievetoberequirementsofrationality—allaredefeasible.

Athirdresponseisalsoworthdiscussing.ThereisadifferentwaytoweakenTestimony,suggestedbyChristensen(2010,2013).12Distin-guishabsoluteandcontributory rules.Absoluteruleshaveaconsequentthatsaysthatyouarerequiredtobeinsomestate,ortoperformsomeaction;contributoryruleshaveaconsequentthatsaysthatyouhavea reasontobeinsomestate,ortoperformsomeaction.Appliedtoepis-temology,absoluterulessaywhatyoushould(orshouldnot)believe,whereascontributoryrulessaywhatcountsinfavourof(oragainst)abelief.Testimonyisanabsoluterule;butconsideracontributoryver-sionofTestimony:

Contributory Testimony:Ifanagent’ssituationincludestes-timonythatx,thentheagenthasa reasontobelievethatx.

Paradoxisavoidedbecauseonecanhaveareasontobelievexandareasonnottobelievex.ButIwillrejectcontributoryrulesduetocasesof“valence-switching”,i.e.wherewhatisusuallyareasonforbecomesareasonagainst.

Hereisamapofthemainpositions,thesectionsinwhichtheyarediscussed and (tentative) suggestions forwhere somephilosophersmightbeplaced:

11. SeeHolton2002;Väyrynen2006,2009;Schroeder2004section5.

12. SeeDancy(2013section1).Christensen’s(2010p.203–4;2013p.92–3)talkof“ideals”couldbeunderstoodastalkofunhedgedcontributoryrules.Hisearlierwork(e.g.2007)suggeststhatthereisaresidualbad-makingfeatureofviolatingideals/rules,inwhichcasetheruleswouldnotbecontributory.

Page 5: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –5– vol.19,no.3(january2019)

inhercurrentsituation.Anagentcanreflectonhersitu-ationandcometorecognizefactsaboutwhatthatsitua-tionrationallyrequires.Notonlydoesthisreflectionpro-videherwithjustificationtobelievethosefacts;thatjus-tificationisultimatelyempiricallyindefeasible.17(p.276)

Butthisisveryhardtobelieve.Christensen(2013)writes:

Suppose…thatIfollow[ProperlyRestrictedTestimony]and remain absolutely confident in its correctness, de-spite the fact that it’s rejectedbymanyepistemologistsIrespect,andevenrateasmysuperiorsinphilosophicalskill.HowshouldIviewmyownreasoningonthistopic?ShouldI thinkthatwhileI’mgenerallyonlymoderatelyreliable when I think about philosophy, neverthelesswhen I think about arguments for general conciliation,andfornotbeingconciliatoryaboutconciliation,I’mes-peciallyimmunefromerror?Thatseemsextremelydubi-ous.(p.89)

Given the difficulty of formulating rational rules, the claim thatweshouldbecertain,orevenhighlyconfident,ofwhattheyare,eveninthefaceofopposingarguments,seemstomeuntenable.

Furthermore, suppose you do hear testimony against some (cor-rect)ruleofrationality.Titelbaumsuggeststhatyoushouldnotbelievesuchtestimonytoanydegree.Itnaturallyfollowsthatyoushouldnot

17. Titelbaumaddsafootnote,sayingthattherules“couldbeopposedbyempiri-calevidencepointingintheotherdirection….Butthosepropositionaljusti-ficationsareultimatelyindefeasibleinthesensethattheempiricalconsider-ationswillneveroutweighthemandmakeitall-things-consideredrationalfortheagenttoformfalsebeliefsaboutwhathersituationrequires.”Idon’tunderstandthis.Ifempiricalconsiderationscancountagainsttherules,whycan’twe imagine increasingly strong empirical considerations that eventu-allyoutweightherules?Andtherestrictionsheplacesontherulesseemtoensurethatempiricalconsiderationscannotcountagainstrationalrulesatall.Andhislaterclaim(section6)thattheFixedPointThesisleadstotheRightReasonsviewsuggeststhatempiricalevidencecannotcountagainstrationalrules.

RestrictedTestimony cannotundermine itself. It sayswe shouldbe-lievetestimonyonmosttopics,butnotwithregardtothequestionofwhetherweshouldbelievetestimony.ItfollowsthatourcredenceinRestrictedTestimonyshouldstaythesame,eveninthelightofoppos-ingtestimony.Thus,RestrictedTestimonyisindefeasiblebytestimony.

Titelbaum(2015)defendssimilarrestrictionsonallepistemicrules,e.g.:

Restricted Perception:Ifanagent’ssituationincludesaper-ceptionthatx,theagentisrationallyrequiredtobelievethatx—unlessxcontradicts[Perception].(p.273)

Andnotonlycanrulesunderminethemselves—theycanundermineeachother,e.g.youmightbetoldthatyourperceptionisunreliable.ToblockRestrictedTestimonyfromunderminingotherrules,Titelbaumdefends:

Properly Restricted Testimony: If an agent’s situation in-cludestestimonythatx,theagentisrationally…requiredtobelievex—unlessx contradicts [a] truthaboutwhatrationalityrequires.(p.274)

Thisstructureisintendedtogeneralizetoallrules,e.g.:

Properly Restricted Perception: If an agent’s situation in-cludesaperceptionthatx,theagentisrationallyrequiredtobelievethatx—unlessxcontradictsatruthaboutwhatrationalityrequires.

Titelbaumsuggeststhatweshouldhavecredence1intherulesofra-tionality;theyareindefeasible fixed points.Tousehismemorablephrase:mistakesaboutrationalityaremistakesofrationality:

[E]very agent possesses a priori, propositional justifica-tionfortruebeliefsabouttherequirementsofrationality

contradictsTestimony,butIthinkhemustbecommittedtothisline.I’llleavethisimplicitwhennotneeded.

Page 6: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –6– vol.19,no.3(january2019)

This isclose tosaying that fundamental rulesare those thatarenotdefeasible.Butthenthequestionis:Whyshouldwethinkthatthereareanysuchrules?IndeedIwillarguethattherearenone.

Elga argues that fundamental rules are dogmatic using the fol-lowingexample:Imagineamagazine,Consumer Reports,consistentlyrating itself as thebest consumermagazine.19When facedwith thecomplaintthatthemagazineisbiasedwhenitrecommendsitself,Elgaendorsesaneditorsaying:

Toputforwardourrecommendationsabouttoastersandcars is to put them forward as good recommendations.Andwecan’tconsistentlydothatwhilealsoclaimingthatcontrary recommendations are superior. So our alwaysratingourselves #1doesnot result fromanarbitraryoradhocexceptiontoourstandards.Weareforcedtorateourselves#1inordertobeconsistentwithourotherrat-ings.(p.185)

ButIthinkthisbringsouttheimportanceofthedistinctionbetweenfundamentalandnon-fundamental rules.Theadviceofa consumermagazineisnotaplausiblecandidatetobeafundamentalrule,anditiseasytoimaginecaseswhereConsumer Reportsallowsthatitsrec-ommendations canbedefeated. Suppose the editor ofConsumer Re-ports knowsthatherrivalmagazinehasalargerbudgetand,asaresult,makesrecommendationsbasedonmoreevidence.Thiseditorshouldrecommend that consumers rate this rivalmagazine’s recommenda-tionsoverthoseofConsumer Reports.Itwouldbeadhocfortheeditortoignoreevidenceagainsthermagazine’srecommendations.

It is only fundamental epistemic rules that are plausibly (non-ad-hocly) indefeasible. Elga needs to argue that there are simple funda-mentalepistemicrules,and,togettheconsequenceshewantsregard-ingthepeerdisagreementdebate,heneedstoarguethatRestricted

19. BasedonLewis(1971).

believeyouhaveany reason,nomatterhowweak,torejectthatruleofrationality.Weareledtotheviewthatwhat seems like evidence doesn’t even count as evidence. Lasonen-Aarnio18 is sympathetic to this view,writingthatit“restsmerelyonadesiretoavoidparadox”(p.342).Butthispositionseemsatleastasparadoxicalasanyofthealternatives.

Sowheredidwegowrong? Iwillargue in thenext section thatElga’s(2010) initialmovetoarestrictiononTestimony—amoveex-tendedbyTitelbaum—wasastepinthewrongdirection.

4. Elga’s Argument for Ubiquitous Indefeasibility

Whenfacedwitharulethatunderminesitself,Elgamodifiestherulesothatitignoresevidencethatthreatenstoundermineitself.Butthislooksadhoc.Whatreasondowehavetobelievethis,otherthanthefactthatitavoidstheunderminingproblem?Afterall,mostofourbe-liefsaresubjecttodoubtinthelightofopposingevidence,sowhyarebeliefsaboutdisagreementdifferent?

Elgaarguesthatall fundamentalepistemicrulesmustbenon-un-derminable.Hewrites:

Inordertobeconsistent,afundamental…rule…mustbedogmaticwithrespecttoitsowncorrectness.(p.185)

Butwhatdoes‘fundamental’meanhere?Elga(2010)writes:

afundamental[rule]isonewhoseapplicationisnotgov-erned…byanyother[rule]….(p.179)

18. “Itmaycomeasasurprise that insomecasesastatecanbeperfectlyepis-temicallyrationalevenifonehaswhatwouldseemlikestrongevidenceforthinkingthatitisnot”(Lasonen-Aarnio2014p.342,italicsadded).Althoughthenextsentencedoesseemtocounttheapparentevidenceasrealevidence.And she then says, “Thatone shouldbelieve thatone shouldn’tφdoesn’tentailthatoneshouldn’tφ”(p.343).SoitseemsLasonen-Aarniodoesacceptthat there is evidenceagainstφ,whichaffectshigher-levelbeliefsbutnotfirst-orderbeliefs,andthussheacceptsAkrasia.

Page 7: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –7– vol.19,no.3(january2019)

circumstances,hisdutytosavethechild[outweighs]21hisdutytomeetBetty.(p.256)

Let’ssaythatruleswhichcanbeoutweighedarecontributory.Andwecantakeitthatthedistinguishingfeatureofcontributoryrulesisthatthere is a residual bad-making featurewhen they are not followed,evenwhennotfollowingthemistherightthingtodo(duetoother,weightier rules).22When rules are voided, there is no residual bad-makingfeature.InthissectionIwillargueagainstcontributoryrulesinepistemology.

Inmeta-ethics,contributoryrulesareassociatedwithRoss(1930),whoarguedthatwehavenumerousduties,andwhatweoughttododepends on the overall weighting of these duties. In epistemology,theanalogousview is thatwehavenumerousepistemic reasons tobelieve,andwhatweepistemicallyoughttobelievedependsontheoverallweightofthesereasons.Wecanmakethisexplicitbyweaken-ingtheconsequentofourrules:

Testimony:Ifanagent’ssituationincludestestimonythatx,thentheagentisrationallyrequiredtobelievethatx

to

Contributory Testimony:Ifanagent’ssituationincludestes-timonythatx,thentheagenthasa reasontobelievethatx.

(Assume fornowthat the rule isunhedged.)Thisavoids theunder-miningproblemsabove.Anagentwhohastestimonythatxandaper-ceptionthatnotxhasareasontobelievexandareasontobelievenot x.23No paradox;what they should believe depends on the cor-rectweighingofthesereasons.Andanagentwhodoesn’tbelievethat

21. Frederickuses ‘overridden’but Iprefer ‘outweighed’,whichmakesexplicitthattheystillhaveweight.

22. Schaffer2015p.659.

23. Similarly, testimony against Testimony is compatible with Testimony stillprovidinga reasontobelievetestimony.

Testimony(orsomethingsimilar)isamongthem.20Butnosuchargu-mentsareoffered.SorestrictingTestimonytomakeitindefeasibleisadhocafterall.

Someonemightobjectthatwhatisfundamentalissomeotherrulethat is more complicated than Testimony (and Titelbaum takes nostandonwhatthefundamentalrulesare).Butoncewestartmakingqualifications,weareonthepathtotheUber-ruleandtheviewthatallrulesaredefeasible.ThisisthepathIthinkweshouldbeon.Beforepresentingmypositiveview, Iwillargueinthenextsectionagainstcontributory rules inepistemology. (Thoseuninterested in contribu-toryrulescanskipthenextsectionwithoutlossofcontinuity.)

5. Against Contributory Rules

Distinguishtwotypesofdefeasiblerules—rulescanbeoutweighed,ortheycanbevoided(andpossiblyboth).Frederick(2015)makesthedis-tinctionasfollows:

[N]ormally,ifAlfpromisesBettythathewillmeetherforlunch,AlftherebyacquiresadutytomeetBettyforlunch.However,ifAlf’spromisetoBettywasmadeunderthreatofforce,hispromisefailstoengenderthatduty,becausethecircumstancewasduty-voiding.

Normally, ifAlfhasthedutytomeetBetty for lunch,then he ought tomeet Betty for lunch.However, if onhiswaytomeetBettyheseesachilddrowninginapool,andhecansavethechildwithoutmuchrisktohimself,thenAlfhasadutytosavethechild….IfAlfcannotbothsavethechildandmeetBettyforlunch,hehastwodutieswhichconflict.PerhapsallwouldagreethatAlfoughttosavethechild;buthisdutytomeetBettyremains.Inthe

20.Thisisworthemphasizing.AlthoughElgaisnotexplicitthatconciliationismis a fundamental rule, itmust beone if Elga’s argument that fundamentalrulesareindefeasibleistohaveimplicationsforconciliationism.

Page 8: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –8– vol.19,no.3(january2019)

In the basic case, you are standing outside the library,whenyou seeTomGrabit exit,pull abook fromunderhisshirt,cacklegleefully,andscurryoff.ThisgivesyouprettygoodreasontobelievethatTomjuststoleabookfromthelibrary.

Case2 is just the sameas thefirst case, except thatTomhasanidenticaltwin,Tim,fromwhomyoucan’tvi-suallydistinguishhim.Inthiscase,ithasseemedtothejudgmentofmanyphilosophersthatyourvisualevidenceisnotareasontobelievethatTomstoleabook.(p.333–4)

ButSchroederthinksyourvisualevidenceremainsareasontobelievethatTomstoleabook.Todemonstratethis,heextendstheexample:

Considerathirdversionofthecase,exactlyliketheothertwoexceptthatinthethirdcase,inadditiontoTim,Tomhasathird identicalsibling,Tam,fromwhomyoucan’tvisually distinguish him. This third case underwrites acompellingargumentagainsttheintuitivejudgmentthatinthesecondcase,yourvisualevidencewasnoreasontobelievethatTomstolethebook.Forifyougoontocon-clude,inthethirdcase,thatTomstolethebook,thenyouaredoingworsethanifyouhadgoneontoconcludethisinthesecondcase.YourreasontobelievethatTomstolethebookthereforedoesn’tseemtohavegoneawayinthesecondcase;itmerelyseemstohavegottensubstantiallyweaker.(p.334)

Schroederseemstoberightaboutthiscase,wherethedefeaterweak-ensthestrengthofthereason.ButIdon’tseehowhecansaythesameaboutthepreviouscaseinwhichadrugmakesbluethingslookredandredthingslookblue.Inthatcase,thedefeaterchangesthevalenceofthereason,i.e.whatwasareasonforbecomesareasonagainst.Socontributoryrulesarestilltoostrong.

ContributoryTestimonyisarequirementofrationalitystillplausiblyhasareason(aweakone)tobelievetestimony.

Myobjectionisthatcontributoryrulessaythatafeaturethatisareasontobelievexisalwaysareasontobelievex.Anditisplausiblethat insomecasesa feature thatusually isa reasontobelievexbe-comesareasontobelievenotx.Anditisespeciallyplausibleinepis-temologywithregardtoPerception:

[I]nacasewhereI…believethatIhaverecentlytakenadrugthatmakesbluethingslookredandredthingslookblue,theappearanceofared-lookingthingbeforemeisreason forme tobelieve that there is ablue,not a red,thingbeforeme.Itisnotasifitissomereasonformetobelievethatthereissomethingredbeforeme,butthatassuchareasonitisoverwhelmedbycontraryreasons.Itisnolongeranyreasonatalltobelievethatthereissome-thingredbeforeme;indeeditisareasonforbelievingtheopposite.(Dancy2013)

These strongundercutting defeaters24provideanargument for reasons holism:“afeaturethatisareasoninonecasemaybenoreasonatall,oranoppositereason,inanother”(Dancy2004p.7).

TheconsequencehereisthateventhecontributoryversionofPer-ception isrefuted.Therearesituations inwhichtheperceptionofared-lookingthingisno reason at alltobelieveitisred.SimilarexamplesforTestimonycanbedevised,perhapswheretheagentfindsherselfinsidealogicpuzzleontheislandofLiars.Soweshouldrejectcon-tributoryepistemicprinciplesastoostrong.

Schroeder(2011)objectsthatthereisstillareason,justaweakerone.25Heusesthefollowingcase:

24.Undercuttingdefeaters suggest thatone’sground for thebelief isnot suffi-cientlyindicativeofthetruthofthebelief—thegroundherebeingtheexperi-ence.SeePollock1967.

25. Schroeder(2011fn.8)doesnotactuallydenyreasonsholism.Nevertheless,theargumentheofferscanbenaturallyunderstoodasanargumentagainstreasonsholism.

Page 9: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –9– vol.19,no.3(january2019)

6. Hedged Rules

Let’srecap.SofarIhavearguedthatitisadhoctoholdthatsimpleepistemicrulesareindefeasible,andthatepistemicrulesarenotcon-tributory.Mypositiveviewis:For any (simple or complex) epistemic rule, rational agents can acquire evidence that it is not a requirement of rationality, causing them to decrease their credence that it is a requirement of rationality. Ifthecredenceislowenough,theruledoesnotapplytotheagent(byAnti-Akrasia).

Let’sconnectthistofamiliarviewsinethics.RecallFrederick’sex-ample: ifAlf’spromise toBettywasmadeunder threatof force,hispromise fails toengender thatduty,because the situationwasduty-voiding.So,startingwithasimpleethicalrule:

ifyoupromisedtop,thenyouarerequiredtop

it isplausible thatyouhaveno reason top if thepromisewasmadeunderduress.Thesimpleethicalrulecanbevoided.Inordertoallowforthis,weneedthefullruletobe:

ifyoupromisedtop,thenyouarerequiredtop,unlessthepromisewasmadeunderduress.

We’llsaythatthefullruleishedged.28

Inepistemology,usingtheexampleofPerception,abetterspecifi-cationoftheruleswouldmoveusfrom:

28.This idea has been suggested inmeta-ethics byHolton (2002) andHorty(2007).Hortywrites that “thegeneralprinciple that lying iswrongshouldbe taken tomean simply that lying iswrongbydefault—that is, to afirstapproximation,thatoncewelearnthatanactioninvolveslying,weoughttojudgethat it iswrong,unlesscertaincomplicatingfactors interfere”(p.23).Holtonsuggests thatethical rulesneed ‘That’s it’clausesstating that therearenoother ethically relevant features, e.g. ‘Anyaction thathas such-and-suchfeaturesandThat’sIt iswrong’.Onechoice-pointhereiswhetherthehedgelistsamanageablenumberofexceptions(e.g.‘underduress’)orisaplace-holderforanopen-endedlistofexceptions(e.g.‘andtherearenootherrelevantnormativefeatures’).Thisisthetopicofsection8.SeeField(2000p.135andAppendix)forrelatedpoints.

Furthermore,amotivationforwantingcontributoryreasonineth-icsdoesnotapplytoepistemology.Astrongmotivationforcontribu-toryreasonsinethicsisaneedforoutweighedreasons.Dancywrites:

Scanlon[whoistakentorejectcontributoryrules26]has…deprivedhimselfoftheideaofadefeatedreason,andthereby prevented himself even from addressing thequestionwhattheappropriateresponseistosuchathing.Normallywewouldspeakof regretand residualduties,but if all conflict is, as Scanlon suggests,merely appar-ent, there arenodefeated [outweighed] considerationscapableofdemandingregret,andnothingtogeneratearesidualduty.27(Dancy2004p.26)

The idea that contributory reasons keep their force even when de-featedismostplausibleinethics,especiallyifwethinkofourethicalsystemasconsistinginduties.Ourdutiesremaineveniftheyarede-featedbyotherduties.Yourdutynottolieremains,evenifyouhaveastrongerdutytosavealifebylying.

By contrast, our epistemic duties do not seem to have the samepowertoremainevenifdefeated.Forexample,supposeyouseeared-lookingtable inasituationwhereyouknowyouhavetakenadrugthatmakesbluethingslookredandredthingslookblue.Youdon’tbelieve that the table is red.Do you regret that youhavenot liveduptoyourepistemicdutytobelievethingsareastheyappeartobe?Surelynot.Asignaturefeatureofaviolationofacontributoryruleisthatitinvolvesaresidualbad-makingfeature.Buttheredoesn’tseemtobeanythingbadaboutbelievingthetableisblue.Sothecontribu-toryviewlookstobeunmotivatedinepistemology.Sotherejectionofcontributoryrulesismoreplausibleinepistemologythaninethics.

26. IamneutralonwhetherthisisagoodinterpretationofScanlon.

27. This argument is central to Dancy’s position. He later (p. 28–9) rejectsHolton’sPrincipledParticularismforthesamereason.

humdb
Inserted Text
s
humdb
Cross-Out
humdb
Inserted Text
(epistemically or morally)
Page 10: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –10– vol.19,no.3(january2019)

response,Isuggestthattheagentcanmovestraighttoastablestateinwhich E has higher credence than before and that-Perception-is-a-requirement-of-rationality-in-this-situationhaslowercredencethanbefore.

Sosituationswhererulesappear toconflictare in factsituationswheretherulesareincompletelyspecified.Thisavoidsthesmellofadhoccery—wearenotassertingthatrulesareimmunetoconflictingev-idence;weareplacinglimitsonwhentheruleapplies,amovewhichis familiar inethicsandphilosophyofscience.31 It isalsofamiliar inepistemology—itisthestrategysuggestedbyLewis(1980)regardingobjectivechancewhichIdevelopinthenextsection.

7. Two Types of Inadmissibility

Let’sstartwithasimplecredence-chancelink:

Credence-Chance Link:Ifanagent’ssituationincludesfullbeliefthatthechanceofpisx,thentheagentisrationallyrequiredtohavecredenceofpinx.

Thisruledoesnotapplyinallsituations.Anagentcanhaveevidencethatjustifiestheirhavingcredencesthatdifferfromthechances.Callsuch evidence inadmissible. Themost familiar form is evidence thatgivesusdirectinformationabouttheevent.Forexample,ifyouseeafaircoinlandingHeadsinacrystalballyouknowtobereliable,youshouldnothavecredenceof½thatitwilllandHeads.Thisisanoppos-ingdefeater.SotheCredence-ChanceLinkshouldbehedged.Indeed,LewisdefendedahedgedversionoftheCredence-ChanceLink—thePrincipalPrinciple(PP),ofwhichwe’lluseasimplifiedversion:32

31. CompareCartwight(1983),PietroskiandRey(1995).

32. Precisely,itsays:“LetCbeanyreasonableinitialcredencefunction.Lettbeanytime.Letxbeanyrealnumberintheunitinterval.LetXbetheproposi-tionthatthechance,attimet,ofA’sholdingequalsx.LetEbeanyproposi-tioncompatiblewithXthatisadmissibleattimet.ThenC(AIXE)=x”(p.266).Thisbrings invarious featureswhicharen’t relevant toour concerns.Onethat isworthmentioning is thatLewis’s rule requires that agentsnotonlyhavecredencesthatmatchtheknownchances,butalsoupdateinsuchaway

Perception: If an agent’s situation includes a perceptionthatx,thentheagentisrationallyrequiredtobelievethatx

to

Hedged Perception:Ifanagent’ssituationincludesapercep-tionthatx,thentheagentisrationallyrequiredtobelievethatx,unlesstheyhaveevidenceagainstPerception.29

WhatifanagentdoeshaveevidenceagainstPerception?IsuggestwetakeHedgedPerceptiontoinclude:

IftheyhaveevidenceagainstPerception,30thentheircre-dence thatPerception is a requirementof rationality inthissituationshoulddecrease.

Where there ismoderate evidence that Perception is not a require-mentofrationalityinthissituation,therationalagentmightretainamiddlingcredence thatPerception isa requirementof rationality inthissituation. WherethereisoverwhelmingevidencethatPerceptionisnotarequirementofrationalityinthissituation,therationalagentmightdisbelievethatPerceptionisarequirementofrationalityinthissituation. Assuming Anti-Akrasia, (which links beliefs about the re-quirementsofrationalitywithrequirementsofrationality),itfollowsthatPerceptionisnotarequirementofrationalityinthissituation.Afortiori,Perceptionisnotarequirementofrationalityinallsituations.

A referee objects thatwhere believing E is based on Perception,whichisthenunderminedbyE,thesupportforEvanishes,thustheruleisbelievedagain,andthepositionisdiachronicallyunstable.In

29.NoticethatalthoughHedgedPerceptioncontains‘unless’,itisverydifferentfromTitelbaum’s restricted rules.Restricted rules tell us to ignore evidence thatconflictswiththerules;hedgedrulestellustolowercredenceintherules (inoursituation)whenthereisevidenceunderminingthem.

30.IintendthistocovercaseswheretheevidenceisthatPerceptionfailstoap-plyintheircurrentsituation,andcaseswheretheevidenceisthatPerceptionfailstoapplyinallsituations.

Page 11: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –11– vol.19,no.3(january2019)

to have credence of x in p, unless they have evidence-inadmissible-relative-to-the-Credence-Chance-Link, or unless they have evidence-inadmissible-relative-to-the-PP*.

Isuggestthatotherrulesworkthesameway,andthatweneedtogen-eralizethenotionofinadmissibilitytoapplytotheotherrules.

Let’sworkthroughhowthisappliestoTestimony.

Testimony:Ifanagent’ssituationincludestestimonythatx,theagentisrationallyrequiredtobelievethatx.

Like theCredence-ChanceLink, thisneeds tohaveexceptionsbuiltin.Onetypeofexceptionoccurswhentheagenthasotherevidenceagainstx.Perhaps theagenthasdirectly seen thatx is false; this isanopposingdefeater.Adifferent typeof exceptionwouldbe if theagentrationallybelievesthatthepersontestifyingisunreliable;thisisanundercuttingdefeater.OrsupposeanapparentlyreliableagenttellsthemthatTestimonyisfalse;thisisamoregeneralundercuttingdefeater,relevanttoothertestifierstoo.Callevidencethatjustifiesanagent innotbelieving testimony inadmissible-relative-to-Testimony.SoTestimonyshouldbehedgedasfollows:

Hedged Testimony: If an agent’s situation includes tes-timony that x, then the agent is rationally requiredto believe that x, unless they have evidence that is inadmissible-relative-to-Testimony.

Mutatismutandisforothersimpleepistemicrules(includingHedgedTestimony).Thus,Isuggestthatallsimpleepistemicrulesarehedged.

Theunderminingproblemsdisappear,astheproblemcasesdescribedsituationswheretheagentdoeshaveevidenceinadmissibletosomerule,sotheagenthaslowcredencethattheruleisarequirementofrationality,sobyAnti-Akrasia(whichlinksbeliefsabouttherequire-mentsofrationalitywiththerequirementsofrationality),theruledoesnotapply.34Wehavearrivedatmyviewthatepistemicruleswillbe

34.Objection: Our hedged simple rules will almost never apply. Hedged

PP If an agent’s situation includes full belief that thechance of p is x, then the agent is rationally requiredto have credence of p in x,unless they have inadmissible evidence.

Butthereisasecondtypeofinadmissibleevidencethathasnotbeendiscussed in the objective chance literature—theCredence-ChanceLinkcanhaveundercuttingdefeaters.Theremightbeevidencethatre-ducesyourconfidencethataruleisarequirementofrationality.33Sup-poseanapparently reliableagent tellsyou that chance isnot some-thingwhichyourcredencesshouldmatch.Youmightbeconfusedbysuchastatement,butthisconfusionissurelyenoughtojustifyyourcredences’notperfectlymatchingthechances.SuchevidencereducesconfidenceintheCredence-ChanceLink.

Similarly, supposeanapparently reliableagent tellsyou that thePP is false (i.e. evenwith thehedge). Then you should lower yourcredencethatPP isaruleofrationality.SoPPalsoneedstobehedged.Just as the hedge of theCredence-Chance Link generated the PP,weneedanewprinciplegeneratedbythehedgeofthePP.

Oneway tosystematizeall this is tomake theconceptofadmis-sibilityrelativetoarule.SowecanrestatePPas:

PP*: If an agent’s situation includes full belief thatthe chance of p is x, then the agent is rationally re-quired to have credence of x in p, unless they have evidence-inadmissible-relative-to-the-Credence-Chance-Link.

AndPP*needstobehedgedinturn:

Qualified PP*:Ifanagent’ssituationincludesfullbeliefthatthechanceofpisx,thentheagentisrationallyrequired

that,givenanypossibleadmissibleevidence,theywillcontinuetodoso.Wecouldsetupallourrulesinthesameway,butIwillusethesimplerformula-tion.Thankstox.

33. Suchevidencemostdirectlyreducesyourconfidencethataruleisarequire-mentofrationalityinyoursituation;afortioriitreducesyourconfidencethataruleisarequirementofrationalityinallsituations.

Page 12: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –12– vol.19,no.3(january2019)

epistemicrules,thispositionisstilldefensibleasanepistemictheory.SointhenextsectionIwilldefendanUber-rule.

Beforethat,thereisalooseendtotieup.Supposeweaccepthedgedrules—aretheycontributoryorabsolute?Intheethicscase,weused:

Hedged, Absolute: If you promised to p, then you are re-quiredtop,unlessthepromisewasmadeunderduress,etc.

Butthealternativeis:

Hedged, Contributory:Ifyoupromisedtop,thenyouhaveareasontop,unlessthepromisewasmadeunderduress,etc.

The advantage of contributory rules is that they don’t underminethemselvesoreachother;butwehavenoneedofthisadvantagehere,asthehedgecanensurethatthereisnoconflict.Andifthereisnocon-flictingrule,thensurelyapromisetopmeansthatyouarerequiredtop;theHedged,Absoluteruleiscorrect.

Theepistemologycaselookssimilar.

Hedged, Absolute:Ifitlooksred,thenyoushouldbelieveitisred,unlessyourationallybelieveyouhaverecentlytakenadrug,etc.

Hedged, Contributory: If it looksred,thenyouhavearea-sontobelieveitisred,unlessyourationallybelieveyouhaverecentlytakenadrug,etc.

ThestrongerHedged,Absolute rule looksplausible, so I tentativelyendorseit.

8. For the Uber-rule

SofarIhavearguedthatthelinkbetweendescriptiveandnormativeconceptscanbedescribedonlybyatapestryofinterlockinghedged

notthesimpleruleswithwhichwebegan,butcomplexrulesfullofhedges.

Dancy(2004)attributesananalogousviewtoScanlon(1998):35

Scanlon’s view… seems to be that… there are no ac-tualconflicts,onlyappearancesofconflict.Iftwoofour[rules]seemtogetineachother’sway,whatthisshowsisthatatleastoneofthemisincompletelyspecified,andthematter isresolvedbyamorecompletespecification.Supposewe[canhelpsomeoneinneedatthecostofkill-ingsomeoneelse].36Theideahereisthatinaproperun-derstandingofthe[rule]thatrequiresustohelpthoseinneed,therewouldprobablybeincludedanexceptiontothatdutyforallcaseswheretohelponewehavetokillanother.Properlyunderstood,therefore,thedutytohelpcannotconflictwiththedutynottokill.(p.25)

But a new danger emerges if the complexity spirals out of control.Startingwithsimplerules,cantheexceptionsbefinitelystated?Ideally,wewouldliketohavefiniteexceptions,asthiswouldallowamanage-ablesetofrulesthatcouldbeusedtoguideourdeliberation.

I don’t know if this is possible, so Iwill concede the point, anddefendthepossibility that theexceptionsareopen-ended.The ideaisthat,evenifweareleftwithaninfinitelistofexceptionstosimple

Perceptionmightstart:“Ifitlookslikep,thenbelievep,unlessyouhaverea-sontodistrustyourvisionor….”Anyonewhohaseverexperiencedanillu-siondoeshavesomesuchreason.Thispointiscorrect—hedgedsimplerulesmight rarelyapply toanyactualpeople—but this isnotaproblem.Again,comparethePrincipalPrinciple.Lewisformulatedittoapplyonlytoinitialcredencefunctions,soitdoesnotapplytoanyrealpeople.Nevertheless,itis(ifcorrect)asubstantiveandinformativeruleofrationality.

35. DancycitesScanlon(1998p.197–200).ButDancyadmitsthattheinterpretiveclaimisnotbeyonddispute,andIrefrainfromattributingthistoScanlon.

36.Bizarrely,theoriginalsentenceis“Supposewefaceachoicebetweenkillingonepersonandhelpinganother.” I take itDancy intended todescribe theexampleIuse.

Page 13: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –13– vol.19,no.3(january2019)

so theprior distribution encodeswhat agents shouldbelieve givenanyevidence.ThepriordistributioncanencodeanUber-rule.

Arethereanysimplerulesthatapplyinallsituations,i.e.allovertheVenndiagram?Iseenoreasontoexpectso.Wecannotsayyoushouldbelieveyour senses,because there ispossibleevidence/ar-easof theVenndiagramwhereyoushouldnotbelieveyoursenses.Indeed,foranysimpleruleyoumightstate,thereispossibleevidence/areasoftheVenndiagramwhereyoushouldnotfollowthatsimplerule.ThebestwecouldhopeforwouldbepatternsinsomeareasoftheVenndiagramthatcouldbehelpfullydescribedwithsimplerules.

WemightmakeacomparisonwithHumeanlaws.FortheHumean,lawsdonotexplainevents;eventsarefundamentalandlawsarede-rived fromevents, soHumean lawsare just informative summaries.Similarly,inBayesianepistemologythepriorsarefundamental;somesimpleepistemicrulesmightbederivedfromthepriors,but,likeHu-meanlaws,theyarejusthelpfulsummaries.39

Forafurtherintuitiveargument,considersomedifficultquestion,suchas thecorrect credence that climatechange isman-made. Is itplausiblethatafinitenumberofsimpleepistemicruleswouldgener-atetherationalcredence?Itstrikesmeasentirelyimplausible.

Intherestofthissectionwe’llconsidertwoobjectionstotheUber-rulebasedon(i)guidanceand(ii)coherence.

8.1. GuidanceLasonen-Aarnio (2014) offers the most detailed discussion, and

39.Thussimplerulescanstillhelpexplainrationalrequirements;wejusthavetorememberthatthesimplerulesarethemselvesexplainedbythepriors.SothisseemscompatiblewithChristensen’scomment:“Ifweaskwhythedis-agreementofothercompetentthinkerswiththesameevidenceshouldaffectmyconfidence,thecorrectexplanationmaystillbethatsincetheirdisagree-mentisevidencethatmyinitialbeliefwasbasedonanepistemicerror,itcre-atesrationalpressuretogivecredencetotheclaimthatmyinitialbeliefwasbasedonerror,andthat…thiscreatesrationalpressuretobackoffofthatinitialbelieftoatleastsomeextent”(Christensen2013p.93,italicsadded).However,Christensen(2010p.203–4)offerssomeconsiderationsthatsug-gestthatsomesimplerulesareexplanatoryinastrongersensethanIallow.

rules.Cantheserulesbefinitelystated?Ifnot,wehaveaversionofparticularism:

Principled Particularism:Anyfinitesetof ruleswillbe in-sufficienttocaptureallnormativetruths.37

Thustheconnectionbetweendescriptiveandnormativetruthsisex-pressibleonlywithaninfinitelylongrule:anUber-rule.WiththisoneUber-rule,whichpresumablydoesnotundermineitself,theproblemsofunderminingareavoided.

IwilldefendtheUber-ruleinepistemology.AsfarasIcantell,theterm‘Uber-rule’wasintroducedbyChristensen(2010)anddescribedasfollows:

Suppose we specify, for every possible evidential situ-ation inwhich an agentmay find herself,what the ap-propriate doxastic response is. The resultwould be anoverarching rulewhich took into account every sort ofevidence.Wemight thenthinkof thatruleasencodingtheoneandonlytrueepistemic[rule].(p.203)

IaddthattheUber-rulecannotbefinitelyexpressed.Thispositioncanbegeneratedbyconjoininganinfinitenumberofsimplerules,orpos-itingafinitenumberofsimpleruleswithatleastoneinfinitelylonghedge.AstheUber-ruleappliesinallsituations,itisunhedged;Iwillarguebelow(8.2)thatitisalsodefeasible.

SofartheonlydetaileddiscussionsoftheUber-rulehavearguedagainst it.38 Let me sketch a way of thinking about rationality thatmakesanUber-ruleplausible.Insteadofthinkingaboutrationalityasemergingoutofsimplerules,thinkoftheBayesianapproachwhereagents begin with a prior distribution of probabilities. Imagine allepistemicallypossibleworldsonavastVenndiagram.Bayesiansonlyallow updating by conditionalization (or Jeffrey conditionalization),

37. CompareHolton(2002)andMcKeeverandRidge(2006p.16).

38.Christensen(2010,2013),Lasonen-Aarnio(2014).

humdb
Inserted Text
Footnote: Conditionalization could be a fundamental rule, one which uses totality facts about agents' entire epistemic state.
Page 14: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –14– vol.19,no.3(january2019)

However,thispositionmightbehardertodefendinepistemologythaninethics.Inethics,oftenonlyasmallnumberofconsiderationsarerelevanttowhetheraparticularactionisright.But,turningtoepis-temology,considerwhetherIshouldtrustmyvisionandbelievethereisaredtableinfrontofme.Presumablythetrack-recordofmyvisionisrelevant,soeveryvisualexperienceinmylife—andwhetheritwasveridical—willberelevant.Andwealreadyseemtohavearulethatistoocomplicatedtoguideme.

Atthispointwecouldretreattotheviewthatalthoughwecannev-er(ornotalways)saythattherearenootherrelevantfeatures,theseotherrelevantfeaturescanberationallyignored.Defendingtheviewthattherearemorereasonsthanwenormallytakeintoaccount,MarkSchroeder(2005)writes:

IfGodmade a list of all of the pros and cons of somecourseofaction,itmightbeinfinitelylong.Butyoucan’tpossiblytakeeverythingintoaccount—onlythereasonsnearthetopofthelist.(p.15)

Sowecouldbeguidedby theUber-rulebypayingattention to themostimportantpartsoftheruleforoursituation.

Aproblemmightremain if someone insisted thatguidancemustbeprovidedbythefullepistemictheory,i.e.thewholeUber-rule.ThefullUber-rulecannotbefinitelyformulated,andperhapswecannotbeguidedbyruleswecannotformulate.

Butanagentmightbeguidedbytheruleswithoutbeingabletoformulatethem.Forexample,whenyoujudgethatasentenceisun-grammatical,youareguidedbylinguisticrulesthatyouareunabletoformulate.

Furthermore,youmightevenbedisposedtogettheruleswrongifyoutriedtoformulatethem.Forexample,Arpaly(2003)arguesthatHuckleberryFinnisguidedbythetruemoralrulesinnotturninginanescapedslave,evenifhewronglybelievesthatheshouldturnin

focussesontwomainworries.StartwiththeworrythatwecannotbeguidedbytheUber-rule:40

Now,theproblemfortheUber-ruleviewisthatanUber-rulejustdoesn’tseemlikethekindofrulethatcanoffergenuineguidance.Forone,itcannotevenbeexpressedasasetoffinite, informativegeneralisations.…Even ifonearguesthatsubjectsmanagetogenuinelyfollowtheUber-rulebyemployingmoreordinarykindsofepistemicrulesasheuristicguides,thefactremainsthattheyneedguidancetofollowtheUber-ruleitself.Hence,theUber-rule isaveryawkwardcandidate forarulethat is itselfsupposed toplay the roleofofferinggenuineguidance.(p.333)

InmyviewtheUber-ruleneednotofferguidance.IprefertheBayes-ianviewabove,accordingtowhichagentsare“guided”bytheirpriorsandtheevidence.

Still, Lasonen-Aarnio’s worry may survive as the worry that wemeremortalsareunabletobeguidedbythefullidealpriorprobabilityfunction,whichrequiresassigningprobabilitiestoaninfinitenumberofpropositions.Solet’saddressthisworryintheformthatLasonen-Aarnioputsit—thattheUber-rulecannotofferguidance.

Inresponse,itisarguablethatanUber-rulecouldofferguidance.Inthehappiestcases,agentsknowthattheceterisparibusclauseofasimpleruleistrue,i.e.otherthingsareequal,sotheycanbeguidedbytherule.Forexample,someonewhohasconclusivereasontobe-lievethattheirsensesarereliablecanfollowPerception,andPercep-tionwillbeonepartofthelargerUber-rule.ThoughneverguidedbytheentireUber-rule,agentsmightusuallybeguidedbythepartoftherulethatisrelevanttothem—itlooksred,therearenootherrelevantfeatures,sobelieveitisred.41

40.CompareBoghossian2008p.496.

41. SeeVäyrynen(2008)foradetaileddefenceofasimilarpositioninmetaethics.

Page 15: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –15– vol.19,no.3(january2019)

Uber-ruleinthatsituationare.…Nowimaginethatyouhearanepistemologyoracletellyouthattherecommen-dationsmadebytheUber-ruleintheverysituationyouareinrightnowareincorrect.Insofarastheruleiscom-plete…thechartmustsaysomethingaboutyourcurrentsituation.Imaginethat,asthecharttellsyou,therulerec-ommendsbeinginstateS.But insofarastheoracle istobetrusted,doesn’thertestimonyactasahigher-orderdefeaterforanysuchrecommendation?(p.331)

Asstated,Ithinkthisimaginedsituationisincoherent.TheproblemisthatweareimaginingacasewhereyouaretoldthattheUber-ruleisincorrect.Butwehavedefined‘Uber-rule’asthecorrectrule.Soanyrulethatfailstobecorrectisn’ttheUber-rule.SoanyspeakerwhotellsyoutheUber-ruleisincorrectissayingsomethingincoherent,andyoushouldnotfollowtheiradvice.44Evenworse,we’vebeentoldthatitisan“epistemologyoracle”whoistellingustheUber-ruleis incorrect.Presumably,thephrase‘epistemologyoracle’appliesonlytosomeonewho speaks the truth. So ‘epistemology oracle’ cannot be correctlyappliedtoanyonewhotellsustheUber-ruleisincorrect(underanymodeofpresentation),astheUber-ruleisbydefinitioncorrect.

Acoherentscenario inthisarea is thatarationalagentmightbetoldsomething falseabout theUber-rulebyaplausible-lookingbutmisleadinginformant.AndwhatmightbeworryingLasonen-Aarnioisthethoughtthat,inthisscenario,agentscanhaverationaldoubtsaboutthecontentoftheUber-rule.ItlooksparadoxicalfortheUber-ruletoapplyinallsituations,beunhedged,andyetbedefeasible.45

Ithinkparadoxcanbeavoided.Let’sworkthroughacase.Supposethesumtotalofyourevidenceisacurrentredexperience.Supposethe

44. Theymightsay,“Theruleyouarefollowingisincorrect”,butthat’sadifferentsituation,asitinvolvesadifferentmodeofpresentationoftheUber-rule.

45. CompareChristensen(2013):“IftheagentcontinuestofollowtheUber-rulewhiledoubtingitscorrectness,itseemsinevitablethatshewillinsomecasesviolate [Anti-Akrasia]” (p. 93).Theexample showshowwe candoubt theUber-rulewithoutAkrasia.

theslave.42HuckleberryFinn’sinabilitytoformulatetherulesdoesnotstophimfrombeingguidedbytherules.

PerhapsourpositioninepistemologyisanalogoustoHuckleberryFinn’spositioninethics.WecannotformulatetheUber-ruleanymorethanHuckleberryFinncan formulate theethical rules.But it ispos-siblethathiscompassionmakeshimperformtherightaction,sothereisasenseinwhichheisguidedbyethicalrules.Similarly,itispossiblethat our good sense, or epistemic intuition,makesus form rationalbeliefs,sothereisasenseinwhichweareguidedbytheUber-rule.43

Tosumup,althoughtheUber-ruleisincompatiblewithsomeviewsthat involvestrongrequirementsonourability to formulateandbeguidedbynormativerules,suchstrongrequirementscanberejected.

8.2 CoherenceLasonen-Aarnio’sotherworryisthattheremightbenoUber-ruletobe found—“findinga rulenot susceptible todefeat is surelyharderthanmerelydefiningonetobesuch!”(p.331).TheideaseemstobethattheconceptofanUber-rulemightbeincoherent,justasthecon-ceptofasquarecircleis.

Onemight immediatelybe suspicioushere—surely, for anypos-sibleevidentialstate,thereisarationalresponse,andtheUber-rulestateswhattherationalresponsesare.Nevertheless,Lasonen-Aarniooffersthefollowingcasetodefendthisposition:

[A]ssumethatyouarestaringatachartrepresentingtheUber-rule: for each possible epistemic situation… thechartspecifieswhat therecommendationsmadebythe

42. SeeRaz(2000) for furtherdiscussion.Onemightbe tempted toappeal tothefamiliardistinctionbetweenatheoryofrightnessandadecision-makingprocedure(Bales1971).Butweareworkingonlywiththesubjectiveought,sothetheoryofrightnesslooksirrelevant(orperhapsbetter:thedistinctioncollapses).

43. Adifferentproblemwith an infinitely longUber-rule is thatwewouldbeunabletograspepistemicconcepts.Thus,Jackson,PettitandSmith(2000)giveasemanticobjectiontoparticularism.Thisseemstobeagoodreasonforpositingalongbutfiniterule.

Page 16: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –16– vol.19,no.3(january2019)

AsafinalpointindefenceoftheUber-rule,itisworthemphasiz-ing Holton’s point that Principled Particularism is compatible withutilitarianism:

[S]upposeyouwereautilitarian.Thenyoucouldn’tde-terminewhichactiontoperformonthebasisofalistofpleasuresandpainscausedbypossibleactionsofyours.Youwouldneedtoknow,inaddition,thatthesewereallthepleasuresandpainsthateachactioncaused;andthatthesewereallyourpossibleactions.(p.206)

Nomatterhowmuchof theworldwedescribe, it isnot enough toensure that an act is right—rightnessdependson thewholeworld,justasgeneralizationslike ‘allswansarewhite’do.47Similarly, ifwefocusonnarrow-scoperuleswithdescriptiveantecedentswithoutto-talityfacts,thenwecanneverstopatfiniterules—whatyououghttobelievedependsonyourwholementalstate.Butthisdoesn’truleout

Thisfollowsfromtheinternalistintuitionthatagentshaveepistemicaccesstotherationality-makers.Butonecanhaverationalfalsebeliefsaboutwhatothersshouldbelieve:

Titelbaum’s-Principle-I-Deny:Foranygivenagentandsituation,iftheyare mistakenaboutwhatoneshouldbelieveinasituationotherthantheir own,thentheyaremakingamistakeofrationality.

Thedifferencecanbeput in termsof thescopeof thequantifiers.Foranyagentandsituation, thereareepistemic rules theyought tobelieveare re-quirementsofrationality;buttherearenoepistemicrulestheyoughttobe-lievearerequirementsofrationalityinallsituations.Anotherwayofputtingthisisthatyoucanberationallymistakenaboutwhatsomeoneelseshouldbelieve, but not aboutwhat you should believe.Why the asymmetry? Be-causewhatanagentshouldbelievedependsontheir ownhigher-orderbeliefsaboutrationality,whereaswhatanagentshouldbelievedoesnotdependonsomeone else’shigher-orderbeliefsabout rationality.TheasymmetrycanbereadofftheAnti-Akraticrule(thisisanadhominemcriticismofTitelbaum,asAnti-Akrasiaiscentraltohistheory)—itisirrationaltobelievepandbe-lievethatitisirrationaltobelievep;itisnotirrationaltobelievepandbelievethatitisirrationalforsomeoneelsetobelievep.Titelbaum(2015)mentionsthispossibleasymmetrybutimmediatelyrejectsit:“[E]veryplausiblestoryI’vebeenabletocomeupwithisgeneralizable:itappliesjustaswelltoanagent’sconclusionsaboutwhat’srationallyrequiredinothersituationsasitdoestoconclusionsaboutwhat’srequiredinhercurrentsituation”(p.276).

47. CompareSchroeder2011.

Uber-rulesaysthattherationalresponsetothisevidenceistobe90%certainthatthereisaredobjectinfrontofyou.Youarerational,soyouare90%certainthatthereisaredobjectinfrontofyou.Andyouarereflective,soyouhavethesecond-orderbeliefthattheUber-rulesaysthatagentswiththesumtotalofaredexperienceshouldbe90%cer-tainthatthereisaredobjectinfrontofthem.Sayyouaren%certainofthissecond-orderbelief.Nowaddaninformantwhosays(falsely)thattherationalresponsetoyourcurrentredexperienceistobeonly50%certainthatthereisaredobjectinfrontofyou.

Distinguishfirst-orderandsecond-orderresponses.Themost im-mediate response is lowering your second-order credence that theUber-rulesaysthatagentswithexactlytheevidenceofaredexperi-enceshouldbe90%certainthatthereisaredobjectinfrontofthem.Thisfallsfromn%.Thefirst-orderresponseisthatyourcredencethatthereisaredobjectinfrontofyoudropsbelow90%.Ifyoufullytrust-edtheinformant,itwouldfallto50%,butlet’ssayitsettlesat70%.

ThereisnoviolationoftheUber-rule.Yourepistemicpositionhaschanged—yourevidencenowincludestheredexperienceand the tes-timony.Anditiscompatiblewiththisstorythattherecommendationof theUber-rule for someonewith this evidence is “Have 70% cre-dencethatthereisaredobjectinfrontofyou.”

Noparadoxsofar.PerhapstheproblemisthatyoucandoubtthecontentoftheUber-rule.YouarenotcertainthattheUber-rulerecom-mendsa90%credencethatthereisaredobjecttoagentswithjustaredexperience.Similarly,youshouldnotbecertain that70% is therationalcredenceinyourcurrentsituation.ButarationalagentmaydoubttheUber-rulewithoutbelievingthatthebeliefsitprescribesareirrational.Thepossibilitythatyourcredenceshouldbemorethan70%needstoperfectly balancethepossibilitythatyourcredenceshouldbelessthan70%.46

46.WhataboutTitelbaum’smemorablephrase—mistakesaboutrationalityaremistakesofrationality?Icanendorsethisphraseifwefillitoutasfollows:

Titelbaum’s-Principle-I-Accept:Foranygivenagentandsituation,iftheyare mistakenaboutwhattheyshouldbelieveintheirsituation,thentheyare makingamistakeofrationality.

Page 17: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –17– vol.19,no.3(january2019)

usedtorefertoeitheroftheseproperties.(Compare:Thereisnothingrationalaboutanidealgas.Theassumptionoftheidealnessofagasisanalogoustotheassumptionoftheprobabilismofagents.49)

In one sense, this type of revisability is stronger thanQuine al-lowed,foronewayQuinethinkswemightgiveupasentenceisawaythatwewouldnaturallydescribeasachangeinthemeaningofthesentence.(Quinedeniedthatwecouldseparatemeaningchangefrombelief change, due to his rejection of the analytic/synthetic distinc-tion.)WhereasIamhappytotalkaboutpropositionsandinclinedtohold thatnoproposition is immune from revision.But there are re-latedQuineanviewsthatIdonotendorse.

First,Idonotrejecttheapriori.IthinkthePrincipalPrinciple(andotherprinciples)areaprioriyetdefeasible.Somephilosophersiden-tifytheaprioriwithimmunityfromrevision—soIrejectthe“apriori”onlyinthissense.50

Second, I donot reject analyticity. Let an analytic sentencebe asentencethatonecanbeinapositiontojustifiablybelieveinvirtueofunderstandingit.Thisallowstheexistenceofdefeatersthatblockthejustification.Onecanrejectananalyticsentenceasfalseifonemistak-enlydoubtsthatitisanalytic.51Imaginebeingtoldthat‘Allbachelorsaremen’ is falsebyamisleadingbuteminentsource.Orbeingtoldthatyouhaveingestedadrugthatgeneratesfalsebeliefsaboutwhichsentencesareanalytic.Itmightberationaltorejectthesentence,soevenanalyticsentencesarenotimmunefromrevision.52

49.Hájek2006.

50.Casullo(2003)arguesthatthereisnoexperientialindefeasibilityconditionintheconceptofapriorijustification;andSummerfield(1991)andThurow(2006)arguethatapriorijustificationisdefeasiblebyexperience.

51. Alternatively, perhaps analytic sentences canbe rejected in the sense thattheyarebelievedtofailtousefullyapplytotheworld.Forexample,onecanrejectthesentence‘Etherconductsheat’onthegroundsthatthereisnoether.Eklund(2017p.89)holdsthatanalyticsentencescanbefalse.

52.Williamson(2007)writesthat“thecentralideabehindepistemologicalcon-ceptionsofanalyticity is that, insuchcases, failure toassent isnotmerelygoodevidenceoffailuretounderstand;itisconstitutiveofsuchfailure”(p.73).Irejecttheanalyticinthissense.

generalizations.Sothisversionofparticularismisnotdevastatingfornormativetheorizing.

ThiscompletesmydefenceoftheUber-rule.Inthenextsection,Iwillconnecttheresultingpositiontobroaderissuesregardingdefea-sibilityandtheapriori.

9. Against Certainty

IarguedintheprevioussectionthateventheUber-rulecanberatio-nallydoubted.ThissupportsaviewassociatedwithQuine—thatnostatement is immune from revision.Epistemic rules are good candi-datesforstatementsthatareimmunefromrevision,sobyarguingthattheyarenotimmunefromrevision,thegeneralcasethatnostatementisimmunefromrevisionissupported.

Still,Ihaveonlydiscussednarrow-scoperulesconnectingdescrip-tivewithnormative statements.Mypositiondoesnot entail thatnostatementisimmunefromrevision.Forexample,onemightstillholdthatrationalagentsarecertainoftautologies.

Nevertheless,Ithinkrationaldoubtcanberaisedevenabouttau-tologies, and for similar reasons. For example, suppose a heavenlyvoicetellsyouthatp-and-not-p,ortellsyouthatyourcredencesshouldsumto0.8.Bafflingsituationsforsure,buttheyseemtoprovidesomereasontodoubttautologies.48

ThisisachallengetoBayesianism,whichmodelsagentsasproba-bilistic,andsorequiresthatagentsarecertainoftautologies.Ithinkthe Bayesian should respond that their models, like most models,makeidealizations—thatis,theymakeassumptionsthatareknowntobefalseinordertomakethemodeleasiertoworkwith.Probabilismissuchanassumption.SoprobabilisticagentsinBayesianmodelsareidealized;buttheyarenotnecessarilyideal,inthesenseofbeingper-fectlyrational.Wehaveunfortunateterminologywhere‘ideal’canbe

48. SeeWilliams(forthcoming)foradefenceof“rationalillogicality”;Schechter(2013)offersrelatedargumentsagainstbeingcertain.AtoddswiththisistheliteraturebasedonMcFetridge(1990)(e.g.Hale2002,Ahmed2000,Leech2015),whichseemstoassumethattheremustbesomeruleswhichareim-munetodoubt.

Page 18: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –18– vol.19,no.3(january2019)

–––(2011). Disagreement,Question-Begging,andEpistemicSelf-Crit-icism.Philosophers’ Imprint 11(6):1–22.

–––(2013).EpistemicModestyDefended.InJenniferLackeyandDa-vidChristensen(eds.),The Epistemology of Disagreement: New Essays.OxfordUniversityPress.

Dancy,J.(2004).Ethics without Principles.ClarendonPress.–––(2007).DefendingtheRight.Journal of Moral Philosophy4(1):85–98.–––(2013).MoralParticularism.The Stanford Encyclopedia of Philosophy

(Fall2013Edition),EdwardN.Zalta(ed.),URL=<http://plato.stan-ford.edu/archives/fall2013/entries/moral-particularism/>.

Eklund, M. (2017). Choosing Normative Concepts. Oxford UniversityPress.

Elga,A. (2007). ReflectionandDisagreement. Noûs 41(3):478–502.––– (2010). HowtoDisagreeAboutHowtoDisagree. InTedA.Warf-

ieldandRichardFeldman(eds.),Disagreement.OxfordUniversityPress.

Field,H.(2000).ApriorityasanEvaluativeNotion.InPaulBoghossianandChristopherPeacocke(eds.),New Essays on the A Priori.OxfordUniversityPress.

Hájek,A. (2006).MassesofFormalPhilosophy ‘Interview’.http://ci-teseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.124.5339&rep=rep1&type=pdf.

Hale,B.(2002).BasicLogicalKnowledge.Royal Institute of Philosophy Supplements51:279–304.

Hedden, B. (2012).Options and the SubjectiveOught. Philosophical Studies 158(2):343–60.

Holton,R.(2002).PrinciplesandParticularisms.Proceedings of the Aris-totelian Society76(1):191–209.

Hooker, B. and Little, M. (eds.) (2000).Moral Particularism. OxfordUniversityPress.

Horowitz,S.(2014).EpistemicAkrasia.Noûs48(4):718–44.Horty,J.(2007).ReasonsasDefaults.Philosophers’ Imprint 7(3):1–28.Huemer,M.(2000).DirectRealismandtheBrain-in-a-VatArgument.

Philosophy and Phenomenological Research61(2):397–413.

10. Conclusion

Theproblemsofthispaperweregeneratedbytheparadoxthatemerg-esinsituationswhereepistemicrulesunderminethemselvesoreachother.Onewaytoavoidparadoxistomaintainthatepistemicrulesareindefeasibleandignoreallopposingevidence.Ihavearguedinsteadthatweshouldthinkofsimpleepistemicrulesashedgedrules.Theyapplyonlyifagentsdon’thaveevidencethatisinadmissiblerelativetothoserules.IhavedefendedtheviewthattheonlyrulethatappliesinallsituationsisanUber-rulewhichstateswhatagentsshouldbelievegivenanypossibleevidence.ButeventhecontentoftheUber-rulecanberationallydoubted.53

References

Ahmed,A.(2000).HaleonSomeArgumentsfortheNecessityofNe-cessity. Mind 109(433):81–91.

Arpaly,N.(2002).Unprincipled Virtue: An Inquiry Into Moral Agency.Ox-fordUniversityPress.

Bales,R.E.(1971).Act-Utilitarianism:AccountofRight-MakingChar-acteristics or Decision-Making Procedure?American Philosophical Quarterly8(3):257–65.

Boghossian, P.A. (2008). Epistemic Rules. The Journal of Philosophy 105 (9):472–500.

Broome,J.(1999).NormativeRequirements.Ratio12(4):389–419.Cartwright,N. (1983).How the Laws of Physics Lie.OxfordUniversity

Press.Casullo,A.(2003).A Priori Justification.OxfordUniversityPress.Christensen,D.(2007).DoesMurphy’sLawApplyinEpistemology?

Self-DoubtandRationalIdeals. Oxford Studies in Epistemology 2:3–31.––– (2010). Higher-Order Evidence. Philosophy and Phenomenological

Research 81(1):185–215.

53. IamgratefultoAdamElga,MikeTitelbaum,PekkaVäyrynenandRobbieWil-liams,andtotworefereesfromthisjournalwhogavemetwosetsofdetailedcomments.ItwouldhavebeentedioustoacknowledgealltheplaceswhereIbenefittedfromtheirfeedback.

Page 19: Are There Indefeasible Epistemic Rules? · Section 3 explains the indefeasibility view, and section 4 argues against it. Section 5 argues against contributory rules, and Section 6

darrenbradley Are There Indefeasible Epistemic Rules?

philosophers’imprint –19– vol.19,no.3(january2019)

Strahovnik,V.,Potrč,M.,andLance,M.N.(eds.)(2008).Challenging Moral Particularism.Routledge.

Summerfield,D.M.(1991).ModestAPrioriKnowledge.Philosophy and Phenomenological Research51(1):39–66.

Thurow,J.(2006).ExperientiallyDefeasibleAPrioriJustification.The Philosophical Quarterly56(225):596–602.

Timmermann, J. (2013).KantianDilemmas?MoralConflict inKant’sEthicalTheory.Archiv für Geschichte der Philosophie95(1):36–64.

Titelbaum,M.G. (2015).Rationality’sFixedPoint (Or: InDefenseofRightReason). Oxford Studies in Epistemology5:253–94.

Väyrynen,P. (2006).MoralGeneralism:Enjoy inModeration.Ethics 116(4):707–41.

–––(2008).UsableMoralPrinciples,inStrahovniketal.2008.––– (2009).ATheory ofHedgedMoral Principles.Oxford Studies in

Metaethics4:91–132.Williams,R.(forthcoming).RationalIllogicality.Australasian Journal of

Philosophy96(1):127–41.Williamson, T. (2007). The Philosophy of Philosophy. Blackwell

Publishing.Worsnip, A. (2015). The Conflict of Evidence and Coherence.Philo-

sophical and Phenomenological Research 96 (1):3–44. doi:10.1111/phpr.12246.

Jackson,F.,Pettit,P.,andSmith,M.(2000).EthicalParticularismandPatterns,inHookerandLittle2000.

Lackey, J.  and Christensen,D.  (eds.) (2013). The Epistemology of Dis-agreement: New Essays. OxfordUniversityPress.

Lasonen-Aarnio,M.(2014).Higher-OrderEvidenceandtheLimitsofDefeat.Philosophy and Phenomenological Research 88(2):314–45.

Lewis,D.(1971).ImmodestInductiveMethods.Philosophy of Science 38(1):54–63.

––– (1980). ASubjectivist’sGuidetoObjectiveChance. InRichardC.Jeffrey(ed.), Studies in Inductive Logic and Probability.UniversityofCaliforniaPress.

McFetridge,I.G.(1990).LogicalNecessity:SomeIssues.InJohnHal-daneandRichardScruton(eds.),Logical Necessity & Other Essays, volume11ofAristotelianSocietySeries.AristotelianSociety.

McKeever, S. andRidge,M. (2006).Principled Ethics: Generalism as a Regulative Ideal.OxfordUniversityPress.

Pietroski,P.M.andRey,G.(1995).WhenOtherThingsAren’tEqual:SavingCeterisParibusLawsfromVacuity.The British Journal for the Philosophy of Science46(1):81–110.

Pollock,J.L.(1967).CriteriaandOurKnowledgeoftheMaterialWorld.The Philosophical Review76(1):28–60.

Pryor,J. (2000). TheSkepticandtheDogmatist. Noûs 34(4):517–49.Ross,W.D.(1930).The Right and the Good.OxfordUniversityPress.Schaffer, J. (2015).WhatNot toMultiply withoutNecessity. Austral-

asian Journal of Philosophy 93 (4):644–64.Schechter, J. (2013). Rational Self-Doubt and the Failure ofClosure.

Philosophical Studies163(2):429–52.Schroeder,M.(2004).TheScopeofInstrumentalReason.Philosophical

Perspectives18(1):337–64.––– (2005). Realism andReduction: TheQuest for Robustness.Phi-

losophers’ Imprint5(1):1–18.–––(2011).Holism,Weight,andUndercutting.Noûs45(2):328–44.––– (2013). Scope for Rational Autonomy. Philosophical Issues 23

(1):297–310.