An Explanatory Account of Practical Reasons
-
Upload
khangminh22 -
Category
Documents
-
view
0 -
download
0
Transcript of An Explanatory Account of Practical Reasons
TheLondonSchoolofEconomicsandPoliticalScience
AnExplanatoryAccountofPracticalReasons
DerenCemHalilOlgun
AthesissubmittedtotheDepartmentofPhilosophy,LogicandScientificMethodoftheLondonSchoolofEconomicsandPoliticalScienceforthedegreeofDoctorofPhilosophy,December2017.
2
Declaration
I certify that the thesis I havepresented for examination for thePhDdegreeof the London
School of Economics and Political Science is solely my own work other than where I have
clearlyindicatedthatitistheworkofothers(inwhichcasetheextentofanyworkcarriedout
jointlybymeandanyotherpersonisclearlyidentifiedinit).
Thecopyrightofthisthesisrestswiththeauthor.Quotationfromitispermitted,providedthat
full acknowledgement ismade.This thesismaynotbe reproducedwithoutmypriorwritten
consent.
Iwarrantthatthisauthorisationdoesnot, tothebestofmybelief, infringetherightsofany
thirdparty.
Ideclarethatmythesiscontains91,596words(excludingbibliographyandappendices).
3
Abstract
If I takemyumbrella,having seen that it’s rainingoutside,wemight say thatmy reason for
takingmyumbrellawasthatitwasraining.However,ifI’dbelievedthatitwasrainingwhenit
wasn’t,wemight say thatmy reason for takingmyumbrellawas that I believed that itwas
raining.Inthefirstcase,myreasonforactingseemstobeafeatureoftheworld,whilstinthe
seconditseemstobeafeatureofmypsychology.
Accordingtomosttheoriesofreasons,wearemistakenaboutwhatmyreasonforactingwas
inoneofthesecases.However,Iargue,thesetheoriesallentailseveralawkwardclaims.
I argue that there is a theory of reasons that can reconcile these two accounts ofwhatmy
reasonforactingwaswithoutentailingsuchawkwardclaims.Iarguethatwhatthefactthatit
israiningandthefactthatIbelievethatitisraininghaveincommonisthat,intheirrespective
cases, they each explainwhy itwas rational forme to takemy umbrella andwhy I took it.
Moregenerally,Iarguethatthereisatleastasenseinwhichallpracticalreasonsexplainwhy
itis,insomerespect,rationalfortheagenttodotheactionsforwhichtheyarereasons.
Themajorchallengeforthisaccount is theclaimthatonly featuresofanagent’spsychology
canexplainwhytheyactorwhyit isrationalforthemtoact. Iprovideaformalconstrualof
thischallengeandarguethatthefactthatitisrainingcanexplainwhyItakemyumbrellaand
whyitwasrationalformetodoso,byexplainingthefactthatIbelievedthatitwasraining.
4
ContentsLISTOFTABLES 7
LISTOFFIGURES 8
ACKNOWLEDGEMENTS 9
INTRODUCTION 10
1 WHYDOWENEEDANEWTHEORYOFREASONS? 112 PLURALISMABOUTREASONS 113 EXPLANATORYRATIONALISM 124 MYTHEORYOFREASONS 155 THATWHICHIPASSOVERINSILENCE 156 ANOVERVIEWOFTHISDISCUSSION 16
CHAPTERSUMMARY 20
(I)ONTHEORIESOFREASONS 23
1 HOWMANYKINDSOFREASONARETHERE? 242 CLAIMSABOUTREASONS 333 CATEGORISINGTHEORIESOFREASONS 384 NORMATIVEANDMOTIVATINGREASONS 395 CONCLUSION 46
(II)REASONSTOACTTHATMAKEACTIONSWORTHDOING 47
1 SALLYANDTHENON-EXISTENTBEAR 472 HOWREASONEXPRESSIONSRELATE 503 FAVOURISMABOUTREASONSTOACT 514 THEPROBLEMSFORFAVOURISM 555 RESPONSESTOTHEPROBLEMSFORFAVOURISM 576 CONCLUSION 59
(III)ACTINGFORPSYCHOLOGICALREASONS 60
1 SOMEPRIMAFACIEREASONABLECLAIMS 602 PSYCHOLOGISMABOUTTHEREASONSFORWHICHWEACT 643 THEPROBLEMSFORPSYCHOLOGISM 654 RESPONSESTOTHEPROBLEMSFORPSYCHOLOGISM 675 CONCLUSION 68
(IV)ACTINGFORWHATYOUBELIEVE 69
1 SOMEPRIMAFACIEREASONABLECLAIMSABOUTREASONS 692 DELIBERATIVISMABOUTTHEREASONSFORWHICHWEACT 723 WHATSALLYANDEDMUNDTOOKTOFAVOURACTING 734 THEPROBLEMSFORDELIBERATIVISM 755 RESPONSESTOTHEPROBLEMSFORDELIBERATIVISM 776 CONCLUSION 79
(V)ONTHEPLURALITYOFREASONS 80
1 THESENSEOFANEXPRESSION 812 EXPANDINGTHECATEGORISATIONSCHEMA 813 FAVOURIST/DELIBERATIVIST(F/D)PLURALISM 83
5
4 FAVOURIST/PSYCHOLOGIST(F/P)PLURALISM 855 WHYBEAPLURALIST? 876 PLURALISMISNOPANACEA 897 ACHALLENGEFORPLURALISM 928 CONCLUSION 93
(VI)ANEWFAMILYOFCLAIMSABOUTREASONS 94
1 PROTANTORATIONALACTION 942 EXPLANATORYRATIONALISM 973 APROBLEMFOREXPLANATORYRATIONALISM 984 ANOUTLINEOFWHATFOLLOWS 99
(VII)WENEEDTOTALKABOUTEXPLANATION 102
1 WHATDOIMEANBY‘EXPLAINS’? 1022 ONTOLOGICALASSUMPTIONS 1033 FULLEXPLANATIONANDPARTIALEXPLANATION 1044 OVERDETERMINATIONANDOVEREXPLANATION 1085 SUMMARY 110
(VIII)THEEXPLANATORYEXCLUSIONPROBLEM 111
1 ANOVERVIEW 1122 THEARGUMENTFROMFALSEBELIEF 1143 THEARGUMENTFROMIMPOTENTFACTS 1224 THEARGUMENTFORPREMISE1 1255 THEEXCLUSIONPRINCIPLE 1266 THEEXPLANATORYEXCLUSIONPROBLEMFOR(R1) 1287 CONCLUSION 128APPENDIX 129
(IX)OTHERUSESFORTHEEXPLANATORYEXCLUSIONPROBLEM 130
1 THEGENERALFORMOFTHEEXPLANATORYEXCLUSIONPROBLEM 1312 THEEXPLANATORYEXCLUSIONPROBLEMFOR(R2) 1323 THEEXPLANATORYEXCLUSIONPROBLEMFOR(R3) 1344 THEEXPLANATORYEXCLUSIONPROBLEMFOR(R4) 1365 THEARGUMENTFROMILLUSION 1396 CONCLUSION 140
(X)HOWNORMATIVEREASONSDON’TEXPLAIN 141
1 NORMATIVEREASONEXPLANATIONS 1422 THEORIESOFNORMATIVEREASONEXPLANATION 1433 ELLIPTICALTHEORIES 1444 DIRECTTHEORIES 1475 CONCLUSION 150APPENDIX 150
(XI)THEEXCLUSIONPRINCIPLEISFALSE 159
1 TWOCOUNTEREXAMPLESTOTHEEXCLUSIONPRINCIPLE 1592 WHYTHESEARECOUNTEREXAMPLESTOTHEEXCLUSIONPRINCIPLE 1623 WHAT’SWRONGWITHTHEEXCLUSIONPRINCIPLE 1644 WHEREDIDWEGOWRONG? 1665 WHICHEXPLANATIONAREDISTALEXPLANATIONSPARTOF? 1676 CONCLUSION 167
6
(XII)EXPLAININGWHYWEACT 168
1 WHENNORMATIVEREASONSEXPLAIN 1692 IMPLICATIONSFOREXPLANATORYRATIONALISM 1713 IMPLICATIONSFORANTI-PSYCHOLOGICALTHEORIESOFREASONS 1724 CONCLUSION 173APPENDIX 173
(XIII)EXPLAININGWHYITISRATIONALTOACT 177
1 ANOTHERINDIRECTTHEORY 1772 ISEXPLANATIONTRANSITIVE?ANAPPARENTDILEMMA 1783 THEAPPARENTDILEMMAISNOTADILEMMA 1804 THECHALLENGE 1835 THEUNSUCCESSFULNATURALSTRATEGY 1836 THEMYSTERIOUSSTRATEGY 185
(XIV)THEMYSTERYRELATION 186
1 THEMYSTERYRELATIONANDJUSTIFIEDBELIEF 1872 THEMYSTERYRELATIONANDKNOWLEDGE 1893 THEMYSTERYRELATIONANDOPPORTUNITIESTOKNOW 1904 THEMYSTERYRELATIONANDACTINGFORAREASON 1925 ASUMMARYOFTHEEXAMPLES 1936 MYSTERYRELATIONSAREEXPLANATORYRELATIONS 1947 MYSTERYRELATIONSARETRANSITIVE 1968 CONCLUSION 200
(XV)MYSTERYRELATIONSANDWHYITISRATIONALTOACT 201
1 EXPLAININGWHYITISRATIONALTOACT 2022 ARELATIONINCOMMON 2033 WHENNON-PSYCHOLOGICALFACTSEXPLAINWHYITISRATIONAL 2094 CONCLUSION 211APPENDIX 211
(XVI)ANEWTHEORYOFREASONS 217
1 EXPLANATORYRATIONALISM:REVISITED 2182 SOLVINGTHEPROBLEMS 2213 NEWPLURALISM 2224 THECHALLENGEFORPLURALISM 2235 CONCLUSION 225APPENDIX 226
BIBLIOGRAPHY 232
7
ListofTablesTableI-1:The‘standard’categorisationoftheoriesofreasons.................................................23
TableI-2:Reasonexpressionsandthekindsofreasontheypickout........................................32
TableI-3:Aprovisionalcategorisationschema..........................................................................33
TableI-4:Themainclaimsabouteachreasonexpression.........................................................36
TableI-5:Arevisedcategorisationschema................................................................................38
TableI-6:Prominenttheoriesofreasons,categorisedinmyproposedschema........................39
TableI-7:The‘assumedtheory’ofreasons................................................................................41
TableI-8:Theconventionalinterpretationoftheassumedtheory............................................43
TableII-1:Howproponentsoffavourismrespondtosomeproblemsforit..............................58
TableV-1:Acategorisationschemathataccommodatespluralisttheoriesofreasons.............82
TableV-2:The,univocal,'ReceivedView'representedinthenewschema...............................82
TableV-3:Anexamplepluralisttheory......................................................................................83
TableV-4:PureF/Dpluralism.....................................................................................................85
TableV-5:PureF/Ppluralism.....................................................................................................87
TableVI-1:ThewaysinwhichFevzi’sactionsare(oraren’t)rational........................................97
TableVI-2:Explanatoryrationalism............................................................................................97
TableXIV-1:Thecomponentfactsineachexample.................................................................197
TableXIV-2:Thetransitivityofthemysteryrelation................................................................197
TableXV-1:Asummaryofwhatexplainsandwhatdoesn'tineachcase................................211
TableXVI-1:Explanatoryrationalism........................................................................................218
TableXVI-2:Anapplicationofexplanatoryrationalism...........................................................221
TableXVI-3:NewPluralism.......................................................................................................223
TableXVI-4:Otherexamplesforexplanatoryrationalism........................................................226
8
ListofFiguresFigureVIII-1:TheargumentforTheExplanatoryExclusionProblem.......................................114
FigureX-1:Ellipticaltheoriesofnormativereasonexplanation...............................................145
FigureX-2:Directtheoriesofnormativereasonexplanation...................................................147
FigureXI-1:TheHyperAcceleratedDragon..............................................................................161
FigureXII-1:Theindirecttheoryofnormativereasonexplanation..........................................169
FigureXII-2:Theindirecttheoryofbeliefexplanation.............................................................170
FigureXII-3:TheexplanationofwhyIcongratulatedmyfriend..............................................170
FigureXII-4:TheexplanationofwhySallyran..........................................................................171
FigureXII-5:TheexplanationofwhyEdmundskatedattheedgeofthelake.........................171
FigureXII-6:Explainingwhyweact..........................................................................................172
FigureXIII-1:Explainingwhyitisrationaltoact.......................................................................178
FigureXIII-2:Achainofexplanatoryrelationsthatarealltransitive.......................................183
FigureXIII-3:Achainofexplanatoryrelationsthatarenotalltransitive.................................183
FigureXV-1:Theexplanatoryrelationsintheawardcase.......................................................209
FigureXV-2:Theexplanatoryrelationsinthecarbonmonoxidecase......................................209
FigureXV-3:TheexplanatoryrelationsintheEvacase............................................................209
FigureXVI-1:Theexplanatoryrelationsinvolvedwhenitwasraining.....................................219
9
AcknowledgementsThis thesis exists, in large part, because of the advice and support of my friends and
colleagues. I am grateful to all of them, but I would like to thank the following people, in
particular.
First, Iwould like to thankmyprimarysupervisor,ChristianList.Were itnot forhis rigorous
and precise feedback, thisworkwould have been far poorer, andwere it not for his open-
mindednessitwouldhavecertainlylackedwhateveritnowhasbywayoforiginality.Iwould
also like to thankmy secondary supervisor, RichardBradley, forhis astute criticisms, for his
encouragement, and for being so generous with his time. And I would like to thank my
examiners, Maria Alvarez and Jonathan Dancy, for their thoughtful engagement with this
thesis,andfortheirwordsofsupport.Havingspentsomuchtimethinkingabouttheirwork,it
wasanhonourtoheartheirthoughtsonmine.
I amenormously grateful toBryanRoberts for helpingme through several iterationsof The
ExplanatoryExclusionProblem.Withouthis guidancemyelaborationof thatproblemwould
havebeentwiceaslongandhalfasgood.IamalsogratefultoPeteFaulconbridgeandJames
Nguyen for their thoughtful responses to an early draft of chapters (XV) and (XIV); their
insights transformed my arguments. And I am grateful to Todd Karhu for his considered
engagementwithallofthemyriadproblemsIputtohim;hisinfluenceonthisthesisissubtle
butpervasive.
I would like to thank the donor who funded my scholarship, and who wished to remain
anonymous; their generositymade this work possible. I would also like to thank all ofmy
colleagues at the Office of Rail and Road, and Chris Hemsley in particular, for being so
supportiveandsoaccommodating.
IamluckytohaveenjoyedthefriendshipofmyfellowPhDstudents.Itwouldberemissofme
not to mention, in particular, and in addition to those already mentioned, Susanne Burri,
Goreti Faria, Johannes Himmelreich, Silvia Milano, Mantas Radzvilas and Nicolas Wüthrich.
Theircompanionshipthroughthepeaksandtroughsofthisworkwasinvaluable.
Finally,Ithankmypartner,AlexBobocica,whohasbeensupportive,patient,andconsiderate
throughout.Idon’tknowhowIwouldhavedonethiswithouther.
10
IntroductionIf I takemyumbrella,having seen that it’s rainingoutside,wemight say thatmy reason for
taking my umbrella was that it was raining. However, if it hadn’t been raining, but I’d
mistakenlybelievedthat ithad,wemight insteadsaythatmyreasonfortakingmyumbrella
wasthatIbelievedthatitwasraining.
Inthefirstcase,myreasonforactingseemstobeafeatureoftheworld,whilstinthesecond
it seems to be a feature of my psychology. Can these two different accounts of what my
reasonforactingwasbereconciledwithinasingletheoryofwhat it istobeareason?Most
theorists thinknot;most theorists think that in oneor theother of these caseswe are just
wrongaboutwhatmyreasonforactingwas.
First,therearethosetheoristswhotakewhathappenswhenIammistakentobeindicativeof
whathappenswhenI’mnot.Theyinsistthat,irrespectiveofwhetherornotitwasraining,my
reason for taking my umbrella was that I believed that it was raining (and not that it was
raining).Theoristsofthissort,so-called‘psychologists’,suggestthatourreasonsforactingcan
onlybefeaturesofourpsychology.
Second, and in contrast, there are those theorists who take what happens when I’m not
mistaken to be indicative of what happens when I am. They insist that, in both cases (i.e.
regardlessofwhetherornotitwasraining),myreasonfortakingmyumbrellawasthatitwas
raining (and not that I believed that it was raining). These theorists, whom I call
‘deliberativists’,insistthattheconsiderationinlightofwhichanagentactsistheirreasonfor
acting.
Third,therearethosetheorists,whomIcall ‘favourists’,whodonotseektolearnsomething
fromonecaseabouttheother; instead,they insistthatourreasonsforactingmustcount in
favourofouractions.Likedeliberativists,favouristsarguethat,whenitwasraining,myreason
fortakingmyumbrellawasthatitwasraining.However,unlikedeliberativists,favouristsinsist
thatifItakemyumbrellabecauseofamistakenbeliefthatitisrainingthenIdon’ttakeitfora
reason.
Incontrasttoallofthesetheorists,Ithinkthatthetwodifferentaccountsofwhatmyreason
for takingmyumbrellawascanbe reconciledwithina single theoryof reasons. I argue that
whatthefactthat it israiningandthefactthat Ibelievethat it israininghave incommonis
that, in their respective cases, they each explain why it was rational for me to take my
umbrella and why I took it. I suggest that it is in virtue of standing in those explanatory
11
relationstomyactionthatthosefactswere,intheirrespectivecases,myreasonfortakingmy
umbrella.Moregenerally,Iarguethatthereisaclearsenseofwhatitistobeareasonthereis
for one to act, or a reason one has to act, or a reason for acting, according towhich such
reasonsexplainwhyitis,inacertainsense,rationalfortheagenttodotheactionsforwhich
theyarereasons.Icallthisaccount‘explanatoryrationalism’.
1 Whydoweneedanewtheoryofreasons?
If there are already three, distinct, andpopular theories of the relation between an agent’s
reasonforactingandtheiraction,whygolookingforafourth?Whydoweneedanewtheory
ofreasons?
Weneedanewtheoryofreasonsbecausetheexistingtheoriesare,asJonathanDancyputsit,
‘awkwardinthesortofwaythatisusuallyduetobadtheory.AsAristotlesaid,theyleaveone
sayingthingsthatnobodywouldsayunlessdefendingatheory.’(2008a,267)Forinstance,itis
awkwardtosay,asthepsychologistmust,thatmyreasonfortakingmyumbrellacouldnever
bethatitwasraining.Anditisawkwardtosay,asthedeliberativistmust,thatIcouldtakemy
umbrellaforthereasonthatitwasraining,eventhoughitwasn’t.Anditislikewiseawkward
tosay,asthefavouristmust,thatalthoughItakemyumbrelladeliberately,purposefullyand
intentionally, I don’t take it for a reason. And indeed, as I will show, these theories are
awkwardinyetotherwaysstill.
IstresstheawkwardnessofthesetheoriesnotbecauseItakethattobethelastwordontheir
plausibility;clearly,onecouldjustacceptthisawkwardnessasapricethatisworthpayingfor
anotherwiseconvenientaccountofwhatit istobeareason.Myaimisrathertomakeclear
that foreachof thesetheories there isaprice thatmustbepaid,andthat thepriceofeach
theoryissufficienttomotivatethesearchforlesscostlyalternatives.AsMariaAlvarezputsit:
prima facieparadoxical claimsarenot, ‘adecisiveargumentagainst theviews thatgenerate
thembut[they]seemtofavourconsideringwhetherthereisaplausiblealternativeviewthat
doesnotcommitonetosuchclaims.’(2016b,11)Soisthere?
2 Pluralismaboutreasons
Well,hereisanalternativethatsometheoristsfavour:perhapstherearejustdifferentsenses
ofwhatitistobeareason.Forinstance,perhapsthereisasenseofwhatitistobeareasonin
whichreasonsarewhatpsychologismsaystheyare,andperhapsthereisadifferentsenseof
whatitistobeareasoninwhichreasonsarewhatfavourismsaystheyare.Accordingtosuch
an account, if I takemy umbrella because I mistakenly believe that it is raining, there is a
(psychologistic) sense inwhich I take it fora reasonanda (favourist) sense inwhich Idon’t;
12
likewise (topickadifferentexpression), there isa sense inwhich Ihad a reason to takemy
umbrella and a sense in which I didn’t. According to this pluralist theory of reasons, the
awkwardness that univocal theories of reasons face is merely the result of conflating two
differentsensesofasingleexpression;pluralisttheoriesthuspurporttobeexactlythesortof
plausiblealternativethatwearelookingfor.Butarethey?
Well,IagreethatwhenImistakenlybelievethatitisrainingthereisasenseinwhichItakemy
umbrellaforareasonandasenseinwhichIdon’t;so,tothatextent,Iampersuadedthatour
eventualtheoryofreasonsoughttobepluralist.However,Iamnotpersuadedthatweshould
relyonpluralismasameansofavoidingtheproblemsthatunivocaltheoriesofreasonsface.In
particular, it is not at all clear tome thatwhenever a univocal theory ends up entailing an
awkwardclaimitisbecauseithasconflatedtwodifferentsensesofasingleexpression.And,in
the light of that scepticism, pluralism seems less like a way of avoiding awkwardness, and
morelikeacapitulationtoit.
3 Explanatoryrationalism
Wheredoesallthisleaveme?Itleavesmesearchingforaunivocalaccountofwhatitistobea
reasonthatdoesn’tforcemetosayawkwardthings(or,moremodestly,thatdoesn’tforceme
tosay thesameawkwardthingsas theseother theories).To thatend, Ipresentexplanatory
rationalism:
Reasonexpression Explanatoryrationalism
Foranyp,pisareasonforAtoφ… …ifandonlyifpexplainswhyitisprotantorationalforAtoφ.
Foranyp,pisareasonforA’sφing… …ifandonlyifpexplainswhyitisprotantorationalforAtoφandpmakesA’sφing,insomerespect,worthdoing.
Foranyp,pisareasonAhastoφ… …ifandonlyifpexplainswhyitisprotantorationalforAtoφ.
Foranyp,pisA’sreasonforφing… …ifandonlyifpexplainswhyitisprotantorationalforAtoφandexplains(intherightway)whyAφ’d.
Table1:Explanatoryrationalism1
I will eventually argue that explanatory rationalism avoids all the problems that face other
theories because it is able to reconcile the idea that agents always act for psychological
reasonswith the idea that theymay sometimes also act for non-psychological reasons. But
1Twopointsareworthnoting:First,IsaythatanactionisprotantorationalforAifandonlyifAtakesitto be, in some respect, worth doing. See §(VI)1 for further discussion. Second, the categorisationschemausedinTable1isunusualinsofarasitallowsforthepossibilityofdistinguishingbetweeneachof the expressions listed (it is conventional to take at least some of these expressions to be co-extensive).Idefendthisapproach,atlength,in§(I).
13
beforeIreachthatconclusion,IaddresswhatItaketobethemajorproblemforexplanatory
rationalism.
3.1 Theproblemforexplanatoryrationalism
Explanatoryrationalismclaims,amongstotherthings,thatanagentactsforthereasonthatp
onlyifpexplainsbothwhytheyactedandwhyitwas(protanto)rationalforthemtosoact.
Thismeansthat ifexplanatoryrationalismistobeconsistentwiththeclaimthatwhenIsaw
that itwasrainingmyreasonfortakingmyumbrellawasthat itwasraining(asI intenditto
be), then itmust be possible for the fact that itwas raining to explainbothwhy I tookmy
umbrellaandwhyitwasrationalformetotakemyumbrella.
Thereis,however,awell-rehearsedargumentagainstthispossibility.Thisargumentsaysthat
facts about things that are external to an agent’smind (e.g. the fact that it is raining) can’t
explainwhytheydidsomethingorwhyitwasrationalforthemtodoit,sincewhatanagent
does,orwhatitisrationalforthemtodo,onlydependsontheirmind.
For instance,noticethatevenif it isn’training,so longas Ibelievethat it israiningIwillstill
takemy umbrella; and Iwill take itbecause I believe that it is raining. Further, notice that,
given that Ineed tobelieve that it is raining inorder to takemyumbrella,even if I takemy
umbrella when it is raining Imust still take it because I believe that it is raining. Thus, the
argumentgoes, if thefactthat Ibelievethat it israiningcanexplainwhyItookmyumbrella
whetherornotitisraining,thenthefactthatitisrainingdoesnorealworkinexplainingwhyI
tookmyumbrella,and,therefore,itdoesnotexplainwhyItookit.2
Thisargumentisaproblemforanytheorythatsaysthatanagent’sreasonforactingcouldbe
somethingotherthanafeatureoftheirpsychology(asexplanatoryrationalism,favourismand
deliberativismalldo), and is thusprobably themotivatingargument forpsychologismabout
agents’reasonsforacting.Thus,inordertosaveexplanatoryrationalism,Ineedtosayhowit
isthat(inspiteoftheargumentabove)thefactthat it israiningcanexplainbothwhyItook
myumbrella,andwhyitwasrationalformetotakeit.
3.2 Myresponsetotheproblem
My response to thisproblem is to say that the fact that it is rainingexplainswhy I tookmy
umbrellabyexplainingwhyIbelievedthatitwasraining,whichinturnexplainswhyItookmy
umbrella.Likewise,Isuggest,thefactthatitwasrainingexplainswhyitwasrationalformeto
takemyumbrellabyexplainingwhyIbelievedthatitwasraining,whichinturnexplainswhyit2This argument can likewise be applied to the explanation ofwhy itwas rational forme to takemyumbrella.
14
was rational forme to takemyumbrella. I argue that themistake in theargumentabove is
thatitrejectsthepossibilityofsuchdistalexplanations.
Thisresponsereliesonthetransitivityoftheexplanatoryrelationsinvolved:iftheexplanatory
relationsbetweenthefactthatitwasraining,thefactthatIbelievedthatitwasraining,and
thefactthatItookmyumbrellaweren’ttransitive,thenthefactthatitwasrainingwouldnot
explainwhyItookit.
However,explanatory relationsarenotalways transitive.Forexample, if Imistake thespray
from a sprinkler for rain, then the fact that the sprinkler is spraying in front ofmywindow
explainswhy I believe that it is raining,which explainswhy it is rational forme to takemy
umbrella.But,thefactthatasprinklerissprayinginfrontofmywindowdoesnotexplainwhy
itisrationalformetotakemyumbrella,inspiteoftheexplanatorychainconnectingthetwo
facts.So,theexplanatoryrelationsinvolvedaren’ttransitivewithoneanother.
Whyisitthatthetransitivityofexplanationfailsinthiscase,butapparentlysucceedswhenI
actually see rain? That is, given that the fact that the sprinklerwas spraying in front ofmy
windowandthefactthatitwasrainingboth(intheirrespectivecases)explainwhyIbelieved
that it was raining, why is it that (as I have suggested) only the latter explains why it was
rationalformetotakemyumbrella?Itisbecause,Iwillargue,theexplanatoryrelationshipsin
thecaseofthelatter,unliketheformer,areallofacertain,transitivesort.Inparticular,Iargue
thatthereisamysterious,non-causalexplanatoryrelationthatobtains,interalia,betweenthe
factthatpandthefactthatanagentbelievesthatpwhenanagentknowsthatp.
Ithusarguethat,inthecasewhenIsawrain,thefactthatitwasrainingexplainswhyitwas
rational for me to take my umbrella because it ismysteriously related to the fact that I
believed that it was raining (which in turn explains why it was rational for me to take my
umbrella).AndIarguethat,conversely,inthecaseinwhichIwasmistaken,thefactthatthe
sprinklerwasspraying in frontofmywindowdoesnotexplainwhy itwasrational formeto
takemyumbrellabecauseit isnotmysteriouslyrelatedtothefactthatIbelievedthatitwas
raining(it ismerelycausallyrelatedto it). Moreover, Iargue, inbothofthesecasesthefact
that it appeared to me as though it was raining is mysteriously related to the fact that I
believedthat itwasraining,sothat inbothofthesecasesthefactthat itappearedtomeas
though itwas rainingexplainswhy itwas rational forme to takemyumbrella (and, indeed,
explainswhyItookit).
15
4 Mytheoryofreasons
Thus, according to explanatory rationalism, when I saw that it was raining, my reasons for
takingmy umbrella include the fact that it was raining, the fact that it appeared tome as
thoughitwasraining,andthefactthatIbelievedthatitwasraining.Incontrast,whenImerely
sawthesprayofthesprinkler,myreasonsfortakingmyumbrelladonotincludethe‘fact’that
it was raining (not least because it wasn’t raining) nor do they include the fact that the
sprinklerwas spraying in front ofmywindow, but they nonetheless include the fact that it
appearedtomeasthoughitwasraining,andthefactthatIbelievedthatitwasraining.
Unlike anyother univocal theoryof reasons, explanatory rationalism is thus consistent both
with theclaimthatagentsalwaysact forpsychological reasonsandwith theclaimthat they
sometimesactfornon-psychologicalreasons.And,indeed,itispreciselythisthatmeansthat
explanatory rationalism avoids the particular awkward claims that other theories face, and
which,moregenerally,makesit immunetomanyofthechallengesonewouldnormally level
against such theories. Thus, I argue, explanatory rationalism is the best univocal theory of
reasons.
However, since, as I noted, I am persuaded that theremay be two distinct senses to each
reasonexpression,myowntheoryofreasonsiswhatIcall‘newpluralism’.Newpluralismsays
thatonesenseofeveryreasonexpression isexplanatoryrationalist,whiletheothersense is
favourist; and this allowsme to say that, for instance, when I takemy umbrella because I
mistakenlybelievethatitisraining,thereisasenseinwhichmyreasonfortakingmyumbrella
isthatIbelievethatitisraining,andthereisasenseinwhichItakemyumbrellafornoreason.
The virtue of new pluralism over existing pluralist theories is that it does not rely on the
plurality of senses to avoid the awkwardness that univocal theories face – explanatory
rationalismalreadyavoidsthatawkwardnessonitsown.
5 ThatwhichIpassoverinsilence
Thereare a few topics aboutwhich this discussionmakesno claim. The first of these is the
ontology of reasons. For themost part Iwill talk as though reasons are propositions, if not
facts,however, this ismostly for convenience.Whilst therehasbeensomedebatebetween
those who take reasons to be facts (or states of affairs) and those who take them to be
psychological states, I make no particular claim about that. My theory is not about what
reasonsare; it isabouttherelation inwhichreasonsstandtotheactionsforwhichtheyare
reasons.SinceItakereasonstostandinexplanatoryrelationstotheactionsforwhichtheyare
16
reasons,ifyouacceptmyaccountthenitwillbe(atleastinpart)yourviewontheontologyof
explanantiathatdetermineswhattheontologyofreasonsis.AndaboutthatImakenoclaim.
This leadsmetothesecondtopicaboutwhich Imakenoclaim:whatthecorrectanalysisof
explanationis.WhilstIdiscussexplanationinsomedepth,Imakenoclaimsaboutthespecific
relationsthatunderpinexplanatoryrelations,andwhattheconditionsforthemare.However,
itmaybethatsomeofthepurportedinstancesofexplanationthatIciteareincompatiblewith
some accounts of explanation. To my knowledge, there are no instances of this kind that
wouldbeofseriousconcernformyargument,anditismybeliefthatanalternativeconstrual
of my argument could be made whatever one’s theory of explanation. However, I may be
wrong;theremaybeaccountsofexplanationthatareinconsistentwithwhatIwantfrommy
theoryofreasons,inwhichcase,sobeit.
Thirdly, I will also leave the question of whether or not desires are reasons well alone. Of
course,weregularlysaythingslike:‘myreasonforgoingtothegymisthatIwanttogetfit’;
and, ‘my reason for going to the shops is to buy eggs’. Whilst I think that explanatory
rationalismhassomething tosayabouthowwe interpret thesesentences, Iwon’tdiscuss it
here,andIwon’tmentionsuchsentencesfurther.
Finally, I will not discuss reasons for belief. Many hold that reasons for belief should be
analysableinthesamesortofwayaspracticalreasons,andthisisaviewthatIshare.Itwould
be tempting, then, to extend explanatory rationalism into an analysis of what it is to be a
reasontobelievesomething,and, indeed, Ithinkthatsuchanaccountcouldbe illuminating.
However,Idon’tdiscussthathere.
6 Anoverviewofthisdiscussion
The structureof thisdiscussion is as follows. In§(I), I setoutmyapproach to talkingabout
theoriesofreasons.InparticularIcategorisetheoriesofreasonsaccordingtotheclaimsthey
make about each of the following expressions: ‘the reasons there are for one to act’; ‘the
reasonsfororagainstacting’;‘thereasonsonehastoact’;and‘one’sreasonforacting’.Thisis
at odds with the prevailing tendency to sort reasons into ‘normative’ and ‘motivating’;
however, for reasons that I will make clear, I prefer to eschew that terminology in my
categorisationschema.
17
Thediscussionof§(I)highlightsthreeclaimsaboutreasonsthatareprobablythemostwidely
held(thoughnotnecessarilybythesametheorists).Theseare:
Favourism about reasons to act: For anyp, p is a reason forA toφ if and only ifpmakesA’sφing,insomerespect,worthdoing.
Psychologismaboutthereasonsforwhichweact:Foranyp,pisareasonforwhichAφ’difandonlyifpisafeatureofA’spsychologythatrationalisesφingandexplains(intherightway)whyAφ’d.
Deliberativismaboutthereasonsforwhichweact:Foranyp,pisareasonforwhichAφ’difandonlyifpisaconsiderationinlightofwhichAφs.
In§§(II), (III)and(IV), respectively, Ishowthateachof theseaccounts is inconsistentwitha
numberofprimafaciereasonableclaims.Theproblemssetout inthesechaptersaremostly
alternative construals of arguments that have already been made against each of these
accounts,andtheyformthebasisofmyclaimthatallexistingtheoriesofreasonsare,insome
sense,awkward.
In§(V),Iconsiderwhetherweshouldjustadoptpluralismasasolutiontotheawkwardnessof
univocal theories. Iconcludethatweshouldnot:pluralism,at leastoftheconventionalsort,
fallsshortofwhatweneedfromanewtheoryofreasons.
In §(VI), I set the agenda for the remainder of the discussion. I introduce explanatory
rationalism, and I giveanoverviewof themainproblem for it (highlighted in thediscussion
above),whichIcall‘TheExplanatoryExclusionProblem’.Iprovideanoutlineofmyapproach
todiscussing,and,ultimately,solvingthisproblem,whichisasfollows.
In §(VII), Imake some assumptions about the structural principles and logical properties of
explanation.Whilemydiscussionreliesoncharacterisingexplanationinthisway,Idonotthink
thateithermysolutiontoTheExplanatoryExclusionProblem,ormytheoryofreasonsmore
generally,dependsonexplanatoryrelationsbeingsocharacterisable.
In§(VIII),IusetheassumptionsoftheprecedingchaptertoprovideaformalconstrualofThe
ExplanatoryExclusionProblem. Inparticular, I showhow its conclusion (that factsabout the
worldexternaltoanagent’smindcannotexplaintheagent’sactions)canbearrivedat from
twoseeminglytrivialclaimsaboutwhatexplainsanagent’sactionwhentheyactinerrororin
ignorance,togetherwithfiveseeminglyplausibleprinciplesofexplanation.
In§(IX), IshowhowTheExplanatoryExclusionProblemcanalsobeusedtoarguethatfacts
abouttheworldexternaltoanagent’smindcannotexplainwhyitisrationalforthemtoact.
18
In§(X),Iconsidertwopopularaccountsofhow,forinstance,thefactthatitisrainingexplains
whyItookmyumbrella.Thefirstoftheseaccountssaysthatitdoessoelliptically;thesecond
saysthatitdoessodirectly.Iarguethatbothoftheseaccountsareflawed.
In§§(XI)&(XII),IdevelopmyaccountofhowthefactthatitisrainingexplainswhyItookmy
umbrella;Isaythatitexplainsitindirectly.Myargumentproceedsintwostages.First,in§(XI),
IarguethatweshouldrejecttheconclusionofTheExplanatoryExclusionProblembecauseitis
based on a false principle of explanation, the exclusion principle. The exclusion principle
implies that only themostproximal explanations of some explananda explain it; but this is
mistaken – I argue thatmost of the explanationswe are interested in are, to some extent,
distalexplanations.Then, in§(XII), Ishowhowthat insighthelps informtheaccountofhow
thefactthatitisrainingcanexplainwhyItookmyumbrella.Specifically,Isuggestthatthefact
that it is raining is adistal explanation ofmy action; it explainswhy I tookmy umbrella by
explaining amore proximal explanation of why I tookmy umbrella (namely, the fact that I
believedthatitwasraining).
In§(XIII),Isuggestthatthesameaccountexplainshowitisthat,forinstance,thefactthatitis
rainingcanalsoexplainwhyitisrationalformetotakemyumbrella.Thatis,thefactthatitis
rainingexplainswhyIbelievethatitisraining,whichinturnexplainswhyitisrationalforme
totakemyumbrella.
However,Inote,thisdoesnotmeanthatifanagent’sbeliefexplainswhyitisrationalforthem
todosomeaction,thenanythingthatexplainswhytheyhavethatbeliefwillalsoexplainwhy
it is rational for them to do that action. That is, I note, not all explanatory relations are
transitive.Ithensetthestagefortheremainderofthediscussion,whichseekstoprovidean
accountofwhentheexplanatoryrelationsinvolvedare,andwhentheyaren’t,transitive.
Myaccountproceeds in twostages.First, in§(XIV), I introducethemystery relation. Iargue
thatthemysteryrelationisanon-causal,transitive,explanatoryrelationthatrelates:thebelief
thatptosomejustificationforitwhenthatbeliefisjustified;thebeliefthatptothefactthatp
whenthebeliefthatp isknowledgeable;ajustificationforthebeliefthatptothefactthatp
when that justification affords theopportunity for knowledge; and an action to somebelief
thatexplainswhythatactionisrationalwhenthatactionisdoneintentionally.Then,in§(XV),
Iarguethatthemysteryrelationistransitivewiththenon-causalexplanatoryrelationinvolved
inexplainingwhysomeactionisrational,whereasmerelycausalexplanatoryrelationsarenot.
Thisiswhyfactsthatmerelycausallyexplainourbeliefs(suchasthefactthatthesprinkleris
19
sprayinginfrontofmywindow,orfactsthatdeviantlycauseourbeliefs)donotexplainwhyit
isrationalforustoact.
Finally, in §(XVI), I revisit explanatory rationalism. I show how it responds to the problems
faced by other theories. I suggest that it is to some extent immune from the conventional
proceduresforgeneratingproblemsfortheoriesofreasonsbecauseitisconsistentbothwith
the claim that agents always act for psychological reasons and with the claim that they
sometimes act for non-psychological reasons. In light of these arguments, I suggest that
explanatoryrationalismisthebestunivocalaccountofwhatitistobeareason.
Iconcludebyreturningtothe intuitionthat,whensomeoneactsona falsebelief, there isa
sense in which they act for a reason and a sense in which they don’t. Since I share this
intuition,Iadvocatenewpluralism,whichsaysthatexplanatoryrationalismtellsusonesense
ofwhatitistobeareason,whilstfavourismtellsustheother.Ishowthatnewpluralismdoes
notfacethesameweaknessesasotherformsofpluralismbecausewheneveranagentactsfor
areasoninthefavouristsense,theyalsoactforareasonintheexplanatoryrationalistsense.
20
ChapterSummary(I)ONTHEORIESOFREASONS
InwhichIsayhowweshouldcategorisetheoriesofreasons.Iarguethat,ifourcategorisationschema is to capture at least themain theories of reasons, then it should by allow for thepossibility of asmany kinds of reason as there are reason expressions. I say that instead ofdistinguishing between psychological and anti-psychological theories we should categorisetheoriesofreasonsaccordingtowhattheysay,foreachreasonexpression,abouttherelationbetweenthereasonspickedoutbythatexpressionandtheactionsforwhichtheyarereasons.And I say thatweshouldeschewthe terminologyof ‘normative’and ‘motivating’ reasons inourcategorisationschemaandinourtheorising,because,astheyarestandardlydefined,theymakesubstantiveclaimsaboutwhat it is tobea reason that it isproperly thebusinessofatheoryofreasonstodetermine,andthat,moreover,thoseclaimssystematicallydisadvantage‘psychological’ theories of reasons. To frame the discussion of practical reasons in terms‘normative’and‘motivating’reasonsis,Isuggest,aTrojanhorsethatearnesttheorisingoughttoreject.
(II)REASONSTOACTTHATMAKEACTIONSWORTHDOING
Inwhich I showwhat it costs to think that if there is a reason todo someaction then thataction is, in somerespect,worthdoing. I showhow ‘favourismabout reasons toact’ (whichentails that reasons to act make actions worth doing) clashes with some prima faciereasonableclaimsaboutacaseinwhichsomeoneactsonafalsebelief.Isetoutwhichclaimsthe proponent of this view must choose between rejecting and I categorise the commonchoicesfromtheliterature.
(III)ACTINGFORPSYCHOLOGICALREASONS
In which I showwhat it costs to think that the reason for which an agent acts is always afeature of their psychology. I showhow ‘psychologism about the reasons forwhichwe act’clasheswithsomeprimafaciereasonableclaims.Inparticular,Ishowthatisinconsistentwiththeideathatweareoftenabletoactforreasonsthatmakeouractionsmorallyworthyand,moregenerally,worthdoing.
(IV)ACTINGFORWHATYOUBELIEVE
InwhichIshowwhatitcoststothinkthatareasonforwhichanagentactsisthecontentofthe belief they acted on. I show how ‘deliberativism about the reasons for which we act’clashes with some prima facie reasonable claims about the factivity of reasons, theexplanatorypowerof thereasons forwhichweact, thefactivityofexplanationandwhatanagent’s reasons for acting are inGettier cases. I set outwhich claims the proponent of thisviewmustchoosebetweenrejecting.
(V)ONTHEPLURALITYOFREASONS
InwhichIexplainwhatapluralisttheoryofreasonsisandwhy‘goingplural’isnotapanacea.Isuggestthatagivenreasonexpressioncouldhavemorethanonesense,andIshowhowwecanaccommodatetheoriesofreasonsthatacceptthatidea,i.e.pluralisttheoriesofreasons,inourcategorisationschema.Idiscusssomeexamplesofpluralisttheoriesfromtheliterature.Ishow how pluralist theories can solve some of the problems discussed in the previouschapters.Iexplainwhypluralismisnot,however,enough,andIsuggestthatourinvestigationshouldgobeyondfavourism,psychologismanddeliberativism.
21
(VI)ANEWFAMILYOFCLAIMSABOUTREASONS
InwhichIsetoutanewfamilyofclaimsaboutreasons,andintroducethemajorchallengetoit. Idefine ‘pro tanto rational’actionsasactions thatanagent takes tobe, in somerespect,worthdoing. I set out a new family of claims about reasons, explanatory rationalism,whichsaysthatallpracticalreasonsexplainwhytheactionsforwhichtheyarereasonsareprotantorational.Iintroducethemajorchallengeforexplanatoryrationalism,TheExplanatoryExclusionProblem,whichargues thatonly featuresof an agent’s psychology couldexplaineitherwhythey do something or why it was rational for them to do it. I set out the program for theforthcomingchapters.
(VII)WENEEDTOTALKABOUTEXPLANATION
Inwhich Imake someassumptions aboutexplanation. I saywhat Imeanby ‘explains’ and IstatethatIwilltalkasthoughexplanandaarefactsandexplanantiaarepropositions(whetherornot theyare). I distinguish two sortsof explanatory relation, ‘fully explains’ and ‘partiallyexplains’,wherea fullexplanation issufficient for thetruthof the fact that itexplainsandapartial explanation is an element (or subset) of a full explanation, and I make someassumptions about the logical properties of these relations. Lastly, I say that some fact is‘overexplained’justincasetherearetwogenuinelydifferentfullexplanationsofthatfact.
(VIII)THEEXPLANATORYEXCLUSIONPROBLEM
InwhichIsetoutTheExplanatoryExclusionProblem,whichis, insomeformoranother,themotivatingargumentforpsychologistictheoriesofreasons.IprovideaformalconstrualoftheProblem, showing how it results from two seemingly trivial claims about what explains anagent’s action when they act from error and from ignorance together with five seeminglyplausibleprinciplesofexplanation.IshowhowtheProblemimpliesthatIdidnotcongratulatemyfriendbecauseshehadwonanaward,butonlybecauseIthoughtshehad.
(IX)OTHERUSESFORTHEEXPLANATORYEXCLUSIONPROBLEM
InwhichIshowhowTheExplanatoryExclusionProblemcanbeusedtoarriveatsomeotherconclusionsthatareinconvenientforexplanatoryrationalism.IsetoutthegeneralformoftheProblem,followedbythegeneralformoftheargumentforthefirstpremiseoftheProblem.IshowtheProblemcanbeusedtoarguethat the fact that I readthatmy friendhadwonanawarddoesnotexplainwhy Icongratulatedher,andthatneitherthatfact,northefactthatshehadwonanaward,canexplainwhyitwasprotantorationalformetocongratulateher.
(X)HOWNORMATIVEREASONSDON’TEXPLAIN
In which I reject two accounts of how normative reasons explain. I re-introduce talk ofnormativereasons,definingthemasthingsthatmakeactions,insomerespect,worthdoing.Iaskhowitisthatwemanagetoexplainouractionswhenwesaythatweactedbecauseofanormativereasontherewastoact;forinstance:howisitthatIexplainwhyItookmyumbrellawhen I say that I took it because it was raining? I suggest that the fact that it was rainingexplainswhyItookmyumbrellaeither‘elliptically’, ‘directly’or‘indirectly’. Inotethatwhichanswer one accepts will depend on one’s response to The Explanatory Exclusion Problem:elliptical theorists accept the conclusion of the Problem, direct theorists reject the firstpremise, and indirect theorists reject the second. I set out the problemswith elliptical anddirecttheories.
(XI)THEEXCLUSIONPRINCIPLEISFALSE
In which I show that the exclusion principle is false. I provide two counterexamples to theexclusion principle, one involving causal explanation and another involving non-causalexplanation. I suggest that they are counterexamples because in each case the purportedlyexcluded fact explains the explanandum by explaining something that, in turn, explains the
22
explanandum. I suggest that theproblemwith theexclusionprinciple is that itdiscriminatesagainst all but the most proximal explanations of any given explanandum, and that this isproblematic at leastpartlybecauseweare typically interested inmoredistal explanations. Iexplainwhereour reasoningwentwrongandwhich full explanationanapparentlyexcludedfactispartof.
(XII)EXPLAININGWHYWEACT
Inwhich I say how normative reasons (and the appearance of them) explainwhywe act. Isuggest that normative reasons explain an agent’s action by explaining their belief that, inturn,explainstheagent’saction.Isuggestthattheyexplainanagent’sbeliefbyexplainingtheappearanceofthemthat,inturn,explainstheagent’sbelief.Isetouttheimplicationsofthisviewforexplanatoryrationalismandforanti-psychologicaltheoriesofreasonsmoregenerally.
(XIII)EXPLAININGWHYITISRATIONALTOACT
InwhichIsaywhensomethingexplainswhyit’srationaltoact,andwhenitdoesn’t.Isuggestthat normative reasons or appearances explainwhy it is rational to act only if they explainthosebeliefsthatinturnexplainwhyitisrationaltoact.Inotethatitistemptingtoinferthatif an agent’s belief explains why it is rational for them to do some action then whateverexplainsthatbeliefalsoexplainswhyitisrationalforthemtodothataction.Ishowhowthatinference leads to an apparent dilemma for explanatory rationalism. I counsel against thatinference, by noting that different kinds of explanatory relationsmay not be transitivewitheachother. I thensetout the taskahead: showing that theexplanatory relationsconcernedaretransitivewhen,andonlywhen,explanatoryrationalismneedsthemtobe.
(XIV)THEMYSTERYRELATION
In which I introduce the mystery relation. I suggest that a mysterious, non-causal relationobtainsbetweenabeliefandthejustificationthatitisbasedonwhenthatbeliefisjustified.Iargue that the mystery relation must be non-causal, because, as deviant causal chainsdemonstrate, a merely causal relation between a belief and some justification for it is notsufficient for that belief to be justified. I suggest that this exact same mysterious relationrelates: the belief that p to the fact that p when the belief that p is knowledgeable; ajustification for the belief that p to the fact that p when that justification affords theopportunityforknowledge;andanactiontosomebeliefthatexplainswhyitisrationalwhenthatactionisdoneintentionally.Iargue,furthermore,thatthismysteryrelationisatransitive,explanatoryrelation.
(XV)MYSTERYRELATIONSANDWHYITISRATIONALTOACT
Inwhich I say thatmystery relations are transitivewith the explanatory relation involved inexplainingwhyitisrational.Ilabelthesortofexplanatoryrelationthatobtainsbetween(i)thefactthatIbelievethatitisrainingand(ii)thefactthatitisprotantorationalformetotakeanumbrella,the‘E’-relation’.IarguethatthemysteryrelationistransitivewiththeE’-relation.Ishowhowthisaccordswithourintuitionsinsomeoftheexamplesalreadyconsidered.
(XVI)ANEWTHEORYOFREASONS
InwhichIsetoutmytheoryofreasons.IdiscusswhatexplanatoryrationalismsaysabouttheapplicationofeachreasonexpressiontothecasewhereItakemyumbrellahavingseenthatitis raining. Ishowhowexplanatoryrationalismsolvestheproblemsfacedbyothertheories. Isuggest that the best theory of reasons is a pluralist theory of reason that combinesexplanatory rationalism and favourism; I call this theory ‘new pluralism’. I show howexplanatory rationalism enables new pluralism to meet the main challenge to pluralisttheories.
23
(I)
OnTheoriesofReasons
InwhichIsayhowweshouldcategorisetheoriesofreasons.Iarguethat,ifourcategorisationschemaistocaptureatleastthemaintheoriesofreasons,thenit should by allow for the possibility of asmany kinds of reason as there arereasonexpressions. I say that insteadof distinguishingbetweenpsychologicaland anti-psychological theories we should categorise theories of reasonsaccording to what they say, for each reason expression, about the relationbetween the reasons picked out by that expression and the actions forwhichthey are reasons. And I say that we should eschew the terminology of‘normative’ and ‘motivating’ reasons in our categorisation schemaand in ourtheorising, because, as they are standardly defined, they make substantiveclaimsaboutwhatitistobeareasonthatitisproperlythebusinessofatheoryof reasons to determine, and that, moreover, those claims systematicallydisadvantage ‘psychological’ theories of reasons. To frame the discussion ofpracticalreasons interms‘normative’and‘motivating’reasons is, Isuggest,aTrojanhorsethatearnesttheorisingoughttoreject.
Within the domain of practical reasons, it is common to distinguish two kinds of reason:
normative and motivating. It is typical1to then categorise theories of reasons according to
whether or not they take reasons of each kind to be (exclusively) features of the agent’s
psychology.Ifwesaythatatheoryis‘psychological’withrespecttoagivenkindofreasonifit
says that a reason of that kind is always a feature of the agent’s psychology, and ‘anti-
psychological’ if it denies that view, we have the following, ‘standard’ categorisation of
theoriesofreasons:
Theory Normativereasons Motivatingreasons
‘TheReceivedView’2 Anti-psychological Psychological
‘Anti-psychologism’ Anti-psychological Anti-psychological
‘Extremepsychologism’3 Psychological Psychological
TableI-1:The‘standard’categorisationoftheoriesofreasons
Despite its ubiquity, this categorisation schema is, as I will argue, only useful within the
confinesofafairlynarrowtheoreticalframework.Inparticular,itisnotwellsuitedtothetask
(forwhichitmaywellnothavebeenintended)ofdistinguishingmanyofthemaintheoriesof1(E.g.Dancy2000;Sandis2013;O’Brien2015)2ThisviewisthetargetofDancy’scriticisminhisPracticalRealityand iswhatMitova(2015)callsthe‘Standard Story’. It is typically attributed to Smith (1987), however, in §(V) I will challenge thisattribution.3‘Extremepsychologism’isanamethatIhaveborrowedfromMitova(2015),whodefendsthatview.
24
reasons fromoneanother,because: (i) itonlyallowsfor twokindsofpractical reason,while
someprominenttheoriesallowformore;(ii)thecategorisationoftheoriesintopsychological
and anti-psychological cannot discriminate between distinct anti-psychological theories; and
(iii) given the way that ‘normative’ and ‘motivating’ reasons are now standardly defined,
assuming that they are picked out by any given reason expression involves a theoretical
commitmentthatatleastsometheoriesofreasonswouldandshouldreject.
IncriticisingthiscategorisationschemaImeantocriticisenoworkinparticular.Itmaybethat
the schema, so represented, isa strawman in so farasnooneholds it in itsentirety (orat
leastnot for thepurposeofsortingdifferenttheoriesof reasons).Nonetheless, I take it that
everythingthatIcritiqueisheldbyatleastsome,andsomeofthatwhichIcritiqueisheldby
many–so,sincemydiscussionisoftheconstituentpartsoftheschema,whetherornotthe
complex isgenerallyendorsed isneitherherenor there. If it turnsout thateveryoneagrees
thatthestandardschemaisofnouseincategorisingdistincttheoriesofreasonsthenwhatthe
discussionthatfollowswillhavedoneisdemonstratewhythatisthecase,aswellasoffering
analternativethatisbettersuitedtothatend.
Tore-iterate,myaim,inwhatfollows,istodemonstratethatwhatIhavecalledthe‘standard’
schemaisnotwellsuitedtothetaskofdiscriminatingbetweendifferenttheoriesofreasons,
andtoproposeanalternatecategorisationschemathatisbettersuitedtothattask.
1 Howmanykindsofreasonarethere?
Ingeneral,ifweuseoneexpressiontorefertoonethingandadifferentexpressiontoreferto
somethingelse,wehaveaprima faciebasis for thinking that the twothingsareofdifferent
kinds.Ofcourse,itmayturnoutthattwodifferentexpressionsareusedtopickoutthesame
kindofthing(e.g.‘abachelor’and‘asingleman’),butbeforewedeterminewhatkindofthing
theypickout,adifferenceintheformofanytwoexpressionsgivesusaprimafaciebasisfor
thinkingthattheypickoutdifferentkindsofthing.
Conversely, ifweuse thesameexpression to refer to twodifferent things,wehaveaprima
faciebasisforthinkingthatthedifferentthingsareofacommonkind.Ofcourse,itmayturn
outthatthesameexpressioncanbeusedtopickoutdifferentkindsofthing(asinthecaseof
homonymslike‘abat’(ananimal)and‘abat’(anitemofsportsparaphernalia)),butbeforewe
determinethatthatisthecase,thesamenessoftheexpressiongivesusaprimafaciebasisfor
thinkingthatthethingsthatitpicksoutareallofacommonkind.
25
So,whenwewanttoworkouthowmanykindsofreasonthereare,agoodwaytostartisto
lookatthedifferentexpressionsthatweusetopickoutthingsthatwecall‘reasons’.Hereare
whatIwilltaketobethemainones:
- Reasonstherearetoact;
- Reasonsforandreasonsagainstacting;
- Reasonsonehastoact;
- One’sreasonforacting;and
- Reasonswhy.
All of these expressions pick out reasons of one kind or another. Each expression has a
differentform,so,Isuggest,atleastbeforewedeterminewhatkindofreasontheyeachpick
out,wehaveaprimafaciebasisforthinkingthattheyeachpickoutdifferentkindsofreason.
Moreover,becausehomonymsaretheexception,nottherule,wehaveaprimafaciebasisfor
believingthattheyeachpickoutonlyonekindofreason.Ofcourse,furtheranalysismayyield
thefindingthatsomeoftheseexpressionspickoutthesamekindofreason,orthatsomepick
outmore thanonekindof reason,but thatshouldbeaconclusionofour theorising,notan
assumptionwithwhichwebegin.So,sincewehavefivedifferentreasonexpressions,wehave
a prima facie basis for thinking that there are five different kinds of reason – one
correspondingtoeachreasonexpression.
Someclarifications,fortheavoidanceofdoubt:firstly, Itaketwokindsofreason,kindAand
kindB,tobedifferentkindsofreasonifandonlyiftheconditionsforbeingareasonofkindA
differ from the conditions for being a reason of kind B. Secondly, saying that there are
differentkindsof reasondoesnotmeanthatoneandthesamethingcannotbeareasonof
each kind, but nor does it mean that they can. Rather, it is for your theory of reasons to
determinewhetherornotreasonsofonekindcanalsobereasonsofanotherkind;anditwill
dosobytellingyouwhattheconditionsforbeingareasonofeachofthosekindsare.Lastly,
tworeasonexpressionspickoutreasonsofthesamekindifandonlyiftheyareco-extensive.
1.1 Reasonexpressionsinthestandardschema
Thestandardschemaassumesthattherearejusttwokindsofreason–thisistheassumptionI
wish to criticise in this section. The standard schema also makes assumptions about the
conditions under which something is a reason of either kind, which further restricts its
usefulness,butIwillreturntothatissuein§4.
26
Giventhateachreasonexpressionpicksoutreasonsofonekindoranother,inassumingthat
there are only two kinds of reason, the standard schemamust assume that several of the
reasonexpressionsconsideredpickout reasonsof thesamekind.Andthat is indeedso: the
standard schema typically assumes (either implicitly or explicitly) that the first three reason
expressions (‘reasons thereare toact’; ‘reasons for/against acting’; and ‘reasonsonehas to
act’)allpickoutreasonsofonekind,whilethelattertwoexpressions(‘one’sreasonforacting’
and‘reasonswhyoneacts’)pickoutreasonsoftheotherkind.
The standard schema may well be correct about the co-extensivity of some of these
expressions – but that is a conclusion to be argued for, it is not something to be assumed
beforeonehasbeguntheorisinginearnest.Onereasonnottomakethisassumptionfromthe
outsetisthatwehaveaprimafaciebasisforthinkingthatthedifferentexpressionsaren’tco-
extensive–namely,thefactthattheyaredifferentexpressions.
However,thedifferencesbetweentheexpressionsnotwithstanding,assumptionsoftheirco-
extensivitymightbetolerableiftheequivalencebetweenparticularexpressionswereobvious,
oratleastun-contentious–butthatiscertainlynotthecase.Indeed,asweshallsee,several
theoriesofreasonsalreadydistinguishbetweenexpressionsthatthestandardschematakesto
beco-extensive,andthosetheoriesconsequentlyescapecategorisationwithinthatschema.
I want to make the case for a categorisation schema that allows for the possibility of
distinguishingbetweeneachofthereasonexpressionsIlist.Ialsowanttomakeclearwherea
theoryofreasonsisrequiredtodrawadistinctionbetweendifferentexpressions.And,lastly,I
wanttosaywhichofthesereasonexpressionsweshouldandwhichweshouldnotincludein
our categorisation schema. The following sections discuss each expression, with a view to
achievingeachoftheseaims.
1.2 Reasonstherearetoact
Thereisareasonformetogoforaswim.ToavoidcontroversyIwon’tsaywhat it is–butI
takeittobeuncontroversialtosaythatif(andonlyif)thereisareasonformetoswimthen
something4is the reason that there is. I take it to be likewise uncontroversial to say that if
somethingisareasonthereisformetoswimthenitisareasonformetoswim.5
4Imakenoassumption(yet)abouttheontologyofthatsomething–itcouldbeafact,astateofaffairs,amentalstateorwhateveryoulike.5These remarks are obviously intended to generalise beyond me and my swimming. But, for theavoidanceofdoubt, I take thestatement that ‘something is reasonthere is forA toφ’ tobe logicallyequivalent to ‘that same thing is a reason forA toφ’ and ‘there is a reason forA toφ’. I take thedifference in the forms of words here to be so slight as to preclude effective distinction – so our
27
Hereisaseemingly‘pre-theoretic’observationwecanmakeaboutthekindofreasonpicked
outbytheexpression‘thereasonstherearetoact’:somethingcanbeareasonforAtoφeven
ifAdoesnotφ.Forinstance,therecanbeareasonformetoswimevenifIdon’tswim.That
is,itdoesnotimpingeonthereason-hoodofareasontoactifonedoesnotdotheactionthat
itwasareasontodo.
Isuggestwesaythatifthereason-hoodofreasonsofacertainkindisnotdependentonthe
occurrence of the action for which they are a reason, then reasons of that kind are
‘independent’oftheactionsforwhichtheyarereasons.6Thus,reasonstoactareindependent
oftheactionsforwhichtheyarereasons.
1.3 Reasonsforandreasonsagainstacting
I’mdebatingwhetherornottoswim:therearereasonsforandreasonsagainstmyswimming7
–don’tworryaboutwhattheyare.Likereasonstherearetoact,reasonsfororagainstacting
are independent of the actions for which they are reasons: something can be a reason for
swimming(orareasonagainstswimming)evenifIdon’t(ordo)swim.
While they differ in their form, I have grouped ‘reasons for’ and ‘reasons against’ because
thereseemstobeaclearsenseinwhichtheirmeaningsareinter-related–theyareopposites.
Sowhateverwelearnaboutthekindofreasonpickedoutbyonewilltellussomethingabout
the kind picked out by the other. As is typical in the literature, I will mostly restrict my
discussiontothe‘reasonsfor’expression.
Aswenoted in §1.1, it is typically assumed that ‘reasons there are to act’ and ‘reasons for
acting’areco-extensiveexpressions(it isalsotypical toassumethat ‘reasonsnottoact’and
‘reasonsagainst acting’are likewiseco-extensive). I think this is amistake– I think that the
prepositionsmattertothemeaning,andthatthetwoexpressionspickoutdifferentkindsof
reason.However,Iamnotarguingforthisclaimatthisstage;fornow,Iamonlyarguingthat
ourcategorisationschemashouldallowforthepossibilityofdifferencebetweenthem.
Thatis,atthisstageIneitheracceptnordenythat‘reasonstherearetoact’and‘reasonsfor
acting’alwayspickoutthesamekindofreason:allthatIamsayingisthat,inspiteofthefact
thattheyarebothindependentoftheactionsforwhichtheyarereasons,thedifferenceinthe
categorisation schema need not distinguish between them. Nonetheless, I state these assumedequivalenceswithaviewtomakingmyassumptionsclear.6 This will contrast them with those kinds of reason (introduced in §1.5) whose reason-hood isdependentontheoccurrenceoftheactionforwhichtheyarereasons,whichIcall‘dependent’reasons.7Again, if thereare reasons for swimming thensomething isa reason for swimming,and if therearereasonsagainstswimmingthensomethingisareasonagainstswimming.Andagain,theseequivalencesgeneralisebeyondmeswimming.
28
form of these reason expressions gives a prima facie basis for thinking that they pick out
different kinds of reason.Moreover, since I eventuallywant to advocate a theory thatdoes
distinguish between the reasons picked out by these expressions, if my theory is to be
represented withinmy own categorisation schema, I had better allow for the possibility of
suchadistinction.
1.4 Reasonsonehastoact
Ihaveareasontoswim.Again,ItakeittobeuncontroversialtosaythatifIhaveareasonto
swimthensomethingisthereasonthatIhavetoswim.8And,likereasonsthereareformeto
swim(andreasonsforswimming),reasonsthatIhavetoswimareindependentoftheactions
forwhichtheyarereasons:onecanhaveareasontoswim(oreat,orgototheshops)without
doingso.
Aswe have seen, it is typical to take the expressions ‘the reasons one has to act’ and ‘the
reasons there are to act’ to be co-extensive. As it happens, I share this view, but we
nonetheless should not start with this assumption, much less incorporate it into our
categorisationschema,forthefollowingreasons:firstly,asI’vealreadystressed,thedifference
betweentheformsoftheexpressionsgivesusaprimafaciebasis forthinkingthattheypick
outdifferentkindsofreason,soit isodd, ifnotcounter-productive,tostart(asthestandard
categorisation schema does) by assuming that they don’t. Secondly, andmore importantly,
severalauthors(e.g.Hornsby2008;Schroeder2008;ComesañaandMcGrath2014)rejectthe
view that these expressions are co-extensive; and their theories consequently escape
categorisationwithinthestandardschema.Andthirdly,asweshallsee,assumingthatthereis
no distinction between the kinds of reasons picked out by these expressions conceals a
possiblesolutiontooneofthemainproblemsforthedominantaccountofthereasonsthere
aretoact.9
So,inspiteofthefactthatthekindsofreasonpickedoutbytheexpressions,‘reasonsonehas
toact’and‘reasonstherearetoact’,arebothindependentofthereasonsforwhichtheyare
actions,wenonethelesshavegoodgrounds forusingacategorisationschemathatallowsat
least for the possibility of distinguishing between the kinds of reason picked out by these
expressions.
8Thismakesnoassumptionabouttheontologyofthatthing(inparticular–sayingthatsomethingisthereason Ihavetoswimdoesnotassumethatthatthingneedbeafactratherthan,say,an intentionalobject).9Seethediscussionofthe‘TheRationalActionProblem’in§(II)4.1.
29
1.5 One’sreasonforacting
Ithinkabouthowswimmingwillimprovemymood,howIwillsleepbetterifIswim,andother
thingsbesidesand,aftersomedeliberation,Idecidetoswim.Iheadoffanddoit:Iswimfora
reason.Asever,Iwon’tsaywhatmyreasonforswimmingwas.10
Some, hopefully, un-contentious equivalences: I swam for a reason if and only if something
wasmyreasonforswimming,andsomethingwasmyreasonforswimming ifandonly if (i) I
swamforthatreasonand(ii)itwasthereasonforwhichIswam.11Thelatterremarkisperhaps
themostcontentious(buthopefullynotespeciallyso)oftheequivalencesIhavedrawn,inso
farasItake‘myreasonforswimming’tobeequivalentto‘thereasonforwhichIran’inspite
oftheobviousdifferenceintheformoftheseexpressions.Itiscertainlyinprinciplepossibleto
distinguishbetweenthem,however,IhaveelectednottoonthebasisthatIknowofnotheory
that does (so the equivalence is seemingly uncontentious) and so, for the sake of brevity, I
assumenodistinction fromtheoutset.Perhaps Iamnot followingmyown instruction–but
my assumption is at least plain, so that those who disagree with it may revisemy schema
knowingasmuch.
While the standard categorisation schema draws no distinction between the reason
expressions discussed in the previous three sections, it does, as I have noted, distinguish
betweenthosereasonexpressionsand‘theagent’sreasonforacting’.Now,whileIwantusto
abandon the standard categorisation schema, I agreewith its implicit verdict that there are
unequivocalgroundsfordifferentiatingthekindsofreasonpickedoutby‘thereasonforwhich
the agent acts’ from the kinds picked out by the expressions already considered, which go
beyondameredifferenceinform.12WhatdoItakethosegroundstobe?
Considerthis:somethingcannotbeone’sreasonforφingunlessoneφs.Thatis,Ican’tswim
for a reasonunless I swim, so something can’t be the reason forwhich I swamunless I did
indeedswim.Andthisistrueeveniftheactionisinthefuture:somethingmightnowbethe
reasonforwhichIwillswim,butitisonlysoifIdoindeedswim(ifIdon’tgoontoswimthenit
isn’t now the reason for which I will swim). So, unlike the reason expressions already
considered,thereason-hoodofthereasonspickedoutbythisexpressionarenotindependent
10In particular, the story of my deliberative process need lead us to no conclusions about what myreasonforswimmingwas.11Again, these remarks should generalise beyond me and my swimming, but, for the avoidance ofdoubt,Itakethestatementthat‘somethingwasA’sreasonforφing’tobelogicallyequivalentto‘thatsamethingisareasonforwhichAφ’d’and‘Aφ’dforareason’.12Icallthisits‘implicitverdict’onlybecausethestandardapproachdoesnotgenerallytakeadifferenceinformtobesufficientgroundsforallowingthepossibilityofadistinction–asthefailuretodistinguishbetweenthereasonexpressionsalreadyconsidereddemonstrates.
30
of theactions forwhichtheyarereasons–theyare, letussay, ‘dependent’onthem, inthe
sensethatthereason-hoodofsuchreasonsdependsontheoccurrenceoftheactionforwhich
itisareason.
So,thekindofreasonpickedoutbytheexpression‘thereasonsforwhichAφs’isdistinctfrom
the kind(s)13picked out by the reason expressions of the previous sections (because the
conditionsforbeingareasonoftheformerkinddifferfromtheconditionsforbeingareason
ofthelatterkind(s)).14Thismeansthatnotonlymustourcategorisationschemaallowforthe
possibilityofdistinguishingthissortofreasonfromthosealreadyconsidered,butanytheory
of reasons should alsomake clear that this expressiondoes pick out a reasonof adifferent
kind.
1.6 Reasonswhy
Therewas a reasonwhy I swam. Likemy reasons for swimming (or the reasons forwhich I
swam),thereasonswhyIswamaredependentonthatwhichtheyareareasonfor:ifthereisa
reasonwhyIswamthenImusthaveswum.
Aswehavenoted,thestandardschemadoesnotsystematicallydistinguishbetweenthekinds
of reasonpickedoutby ‘the reasonwhysomeoneacts’and ‘the reason forwhich theyact.’
Nonetheless,thedifferenceintheirformsstillgivesusaprimafaciebasisfordistinguishingthe
twoexpressions,and this isdespite the fact that theybothpickoutakindof reason that is
dependentonthatwhichitisareasonfor.Moreover,recentscholarshiphasforcefullymadea
case for distinguishing between the kinds of reason picked out by the two expressions that
goesbeyondthemeredifferenceintheirform.15
Firstly,wecanreadilyobservethat,inspiteoftheirbothbeingdependentreasons,something
canbeareasonwhyanagentactswithoutbeingtheirreasonforacting.Forexample,thefact
thatI’mchoppingonionsmaybeareasonwhyI’mcryingwithoutbeingmyreasonforcrying(I
amnotcryingforareason,thoughthereisareasonwhyIamcrying).Likewise,thefactthat
Anthony was given a posthypnotic suggestion might be a reason why he drinks vinegar,
withoutbeinghis reason fordrinking it, and that is soeven ifhedrinks it fora reason.One
13The parenthetical pluralisation is to note that these reason expressions may or may not pick outdifferentkindsofreason.14Weshouldnotethatthisisnottosaythatoneandthesamethingcannotbeareasonofeachkind(asIalreadyremarkedintheclarificationsattheendof§1.1);itisjusttonotethattheexpressionspickoutreasonsofdifferentkinds.Thatis,totheextentthatasinglereasoncanbepickedoutbytheexpression‘the reason for which they acted’ and, for instance, ‘a reason they had to act’ (and, at this stage, Iassume nothing about whether or not they can) it is because that which is picked out is both adependentandanindependentreason–itistwodifferentkindsofreason.15See,inparticular,Alvarez(2010)butalsoAudi(2001),Dancy(2000)andHieronymi(2011).
31
neednot,Ithink,acceptanyparticulartheoryofreasonsinordertoacceptthetruthofthese
claimsand,moreover,theirtruthalreadyprovidesabasisfordistinguishingbetweenthekinds
of reason picked out by the expressions ‘the reasons why one acted’ and ‘the reasons for
whichoneacted,’thatgoesbeyondameredifferenceinform.
Secondly,andmoregenerally,thekindofreasonpickedoutbythe‘reasonwhy’expressionis
different to thosepickedoutbyallof thereasonexpressionsalreadyconsideredbecause its
relata are different. While all the reasons picked out by the reason expressions already
considered relate toactions,a reasonwhy is related toa fact (orproposition). For instance,
compare the relata of these expressions: ‘my reason for swimming’; and ‘the reasonwhy I
swam’–whileswimmingisanaction,‘Iswam’isasentence.
Thedifference in relata is alsoof some relevance toour thirdobservation:because reasons
why relate to facts, the sorts of things they are reasons for need not be in anyway agent
involving.Forinstance,therearereasonswhytheEarthorbitstheSunandtherearereasons
whytheDawlishseawallcollapsedin2014.Neitheroftheseisanagent-involvingoccurrence;
theEarthandtheDawlishseawallaren’t/weren’tagentsandtheydidn’tdowhatthey‘did’for
reasons.
Ifwesaythatreasonsthatareessentiallyagent-involvingare‘practical’kindsofreasons,then
thekindofreasonspickedoutbythe‘reasonwhy’expressionisnotapracticalkindofreason.
This does not mean ‘reasons why’ never relate to agent-involving activities, that is, for
instance,itdoesnotmeanthattherecan’tbereasonswhyagentsdothings;itjustmeansthat
‘reasons why’ are not exclusively agent-involving (because they sometimes don’t involve
agents).
In contrast, there are no reasons for the Earth to orbit the Sun (or reasons for the Earth’s
orbitingtheSun,orreasonstheEarthhastoorbittheSun),andthereisnothingthatcouldbe
called‘theEarth’sreasonfororbitingtheSun’becausethekindsofreasonpickedoutbythese
expressionsarepractical;sotheydon’tapplytothingsthataren’tagents,liketheEarth.Unlike
the‘reasonwhy’expression,then,allthereasonexpressionsoftheprevioussectionspickout
practicalkindsofreason.16
So,thekindofreasonpickedoutbythe‘reasonwhy’expressionisdistinctfromthekindsof
reasonpickedoutbytheexpressionsalreadyconsidered(becauseithasdifferentrelataandit
isn’tpractical).This isn’t tosay thatsomething that isa reasonofapracticalkindcannever
16Ofcoursetheseremarksdonotamounttoaproofthatthekindsofreasonpickedoutbythepreviousexpressionsarepractical–butItakethattobeclearenoughthatwecanassumeitwithoutcontroversy.
32
alsobeareasonwhysomeonedoessomething17;itisjusttoacknowledgethattheconditions
for being a practical reason differ from the conditions for being a reasonwhy. That is, the
conditionsforbeingareasonofthekindpickedoutbythe‘reasonwhy’expressiondifferfrom
theconditionsforbeingareasonofthekindpickedoutbyanyoftheotherexpressions;the
‘reasonwhy’expressionpicksoutadifferentkindofreasontothereasonexpressionsalready
considered.
1.7 Asummaryofreasonexpressions
Isaidthattwokindsofreasonaredifferentkindsofreason if,andonly if, theconditionsfor
beingareasonofonekinddifferfrombeingareasonoftheotherkind.18Thediscussionofthe
previoussectionshasestablishedthatthereareatleastthreekindsofreason:independent&
practical reasons; dependent & practical reasons; and independent & non-practical reasons
(seeTableI-2).
Reasonexpression Dependent/Independent Practical/Non-practical
Reasonstherearetoact Independent Practical
Reasonsforacting Independent Practical
Reasonsonehastoact Independent Practical
One’sreasonforacting Dependent Practical
Reasonswhy Dependent Non-practical
TableI-2:Reasonexpressionsandthekindsofreasontheypickout
In addition, I have argued that, in spite of the fact that the standard categorisation schema
ordinarilydrawsnodistinctionbetweenthereasonstherearetoact, reasonsforacting,and
thereasonsonehastoact(thatis,theexpressionsthatpickoutindependentkindsofreason)
oranagent’sreasonforactingandthereasonswhytheyact(thatis,theexpressionsthatpick
outdependentkindsofreason)weshouldnonethelessuseacategorisationschemathatallows
for thepossibilityofdistinguishingbetween them, in particular, so thatwe candiscriminate
betweenalreadyexistingtheories.19
17Indeed, it is common toassume that reasonsof thekindpickedoutby theexpression ‘theagent’sreason for acting’ are always reasonswhy the agent acts (see §(IV)1.2 for further discussion of thispoint).18Tostressapoint Ihavealreadymade:thisdefinition isnotmeanttoprecludethepossibilityofoneandthesamethingbeingareasonoftwodifferentkinds.19Myargumentswere,inbrief,thatthedifferenceinformprovidesaprimafaciebasisforbelievingthatthey pick out different kinds of reason, that some theories already distinguish between the reasons
33
Thus,Iproposethefollowing,provisionalcategorisationschema:
Reasonsthereareto
act
Reasonsforacting
Reasonsonehasto
act
Reasonsforwhichoneacts
Reasonswhyoneacts
Theoryofreasons Claim… Claim… Claim… Claim… Claim…
TableI-3:Aprovisionalcategorisationschema
In the next section I will consider the character of the claims that differentiate theories of
reasons.Havingdoneso,Iwillsuggestarevisiontothisprovisionalschemathatomitsthenon-
practicalkindofreason,reasonswhy.
2 Claimsaboutreasons
WhatIhavecalledthe‘standardschema’categorisestheoriesonthebasisofwhetherornot
they take reasons of a given kind to be exclusively psychological. In doing so, however, this
schemaobscures themany differences betweendistinct anti-psychological theories. Indeed,
whatthestandardschemacalls‘anti-psychologism’isnotsomuchatheoryofreasonsasitisa
collectionofdifferenttheoriesofreasonsthat,despitesharingacommitmenttothefalsityof
thepsychologicalviewofanykindofreason,varyinmanyotherrespects.
For instance,differentversionsofanti-psychologismdisagreeonwhetherornotreasonscan
befalse,andwhetherornottheexistenceofareasondependsonanagent’sperspective–but
the standard schema is incapable of recognising such disagreement. That is, of course,
perfectly fine so long as one’s focus is on psychologism and its discontents, but it becomes
quite inappropriatewhen onewants to categorise theories of reasons from amore general
perspective.
Furthermore, although there are alternatives to the psychological vs. anti-psychological
categorisation that can differentiate between different forms of anti-psychologism, they, in
contrast,obscurethedifferencesbetweenpsychologicaltheoriesandthedifferentvariantsof
anti-psychologism.Forinstance,Turri(2009)categorisestheoriesofreasonsaccordingtotheir
ontology–separatingtheoriesthattakereasonstobefacts,fromthosethattakethemtobe
mental states and again from those that take them to be intentional objects (that is, the
contents of mental states). However, a strict 20 ontology-based categorisation is no less
therearetoactandthereasonsonehastoact,andthatIwishtodistinguishbetweenthereasonstherearetoactandreasonsforacting.20Isay‘strict’becausemanytake‘factualism’tobetheviewthatreasonsarefactsandthattheyarenotexclusively psychological (indeed, it’s common to use ‘factualism’ to name the view that reasons arefacts that favour actions that are consequently typically non-psychological). This, of course, is not a
34
problematic than the standard schema’s approach, because (as Alvarez (2016b, 3) notes) it
cannotdiscriminatebetweenapsychologicaltheorythatsaysthatreasonsareallfacts(about
anagent’smind)andananti-psychologicaltheorythatsimilarlysaysthatreasonsareallfacts
(aboutallkindsofthings),sinceboththeorieshaveacommon,factualistontology.21
Alternatively, Lord (2015), Alvarez (2016a) and Way and Whiting (2017) divide theories
accordingtowhetherornottheytaketheexistenceofagivenkindofreasontodependonan
agent’s perspective. However, the categorisation of theories into ‘perspectivist’ and
‘objectivist’similarlyfailstodistinguishperspectivistanti-psychologicaltheoriesfrom(equally
perspectivist)psychologicaltheories.
Onecould,perhaps,surmounttheseproblemsbycategorisingthedifferenttheoriesonseveral
or all of these aspects,22but I think that such a response would be to miss the point. The
problemwithalloftheseapproachesisnotthattheyeachonlycaptureoneaspectofvariation
betweentheories.Rather, Isuggest, it isthattheyputthecartbeforethehorseinthesense
that they classify theories of reasons according to their stance on a particular topic, rather
thanwhatdeterminesthatstance,whichis,Isuggest,whatthetheoryactuallyisabout.
Towit,theanswerstothequestionsofwhetherornotreasonsarepsychological,ofwhattheir
ontology is andofwhetherornot their existencedependsonanagent’sperspective are all
determined by the answer to this, more fundamental question: for a given kind of reason,
whatistherelationbetweenreasonsofthatkindandtheactionsforwhichtheyarereasons?
Letuscall this relation the ‘reason-relation’: if you think that the reason-relation foragiven
kindofreasonissuchthat itcanrelatenon-psychologicalthingstoactions,thenyourtheory
willcomeoutanti-psychologicalforthatkindofreason,orifyouthinkthatthereason-relation
for a given kind of reason is such that it is not a factive relation, then you will, perhaps,
conclude that reasons of that kind are not facts (perhaps they are intentional objects – but
thatwilldependonwhatyouthinkthereason-relationis…).Inthisway,whatatheorytakes
thereason-relationtobejustdeterminesitsanswertotheseotherquestions.
Thejobofatheoryofreasonsis,interalia,tosayhowmanykindsofreasonthereareandto
then explainwhat the reason-relation is for each kind of reason. The standard approach to
strictlyontologicalclassification–anontologicalclassificationisonethatclassifiestheoriesonthebasisoftheirontologyalone.21Thequestionofwhattheyarefactsaboutisnotanontologicalone,andsocannotfeatureinastrictlyontologicalclassification.22 For instance, one could differentiate between theories that are perspectivist & psychological;perspectivist&anti-psychological;andobjectivist&anti-psychological.
35
categorising theories puts the cart before the horse because rather than categorising them
accordingtotheirclaimsaboutthereason-relation(whichisthesubstanceofthetheories),it
categorisesthemaccordingtotheconsequencesoftheirclaimsaboutthereason-relation.Of
course,thismaywellbefineforthepurposestowhichthecategorisationishabituallyput(e.g.
discussing the merits of psychological vs. anti-psychological or perspectivist vs. objectivist
theories), but it is not fine for the purpose of actually categorising the different theories of
reasons.23To do that, we should put the horse before the cart and categorise theories of
reasonsaccordingtowhattheysayaboutthereason-relationforeachreasonexpression.
2.1 Familiesofclaims
Sowhatdotheysay?Foranygivenreasonexpression,mosttheoriesofreasonssubscribetoa
viewthatbelongs tooneof three families: ‘favourism’, ‘deliberativism’and ‘psychologism’.24
Thesethreefamiliesdonotexhausttheavailableclaimsaboutanygivenreasonexpression,25
however, they cover, between them, the vast majority of claims made about each reason
expression (both considered andde facto), and they are the focusofmydiscussion inwhat
follows.
Whatthemembersofeachofthesefamilieshaveincommonisaviewofwhatittakestobea
reasonofanygivenkind:claimsinthefavourismfamilyrequirethatareasonofanygivenkind
must make an action, in some respect, worth doing;26claims in the deliberativism family
requirethatareasonofanygivenkindmustbesomethingtheagenttooktomaketheiraction,
insomerespect,worthdoing;andclaimsinthepsychologismfamilyrequirethatareasonof
anygivenkindmustbeafeatureoftheagent’spsychologythatrationalisestheiraction.Table
I-4setsouttheclaimsforeachreasonexpressionineachfamily(atthisstageIwill leavethe
23Nor,indeed,isitfineforreallygettingtogripswitheverythingthatiswrongwithaparticulartheory–sinceitmeansconcentratingonjustoneaspectofthetheory.24Thename‘psychologism’ItakefromDancy(2000),theothersaremyowninvention.25A noteworthy omission is Kearns and Star’s (2008, 2009) account according to which reasons areevidencethatoneoughttoactinacertainway.26Ihavetakensomelibertiesinrepresentingaviewthatistypicallyrenderedas‘reasonscountinfavourofactions’as‘reasonsmakeactionsworthdoing’.Idosobecauseleavingwhatitisto‘countinfavour’ofanactionun-interpretedleavesthetheoryofreasonsunder-determined.Iwant‘favourism’toreflectthewidelyheldinterpretationofthe‘countinginfavourof’relationthattakesittoeitherbetherelationof‘making,insomerespect,worthdoing’oratleastentailingthattheactionis,insomerespect,worthdoing. I specify favourist claims in terms of ‘making, in some respect, worth doing’ rather than‘favouring’ because different interpretations of the ‘counting in favour of’ relation abound – forinstance,KearnsandStar(2008)interpretwhatitisto‘countinfavour’ofactingas‘beingevidencethatone ought to so act’, which is clearly different fromwhatmakes an action worth doing.MeanwhileMitova (2016) advocates an alternative, if unspecified, construal of what it is to ‘count in favour’ ofaction,thatiscertainlynottheideaof‘makingitworthdoing’.Ireturntothisin§4.3.
36
right-hand-sideconditionsofeachaccountunexplained;theywillbeexplainedinsubsequent
chapters).
Family
Reasonexpression Favourism Deliberativism Psychologism
Foranyp,pisareasonforAtoφ…
…ifandonlyifpmakesA’sφing,insomerespect,
worthdoing.
…ifandonlyifAtakesptomakeA’sφing,insomerespect,worthdoing.
…ifandonlyifpisafeatureofA’spsychologythatrationalisesφing.
Foranyp,pisareasonforA’sφing…
…ifandonlyifpmakesA’sφing,insomerespect,
worthdoing.
…ifandonlyifAtakesptomakeA’sφing,insomerespect,worthdoing.
…ifandonlyifpisafeatureofA’spsychologythatrationalisesφing.
Foranyp,pisareasonAhastoφ…
…ifandonlyifpmakesA’sφing,insomerespect,
worthdoing.
…ifandonlyifAtakesptomakeA’sφing,insomerespect,worthdoing.
…ifandonlyifpisafeatureofA’spsychologythatrationalisesφing.
Foranyp,pisA’sreasonforφing…
…ifandonlyifpmakesA’sφing,allthings
considered,worthdoingandexplains(intheright
way)whyAφ’d.27
…ifandonlyifpisaconsiderationinlightof
whichAφs.
…ifandonlyifpisafeatureofA’spsychologythatrationalisesφing
andexplains(intherightway)whyAφ’d.
Foranyp,pisareasonwhyAφ’d…
…ifandonlyifpexplainswhyAφ’d.
…ifandonlyifpexplainswhyAφ’d.
…ifandonlyifpexplainswhyAφ’d.
TableI-4:Themainclaimsabouteachreasonexpression
Favourism, deliberativism and psychologism are, as I have said, families of claims about
reasons;theyarenottheoriesofreasons.Instead,atheoryofreasonscanbeconstructedby
selectingoneclaimfromoneofthethreefamiliesforeachreasonexpression.
Withineachfamilytheconditionsforbeing‘areasonthereistoact’;‘areasonforacting’and
‘areasononehastoact’arethesame28–soatheorythatholdsthesameviewforanytwoof
thoseexpressions(say,favourismaboutreasonstherearetoactandfavourismaboutreasons
onehas toact) takes themtopickout thesamekindof reason.But,of course, thepointof
separatingoutthedifferentreasonexpressions inourcategorisationschemaisthatatheory
neednotalwaysselectaclaimfromthesamefamilyforeachreasonexpression.Thatis,one
couldhold(assomedo)favourismaboutthereasonstherearetoactanddeliberativismabout
the reasonsonehas toact–and todoso is tosay that thoseexpressionspickoutdifferent
kindsofreason.
27It’sworthnoting that apartofwhat is implicitly required for somep thatmakesanagent’s actionworthdoingtoexplaintheiractionintherightwayisforittobeaconsiderationinlightofwhichtheyact.28Theydifferfor‘theagent’sreasonforacting’,aswenotedtheyshould,becausewealreadyknowthatthatexpressionpicksoutadifferentkindofreason.
37
2.2 Revisedcategorisationschema
Finally,note that inTable I-4 there isnodisagreementbetweendifferent familiesabout the
conditions for being a reasonwhy. As I have noted, what sets the ‘reasonwhy’ expression
apartfromtheotherreasonexpressionsconsideredisthatitdoesn’tpickoutapracticalkind
ofreason.Instead,astheclaimsinTableI-4makeclear:reasonswhyarejustexplanations(of
whysomeonedoessomething,ofwhysomethingisacertainway,ofwhysomethingoccurred,
etc.).29Theymaybecausalexplanations(suchasthereasonswhytheEarthorbitstheSunor
thereasonswhytheDawlishseawallcollapsed)buttheyequallymaynotbe:thereasonswhy
it iswrong to tortureanimals seeminglydonotcause it tobewrongand the reasonswhy a
footballplayerisoffsideseeminglydonotcausehertobeoffside–buttheyarereasonswhyit
iswrong or reasonswhy she is offside all the same. Simply put: something is a reasonwhy
someother thing is the case if andonly if it explains (causallyorotherwise)why thatother
thingisthecase.30
Ofcourse, there isscopefordisagreementaboutwhattheproperaccountofexplanation is,
and, acknowledging that,we could further analyse the ‘explains’ relation itself – thiswould
create the possibility of more discriminating categorisations. However, I do not think we
should seek to categorise theories of reasons on the basis of their preferred accounts of
explanation–thatisjustaseparatesubject.
Noneof this is to say thatexplanation isnot relevant to thediscussionofpractical reasons;
indeed,somedefendaccountsaccordingtowhichreason-relationsjustareaparticularsortof
explanatory relation.31Mypoint is only thatwe canomit the ‘reasonswhy’ expression from
our categorisation schema, since whatever dispute there is about it is properly a part of a
separatediscussion.32
29See,forinstance,Raz(2009)andAlvarez(2010)forthisview.30Onthisanalysis,‘pisareasonwhyq’and‘pexplainswhyq’and‘qbecausep’arelogicallyequivalent.31For instance, Broome (2006) argues that a reason to act is a reason why someone ought to dosomething(see§(II)3forfurtherdiscussionofthis).Andindeed,Iwilleventuallydefendtheviewthatpracticalreasonsareallreasonswhyanactionisprotantorational.32 As Raz notes: ‘Whatever one can say about [reasons why] is better explored when studyingexplanations,avoluminousphilosophicalsubject.’(2009,186)
38
Those remarkshavingbeenmade, the revised framework for the schema I amproposing is,
thus,asfollows:
Reasonstherearetoact
Reasonsforacting
Reasonsonehastoact
Reasonsforwhichoneacts
Theoryofreasons Claim… Claim… Claim… Claim…
TableI-5:Arevisedcategorisationschema
3 Categorisingtheoriesofreasons
Forthemathematicallyinclined:withinmycategorisationschema,atheoryofreasonscanbe
specifiedasa4-tuple,witheachmemberofthetuplebeingaclaimaboutareasonexpression
from a particular family, ordered as follows: ‘reasons there are to act’; ‘reasons for acting’;
‘reasonsonehastoact’;‘reasonsforwhichoneacts’.IfweletF,D,andPdenoteFavourism,
DeliberativismandPsychologismabouttherelevantreasonexpression,respectively,then,for
instance,wecanspecifysomedistincttheoriesofreasonsasfollows:(F,F,F,F);(P,P,P,P);(F,
F,D,D);(F,F,D,F).Bywayof interpretation:thefirstandsecondofthesetheoriesmakeno
distinction between any independent kinds of reason,while the third and fourth distinguish
between‘thereasonstherearetoact’and‘thereasonsonehastoact’.
Ihavesetoutthereasonexpressionsthataretobeincludedinourcategorisationschemaand
I have set out the main claims that different theories of reasons make about each reason
expression,aswellasa(formal)wayofdescribingtheoriesofreasonsintermsoftheseclaims.
Usingthisframework,inTableI-6,Icategorisedifferenttheoriesofreasons.
This categorisation is far fromexhaustive.33Moreover, the theories of someof the theorists
that I have grouped together differ in some of their intricacies; more discriminating
characterisations of the different families of claims could, perhaps, separate out those
intricacies–butthatlevelofdiscriminationdoesnotalterthemaincriticismsthatthetheories
are subject to.34Since I only need the categorisation to be fine-grained enough to sort
betweenwhatargumentsapply towhich theories, thiswilldo,and, inparticular, it supports
thatendfarbetterthanthestandardschema.
33Noteworthyomissions includeJohnBroomeandThomasScanlon,whomIstruggledtoplace. Ihavealso omitted Michael Smith from this categorization, whose work is typically associated with ‘TheReceivedView’, because, as Iwill argue in §(V)4.1, headvocates apluralist theoryof reasons,whichevadescategorisationwithinthisschema.34See§(II)3.3formoreonthis.
39
Theoriesofreasons 4-tupledescription
Reasonstherearetoact
Reasonsforacting
Reasonsonehastoact
Reasonsforwhichoneacts
‘TheReceivedView’ (F,F,F,P) Favourism Favourism Favourism Psychologism
Stout(2009),Alvarez(2010),Parfit(2011),Littlejohn(2012)
(F,F,F,F) Favourism Favourism Favourism Favourism
Dancy(2000,2014),Davis(2005),Sandis(2009)
(F,F,F,D) Favourism Favourism Favourism Deliberativism
Schroeder(2008),Comesaña&McGrath(2014)
(F,F,D,D) Favourism Favourism Deliberativism Deliberativism
Hornsby(2008) (F,F,D,D) Favourism Favourism Deliberativism Favourism
Turri(2009),Gibbons(2010),Mitova(2015)
(P,P,P,P) Psychologism Psychologism Psychologism Psychologism
TableI-6:Prominenttheoriesofreasons,categorisedinmyproposedschema
4 Normativeandmotivatingreasons
Theterminologyof‘normative’and‘motivating’reasonsiscommonthroughouttheliterature
onpracticalreasons,and,asIhavenoted,thestandardschematypicallycategorisestheories
accordingtowhat theysayabouteach.Wearenow inapositiontoseewhat iswrongwith
thisterminology.
As Iwill argue, commondefinitions of these terms covertly import a theory of reasons into
one’s categorisation schema that leaves little scope for disagreement. Indeed, I argue that,
from a rhetorical standpoint, we ought to see these definitions less as terminological
housekeepingandmoreasaTrojanhorseleftbytheanti-psychologistsforthepsychologists.
So, I suggest,whenworkingtowardsatheoryofreasons inearnest,weshouldabandonthis
terminology,atleastasitisstandardlydefined.
4.1 TerminologyorTrojanhorse?
The de facto distinction between normative (or justifying) reasons and motivating (or
explanatory)reasonsisprobablyasfollows:normativereasonsexplainwhyanagentoughtto
do something whereasmotivating reasons explainwhy they did it.While this distinction is
40
itself problematic, 35 my focus in this section is on the following, increasingly typical
characterisationsoftheseterms36:
- Foranyp,p is anormative reason forA toφ if andonly ifp counts in favourofA’sφing.
- Foranyp,pisA’smotivatingreasonforφingonlyifpissomethingAtooktocountinfavourofφing.37
Thesecharacterisationsare innocuousas theyare:onecandefineone’s termshoweverone
wishes.However,whatmakestheiruseintheorisingproblematicisthattheirdefinitionsaren’t
typicallyrestrictedtotheseinnocuousstatements,butalsoincludethefollowingclaims38:
- Foranyp,pisanormativereasonforAtoφifandonlyifpisareasonforAtoφ.
- Foranyp,pisanormativereasonforAtoφifandonlyifpisareasonforA’sφing.
- Foranyp,pisanormativereasonforAtoφifandonlyifpisareasonAhastoφ.
- Foranyp,p isA’smotivatingreasonforφing ifandonly ifp isa reasonforwhichAφ’d.
I call these latter remarks ‘claims’ because, by associating these terms also with reason
expressions, these additional remarks go beyond mere definitions and well into what it is
properlythebusinessofatheoryofreasonstodetermine.
35Forinstance:noteverythingthatexplainsanagent’sactionisthesortofthingwewouldwanttocalla‘motivatingreason’(seeAlvarez2010).36Forinstance:‘Mostcontemporaryphilosophersstartbydistinguishingtwotypesofreasonforaction:“normative” reasons– that is, reasonswhich,very roughly, favouror justifyanaction,as judgedbyawell-informed, impartial observer; and “motivating” reasons – which, again roughly, are reasons the“agent”(thatis,thepersonacting)takestofavourandjustifyheractionandthatguidesherinacting.’(Alvarez 2016a, 1) It’s worth noting that this is not howmotivating reasonswere always defined, inparticular,MichaelSmith(1987)hasquiteadifferentnotioninmind,asDarwall(2003)observes–see§(V)4forfurtherdiscussionofthispoint.37Note that this is only a necessary condition, not a sufficient one.As I havenoted, it is common toassume thatmotivating reasons (so defined) also explain an agent’s action – although not everyonetakesthatview(e.g.Davis2005).38Forinstance:‘Therearenormativereasons:reasonsthatthereareforpeopletoact–asitisoftenput,reasonsthat‘favour’doingsomething;andmotivatingreasons:reasonsforwhichanagentacts,thatis,thereasonsthatanagenttakes(perhapsrightly)tofavouractingasshedoesandforwhichsheacts.’(Alvarez2016b,4–5emphasisadded)Similarly: ‘Normative reasonsareconsiderationswhichcount infavourof,oragainst,anaction.Whatyououghttodoisdeterminedbyhowthenormativereasonsforandagainstactingweighup–roughly,yououghttodowhatthebalanceofsuchreasonssupports.Forinstance,ifthereisareasonforyoutotakeanumbrella[i.e.anormativereasontotakeanumbrella],andnostrongerreasonnottodoso,yououghttotakeanumbrella.Normativereasonscontrastwithmotivatingreasons–thereasonsforwhichyouact.Insomecases,thereasonsforwhichyouactare,orcorrespond to, reasons for acting. That is to say, in some cases, your motivating reasons are, orcorrespond to, normative reasons. For instance, that it is rainingmight speak in favour of taking anumbrellaandbethereasonforwhichyoudoso.’(WayandWhiting2017,2–3emphasisadded)
41
Inparticular,aswellasprecludingthepossibilityofdistinguishingbetweenthekindsofreason
pickedoutbytheexpressions‘thereasonstherearetoact’, ‘thereasonsforacting’and‘the
reasons one has to act’ (the issues with which I have already discussed at length), these
‘definitions’gosomewaytosettingoutwhatthereason-relationforeachreasonexpressionis,
whichistosaythattheygosomewaytosettingoutatheoryofreasons.That is,behindthe
standardschema’sseeminglyinnocuousdefinitionsof‘normative’and‘motivating’reasonsis
the‘assumedtheory’ofreasonssetoutinTableI-7.
Theproblemisthatifwestartourtheorisingwiththe‘assumedtheory’inmindthenallthatis
really left todebate is how to interpret the relationof ‘counting in favourof’,which, aswe
shall see, gives anti-psychological theories a significant advantage. And it is precisely this
advantagethatmakesmesuggestthatthisactofseemingterminologicalhousekeepingought
tobeseenasaTrojanhorse,asthenextsectiondiscusses.
Reasonexpression Theassumedtheory
Foranyp,pisareasonforAtoφ… …ifandonlyifpcountsinfavourofφing.
Foranyp,pisareasonforA’sφing… …ifandonlyifpcountsinfavourofφing.
Foranyp,pisareasonAhastoφ… …ifandonlyifpcountsinfavourofφing.
Foranyp,pisaA’sreasonforφing… …onlyifAtookptocountinfavourφing.39
TableI-7:The‘assumedtheory’ofreasons
4.2 WhyitisaTrojanhorse
Whenit israiningandIbelievethat it israining,what is it thatcounts infavouroftakingan
umbrella– is itthefactthat it israiningorthefactthat Ibelievethat it israining?Themost
naturalinterpretationseemstobethat,inordinarycircumstances,itisthefactthatitisraining
thatcountsinfavouroftakingmyumbrella,andnotthefactthatIbelievethatitisraining.
A classic40sort of example illustrates this well: suppose that Sam believes that the security
servicesare trying to readhermind. If itwere true that the security serviceswere trying to
readhermind,thenthefactthattheyweretryingtoreadhermindwouldcountinfavourof
herwearingafoilhat.41However,thefactthatshebelievesthatthesecurityservicesaretrying
39It’sworthnoting thatbecause thisdoesn’tprovidesufficientconditions forbeinganagent’s reasonforacting, theassumedtheory isnotacomplete theoryof reasons– there is,as Iwill laternote,stillsomeroomfordisagreement,albeitmodest.40Alvarez (2016b)attributesthisexampletoAnscombe(1957);Hyman(1999);Raz (1999b)andDancy(2000).41Because, inter alia, foil hats block the radiofrequency electromagnetic radiation that the securityservicesusetoreadminds(orwould,iftheydid).
42
toreadherminddoesnotfavourwearingafoilhat–itfavoursgoingtoseeadoctor.42What
thisexampleshowsisthatthecircumstancesinwhichapsychologicalfact‘countsinfavour’of
anactioninthesamewayasthe(non-psychological)factthatitisraining‘countsinfavour’of
takinganumbrella areunusual – it is typically featuresof theworld that count in favourof
doingthings,not featuresofourpsychology.That is, ifawell-informedobserverwereto list
thethingsthatcountinfavourorcountagainstsomeaction,theywouldrarelylistfactsabout
theagent’sbeliefsasthingsthatcountinfavour(oragainst)theiraction(howevermuchthose
factsarelikelytoaffectwhattheagentactuallydoes).
We can press the point about the interpretation of the ‘counting in favour of’ relation by
consideringwhatitisthatanagenttakestocountinfavouroftheiractions.43Seemingly,what
Samtakestocountinfavourofwearingafoilhatisthatthesecurityservicesaretryingtoread
hermind(andthis issoeveniftheyaren’t), incontrast,whatshetakestocountinfavourof
going to see the doctor is that she believes that the security services are trying to read her
mind.Samisdeliberatingbothaboutthewayshetakestheworldtobeandthefactthatshe
takes it to be that way. Again, what the story about Sam’s deliberation shows is the
unusualnessofdeliberatingaboutfeaturesofone’sownpsychology–wedon’tnormallytake
featuresofourpsychologytofavouragivencourseofaction–itisnormallyonlythecontents
ofourmentalstatesthatfeatureinourdeliberation.
This line of reasoning leads us rather quickly from ‘the assumed theory’ to the following
conclusions:firstly,sincethethingsthatcountinfavourofactions(andaretakentocountin
favourofactions)areusuallynotpsychological,alltheclaimsinthepsychologismfamilymust
befalse(sincetheyonlyallowforpsychologicalreasons).Sowhateverourtheoryofreasonsis,
itmustbeanti-psychologicalwithrespecttoeveryreasonexpression.
Secondly,sincesomethingonlycountsinfavourofanactioniftheactionis,insomerespect,
worthdoing(thefactthat it’srainingdoesn’tcount in favouroftakingmyumbrella if taking
myumbrella is, tonoextent,worthdoing (if I actuallywant togetwet, say)),andsince it is
seeminglyworthdoinginvirtueofthatwhichcountsinfavourofit,thenaturalinterpretation
ofwhatitisto‘countinfavourof’anactionseemstobethis:
42Because,inSam’scase,herbeliefthatthesecurityservicesaretryingtoreadhermindistheproductofadelusionaldisorder.43Notingthat(which,asTableI-7makesclear)beingsomethingthattheagenttakestocountinfavourofanactionisanecessaryconditiononbeingthereasonforwhichanagentdoesthataction.
43
- For any p, p counts in favour of A’s φing if and only if pmakes A’s φing, in somerespect,worthdoing.44
By now we should smell a rat; substituting the above, which I have suggested is the most
naturalinterpretationofthe‘countinginfavourof’relation,intoTableI-7,yieldsthefollowing
theory:
Reasonexpression Theassumedtheory–conventionallyinterpreted
Foranyp,pisareasonforAtoφ… …ifandonlyifpmakesA’sφing,insomerespect,worthdoing.
Foranyp,pisareasonforA’sφing… …ifandonlyifpmakesA’sφing,insomerespect,worthdoing.
Foranyp,pisareasonAhastoφ… …ifandonlyifpmakesA’sφing,insomerespect,worthdoing.
Foranyp,pisA’sreasonforφing… …onlyifAtookptomakeA’sφing,insomerespect,worthdoing.
TableI-8:Theconventionalinterpretationoftheassumedtheory
ThetheorysetoutinTableI-8shouldlookfamiliar;theclaimsofthere-interpretedassumed
theory for the first three reason expressions just are the claims of favourism about those
reasonexpressions.
So,myconcernisthis:theterminologyof‘normative’and‘motivating’reasonsamountstoa
Trojanhorsebecauseonceyouaccept it, themostnatural interpretationof the ‘counting in
favourof’relationthenfixesyourtheoryforthefirstthreereasonexpressions(thatis,itforces
you to accept favourism about those reason expressions). Having accepted the assumed
theoryallthatislefttodebateisthenwhethertoendorsefavourismordeliberativismabout
thereasonsforwhichweact(whicharebothconsistentwiththefinalrowofTableI-845).That
is,whatmakes the terminologyaTrojanhorse is that it forcesmanytheorists46toacceptan
account of the reason-relation for each reason expression that puts them at a systematic
disadvantage – the only way to advocate their theory is for them to adopt an unnatural
interpretationofwhatitisto‘countinfavourof’anaction.
4.3 TheTrojanhorserejected
TheTrojansoughttohaveleftthatdamnedhorsealone,andsotooshouldwe,ifwearetodo
our theorising inearnest.That is, rather thanaccommodatingdifferent theoriesbyadopting
44I say this without meaning to undermine the view that the relation of ‘counting in favour of’ issomehowprimitive. It is, Ithink,stillpossiblethatthefavouringrelationisthemorefundamentaloneevenifthisisthecase–see§(II)3.1forfurtherdiscussionofthispoint.45As fn.39records–thefinalrowomitsasufficientconditionforbeingareasonforwhichtheagentacts,sotheassumedtheoryfallsshortofafulltheoryofreasons.46Thatis,theoriesofreasonsotherthan(F,F,F,F)or(F,F,F,D).
44
anunnaturalinterpretationofwhatitisto‘countinfavourofanaction’,47thebestresponseis
torejectthehorsebyrefusingtoaccepttheassumedtheoryasthestartingpointoftheorising.
Andonceonedoesthat,theoft-repeatedargumentagainstpsychologistictheories48becomes
question-begging:itstartsoffbyassumingthetruthofaposition49thatpsychologistictheories
shouldreject.50
4.4 IsitaTrojanhorseorisitjustthetruth?
ButisitreallyaTrojanhorse?Isn’tcountinginfavourofactingjustwhatitistobeareason?
Thisis,afterall,awell-establishedview,broughttotheforeinthefollowingremarks:
Iwilltaketheideaofareasonasprimitive.Anyattempttoexplainwhatitistobeareasonforsomethingseemstometoleadbacktothesameidea:aconsiderationthatcountsinfavorofit. ‘Counts in favor how?’ onemight ask. ‘By providing a reason for it’ seems to be theonlyanswer.(Scanlon1998,17)
Ifweareaskedwhatreasonsare,itishardtogiveahelpfulanswer.Factsgiveusreasons,wemight say,when they count in favourof ourhaving somebelief ordesire, or acting in someway.But‘countsinfavourof’means‘isareasonfor’.Likesomeotherfundamentalconcepts,suchas thoseof reality,necessity, and time, the conceptof a reasoncannotbeexplained inotherterms.(Parfit2001,18)
The suggestion is that ‘being a reason for’ and ‘counting in favour of’ are interchangeable
expressions.Now,werethatviewimmediatelyobviousthenwhatIhavecalleda‘Trojanhorse’
would be less of a covert assault on psychologistic theories andmore just the inconvenient
truth for them. I don’t, however, think that this view is immediately obvious, particularly
becauseIthinkitisactuallyfalse.
WhileIagreewithScanlonandParfitthatwheneversomethingisareasonforactingitmust
countinfavourofsoacting,51IdisagreewiththeminsofarasIthinkthatitcountsinfavour
soactingasaconsequenceofitsbeingareasonforacting,andnotaconsequenceofitsbeing
areason.Inparticular,Idonotthinkitfollowsthatanythingthatcountsinfavourofanaction
isareasonfordoingit,northatallreasonscountinfavourofactions.52WhatIwanttoargue
47AccordingtowhichthefactthatSambelievesthatthesecurityservicesaretryingtoreadhermindcountsinfavourofwearingatinfoilhat.IthinkthatMitova(2016)prefersthatoption.48BywhichImeantheargumentsetoutin§4.2,whichonecouldputsuccinctlyasfollows:whatcountsinfavourofactionsisoftennotpsychological(asshownbySam’scase)andthatwhichcountsinfavourofanactionisareasontodoit,soreasonscanbethingsotherthanfeaturesofanagent’spsychology,sopsychologism(whichsaysthattheycan’t)isfalse.49I.e.thatsomethingisareasonjustincaseitcountsinfavourofanaction.50Thisis,Ithink,preciselyGibbons’s(2010,354)objectiontotheuseofthislineofreasoningagainsthis‘psychologistic’theory.51Andthatwheneversomethingisareasonagainstactingitmustcountagainstdoingit.52Thatis,whileIagreethatifp isareasonforφingthenpcountsinfavourofφing.Idon’tagreethattheright-to-leftreading(ifpcountsinfavourofφingthenpisareasonforφing)istrue,nordoIagreethatifpisareasonforAtoφthenpcountsinfavourofφing.
45
for now, however, is not the view that ‘being a reason’ and ‘counting in favour of’ aren’t
interchangeableexpressions–butjustthattheviewthattheyareisnotimmediatelyobvious.
Myargument is this: it seemsmore likely that themeaningof ‘counting in favourof’ comes
fromthefactthatsomethingisareasonforactingthanthatitisareasonforacting.
TheOxfordEnglishDictionarydefinesthe‘for’preposition(interalia)asfollows:
‘In defenceor support of; in favour of, on the sideof.Opposed to against.’ (‘For, Prep. andConj.’,n.d.,7aemphasisadded)
Seeminglyanythingthatisforsomething,inthissenseofthepreposition,countsinfavourof
it, invirtueof itsbeing for it (andnotnecessarilyanythingelseabout it).So, I suggest,what
makesa‘reasonforacting’countinfavourofanactionisthefactthatitisforthataction(as
opposedtoagainstit),andnotthefactthatitisareason.Thatis,myclaimisthatweshould
not necessarily analyse ‘being a reason’ as ‘counting in favour of’, ratherwe have stronger
reasonsforanalysing‘beingforsomething’as‘countinginfavourofit’.
This is made clearer by the fact that the ‘for’ preposition lends the meaning of ‘counts in
favour of’ to things other than reasons. For instance, I can say: ‘the factors for and against
acting’; ‘the things for and against acting’; ‘the considerations for and against acting’; ‘what
there is to be said for and against acting’.Wherever this ‘for’ appears,we can say that the
thingthatprecedes itcounts in favourof thatwhich it is for,but thatdoesn’t leadustosay
that what it is to be a ‘factor’, or a ‘thing’, or a ‘consideration’, or ‘something to be said’,
dependsoncountinginfavourofsomeactionindependentofitsbeingafactorfor,orathing
for,oraconsiderationfor,orsomethingtobesaidfordoinganaction.Likewise,Isuggest,the
factthatbeingareasonfordoingsomethingdependsoncountinginfavourofdoingit,should
notleadustoassumethatbeingareasondependsoncountinginfavourofdoingsomething–
insteadweshouldjustadmitthatwheneveranythingisforsomething,itcountsinfavourofit,
andjustbecauseitisforit.53
Tobeclear:myaimwiththeseremarksisnottoshowthatsomereasonsdon’tcountinfavour
ofacting.Myaimisjusttoshowthatitisnotaforegoneconclusionthatallreasonscountin
favour of actions, and it is certainly not something thatwe should assume at the outset of
theorising.One’stheorymightenduparguingthatthereason-relationsarebestunderstoodin
termsof the ‘counting in favourof’ relation (i.e. as favouristsdo)–but that is anargument
thatmustbemade,itisnotthedefaultposition.Thatis,weshouldnotstartoffourtheorising53Whatabouttheagent’sreasonforacting?Doesn’tthatthencountinfavouroftheiraction,sinceitisfor it? Idonot think itdoes, I suggest that thisuseof ‘for’appeals tooneof theprepositions (many)othermeanings;specifically‘ofpurposeordestination.’(‘For,Prep.andConj.’,n.d.,IV)Theassociationofthisexpressionwiththismeaningisdiscussedinmoredetailin§(II)1.2.
46
byassumingeitherthatsomethingisareasontoact,areasonforactingorareasononehasto
actjustincaseitcountsinfavourofacting,orthatareasonforwhichanagentactsisalways
somethingtheytaketocountinfavouroftheiraction.Tostartourtheorisinginthismanneris,
Isubmit,prejudicialtoourenquiry.Andthis,accordingly,iswhyIhaverejectedthestandard
schemaandwhyIeschewtheterminologyof‘normative’and‘motivating’reasons.
5 Conclusion
I have argued that the standard schema is ill-suited to the task of categorising different
theories of reasons and I have proposed an alternative categorisation. My alternative
categorisation distinguishes theories according towhat they say the reason-relation is for a
number of typical reason expressions and it eschews the language of ‘normative’ and
‘motivating’ reasons because those terms (as they are standardly defined) already involve
substantivetheoreticalcommitments.
In thenext three chapters I discuss themain critiquesofwhat are typically taken tobe the
strongest suits of each of the three families, namely (and I discuss them in this order):
favourism about reasons to act; psychologism about the reasons for which we act; and
deliberativism about the reasons for which we act. I set out a number of prima facie
reasonableclaimsandshowhoweachoftheseviewsmustrejectsomesubsetoftheseclaims.
47
(II)
Reasonstoactthatmakeactionsworthdoing
InwhichIshowwhatitcoststothinkthatifthereisareasontodosomeactionthenthatactionis,insomerespect,worthdoing.Ishowhow‘favourismaboutreasons to act’ (which entails that reasons to actmake actionsworth doing)clashes with some prima facie reasonable claims about a case in whichsomeoneactsonafalsebelief.IsetoutwhichclaimstheproponentofthisviewmustchoosebetweenrejectingandIcategorisethecommonchoicesfromtheliterature.
Whilewalkingthroughaforestthatsheknowstocontainbears,Sallyhearswhatsoundslikea
bearrunningtowardsher.1Sheruns, frantically, toanearbysafe-house. In fact,nobearwas
chasingher;itwasanoddrustlingofthetreesthatmadethenoise.DidSallyhaveareasonto
run?Ofcourse–shethoughtabearwaschasingher!Didsherunforareason?Surely!What
washerreasonforrunning?Shethoughtthatabearwaschasingher,orperhapsonemight
saythis:sheheardabear-likesound.
Thisseeminglystraightforwardstoryandthesetypicalintuitionsaboutit,createproblemsfor
whatIhavecalled‘favourismaboutreasonstoact’,accordingtowhichareasonforsomeone
todosomeactionisafactthatmakesit,insomerespect,worthdoing.
The purpose of this chapter is to set out a series of prima facie reasonable claims about
reasons2andtothenshowthat‘favourismaboutreasonstoact’mustrejectatleastsomeof
them.This isnotmeanttobeaconclusiveargumentagainstthisview–onlytoshowthat it
comesatthecostofrejectingsomeprimafaciereasonableclaims.
1 Sallyandthenon-existentbear
1.1 Reasonsandrationality
Given that she knew that safety was nearby, and given that she thought that a bear was
chasingher (addthatshewantsto live, ifyou like), itwasplainlyrational forSallytorun,as
Stoutnotes:
Whatcanbeacceptedwithoutmuchdifficultyisthatherhavingthatbeliefmakesherrunningawayrationally intelligible.Learningthatshethinksabear ischasingher Icanmakesenseofherrunningaway;Icanseethatherbehaviourisrational.(Stout2009,52)
1ThisexampleisadaptedfromStout(2009).2BasedbothonthisstoryaboutSallyandonmoregeneralintuitionsabouthow‘thereasonsthereareforanagenttoact’relateto‘thereasonstheyhavetoact’and‘thereasonsforwhichtheyact’.
48
SinceitwasplainlyrationalforSallytorun,let’ssayitplainly:
(F1) ItwasrationalforSallytorun.3
Now, IsaidthatSallyhadareasontorun,eventhoughnobearwaschasingher.Thisseems
likeanaturalthingtosay,butwhy?Ithinkthatitisbecause,asErrolLordobserves:
It’snaturaltothinkthatwheneverit’srationalformetoφ,Ihavereasonstoφ.(Lord2010,1)4
It’snaturaltosaythatshehasareasontorun,anditsoundsstrangetosaythatshedoesn’t
becausethefactthatitwasrationalforhertodoitsuggeststhatshehadareasontodoit.So,
hereisanotherprimafaciereasonableclaim:
(F2) IfitisrationalforAtoφthensomepwasareasonAhadtoφ.
1.2 Actingforareasonandactingintentionally
IsaidthatSallyranforareason.TodenythisandsaythatSallydidn’trunforareason,thatis,
thatshewasrunningfornoreason,seemstosuggestthatheractionwasunconsidered,that,
atbest,shewasrunningforthesakeofrunning,orrunningonawhim. Ifweweretosayto
someone that Sally ran for no reason, and then add that, by theway, she ran because she
thought a bearwas chasingher, itwould seem to cancel the sense inwhich she ran for no
reason:sheiscertainlynotrunningforthesakeofrunningifsheisrunningbecauseshethinks
thatabearischasingher.
Thepointisnotthatthereisnowayofmakingtheclaimthatsheranfornoreasonintelligible
–it’sjustthatsayingthatsheranfornoreasonandthatsheranbecauseshethoughtabear
was chasing her has an air of contradiction about it. And the reason it has an air of
contradiction about it is that saying that someone didn’t act for a reason implies that their
actionlackedsomesortofrational,intentional,deliberateness–itmakesitseemwhimsicalor
unconsidered,ifnotentirelyunintentional.AsDancynotes:
Intentional,deliberate,purposefulactionisalwaysdoneforareason.(Dancy2000,1)
Why does Dancy choose the particular string of adjectives (‘intentional, deliberate,
purposeful’)hedoes?Ithinkheistryingtoanticipateanobjectiontothesimplerclaimthatall
intentionalaction isdoneforareason. Inparticular:onemightsaythat if Icrossmy legs(to
useDancy’sexample),IactintentionallybutIdon’tactforareason–thesortofactsonedoes3Youcould,ofcourse,addindefeaters(shethoughtthebearwasbetweenherandthesafe-house;shethinksthatstayingstillisthebestwaytoavoidabearattack),butthatwouldbetochangemystory–therearenosuchdefeatershere;themostrationalthingforSallytodoisrun.4For similar remarks on the ‘naturalness’ of this claim, fromphilosopherswith different views aboutreasons seeUnger (1978, 200), Alvarez (2010, 13), Gibbons (2010, 337) and Comesaña andMcGrath(2014,61).
49
on a whimmay be done for no reason, but they aren’t unintentional. Dancy, quite rightly,
doesn’twanttoustothinkofsuchactions(whichO’Shaughnessy(1980)calls‘sub-intentional’
actions), as actions that are done intentionally, purposefully anddeliberately.5It seems that
thereisaclearclassofwhatwemightcallsophisticatedlydeliberateactionsthatweassociate
withactingforareason.6Forthesakeofbrevity IwillshortenDancy’sstringofadjectivesto
justdeliberateactionandformulatethisprimafaciereasonableclaimasfollows:
(F3) IfAφsdeliberatelythenAφsforareason.
This view is shared by philosophers with a wide range of views onwhat reasons are,7and,
indeed,Meleremarksthat ‘theoverwhelmingmajorityofordinaryspeakersofEnglishasked
foragutreactionto[thisclaim]wouldfinditextremelyplausible.’(Mele2007,99)
DidSallyrundeliberately(intentionally,purposefully…)?Ofcourseshedid!Sherespondedto
the situation as she took it to be, she decided on the appropriate course of action and she
actedonthedecisionshe’dmade(albeithurriedly!).8Thuswecansay:
(F4) Sallyrandeliberately.
SothisiswhyitsoundsoddtosaytheSallydidn’trunforareason:itsuggeststhatshedidn’t
rundeliberately(intentionally,purposefully…)–whichsheplainlydid.
1.3 Sally’sreasonsforrunning
TheremarksImadeaboutSally’sreasonsforrunningbear(!)repeating,sinceItakethemto
beprimafaciereasonable.Firstly,Isaidthat:
(F5) Sally’sreasonforrunningwas,interalia,thatshebelievedabearwaschasingher.
5Mele (2007, 99) considers another would-be counterexample to claim that all intentional action isdoneforareason,whichDancy’swordingisseeminglyalsodesignedtoavoid.6It’sworthnoting that the senseof sophisticationhere is purely internal to the agent –whatmakessomethingcountas,inDancy’sterminology,anintentional,deliberateandpurposefulactionisallaboutthewaytheagentreasonedtheirwaytoit,andnothingtodowithitscorrespondencetotheexternalworld.7For instance, it’s worth noting that Maria Alvarez, who rejects (F3) nonetheless notes that ‘Thesuggestionthatsomeonewhoactedmotivatedbyafalsebeliefdoesnotact forareasonmightseemprimafaciewrong.’(2010,141)ThatisallthatIamsaying–thatitisprimafaciereasonabletosaythattheydoactforareason.(SeealsoAnscombe1957,9;Davidson2001c,83;Davis2005,68–69;Gibbons2010,357;Hieronymi2011,410–11)8Sally’s circumstancesareextremebut I don’t think it’sunreasonable to suggest that shedeliberatesaboutwhattodo.Shemightbetorn,forinstance,overwhetheritwouldbebetterto‘playdead’–cansheoutrunthebear?Onemightwell, Isubmit,deliberateevenwhenthestakesarehighandtime isshort – such deliberation need not take a long time, or even havemuch to it (themattermight bestraightforward)forone’sactiontobedeliberate.
50
It flies in the faceofexperience todeny thatwemakesuchremarks,as, indeed, JosephRaz
notes:
There isnodenyingthatweuse locutionsoftheform‘hisreasonforφingwashisbeliefthatp’…(asin‘hisreasonfornotcomingwasthathethoughtyouwouldnotbehere’).(Raz1999b,18)
Ialsosaidthat:
(F6) Sally’sreasonforrunningwas,interalia,thatsheheardabear-likesound.
Thisseemstosuggestthatanappearanceorperceptualexperiencecouldbeanagent’sreason
for acting. I find it as natural to say this as saying that her reason for runningwas that she
believedthatabearwaschasingher,but it isperhapsworthnotingthatreasonsofthiskind
are less commonly discussed in the literature.9Nonetheless, I take this claim to be, equally,
primafaciereasonable.
2 Howreasonexpressionsrelate
In §(I)1, I discussed different reason expressions, but made no mention of the relations
betweenthem.Sohowaretheyrelated?HereiswhatGibbonssuggests:
Ifyougotothestoreformilk…youwillgothereforareason.Soyoumusthaveareason.Sotheremustbeareason.(Gibbons2010,343)
Gibbons’inferenceseemsnatural.Thatis,itseemsprimafaciereasonabletothinkthatifone
actsforareasononemusthavehadareasontoactandifonehadareasontoactthenthere
musthavebeenareasontoact,whichonehad.So:
(F7) IfAφsforareasonthentherewasareason,p,forAtoφ.
Denying thismeans, as Jennifer Hornsby notes (albeit, in the process of denying it), saying
thingslike:‘therewasnoreasontodowhathedid,eventhoughhediditforareason,’(2008,
249) which is, in her own words, ‘prima facie paradoxical.’ The denial of a prima facie
paradoxicalclaimis,itseemstome,atleastprimafaciereasonable.
IthinkthatGibbons’inferencepermitsofsomestrengthening:itisnotjustthatyougotothe
storeforareasonandyouhaveareasonandthereisareason,butthatyourreasonforgoing
there isareasonthatyouhavetogothereand isareasonthatthere is foryoutogothere.
9AlthoughDancyalludestothissortofreasonforactingwhenhenotesthatwhenEdmundstaysawayfromthemiddleofanicylakeafterhisfriendhadtoldhimthattheicewasthinthere,‘hisreasoncouldhavebeensimplythathisfriendhadwarnedhimoff.’(Dancy2014,88)KearnsandStar(2008)defend(at length)anaccountthatagreeswiththis intuitionandWhiting(2014)alsonotesthatweoftenciteperceptualexperiencesasreasons.Lord (2010,6) isalsoclear thatareasontobelievesomething isareasontodothatwhichthebeliefmakesrational–assumingthatexperiencescanbereasonsforbelief,hewouldseeminglyagreewith(F6).
51
These strengthened claims provide us with the following, additional prima facie reasonable
claimsaboutreasons:
(F8) Foranyp,ifpwasareasonAhadtoφthenpwasareasonforAtoφ.10
(F9) Foranyp,ifAφsforthereasonthatpthenpwasareasonforAtoφ.11
Bothof these claimsareat leastprima facie reasonablebecausedenyingeitheramounts to
saying, implausibly,thatsomethingthatisareasonisn’tareason:denying(F8)meanssaying
thatanagenthasareasontoactthatisn’tareasontoact;anddenying(F9)meanssaying,as
Dancynotes,‘anagentcanactforareasonthatisnoreason.’(2000,3)
Indeed,Melefindsitsostrangetosaysuchthingsthathemakesitanexplicitconstraintona
satisfactorytheoryofreasonsthatitconcedethat:‘anythingthatisareasonisnotnoreason.’
(2007,95) Iamnotgoingas farasMele– Inotehis remarksonlytoshowthattheseclaims
enjoyconsiderableintuitiveplausibility,sotheyareatleastprimafaciereasonable,andthatif
atheoryrejectsthem,itisatsomecostthatitdoesso.
3 Favourismaboutreasonstoact
RecallthefollowingclaimaboutreasonsfromTableI-4:
Favourism about reasons to act: For anyp, p is a reason forA toφ if and only ifpmakesA’sφing,insomerespect,worthdoing.
ThisisnotthestandardwayofcharacterisingtheviewofthosetowhomIattributeit.12These
theoristsmoretypicallycharacterisetheirviewastheclaimthatareasontodosomeactionis
thatwhich‘countsinfavour’ofdoingit.
However,IcharacteriseitinthewaythatIdobecausetherearedifferencesofopinionabout
what it is to ‘count in favour’ofanaction,13sotheproblemsfor favourismaboutreasonsto
act(tobediscussed in§4)donotapplytoeveryonewhoholdstheviewthatreasonstoact
count in favour of actions. My aim in characterising the view in this way is thus to group
togetherthosetheorists(ofwhichtherearemany)whoaresusceptibletotheproblemsthat
10Alvarez (2016b, 11) provides a thorough defence of the intuitiveness of this claim. It is alsoworthnotingthatImakenoclaimaboutthetruthorfalsityoftheright-to-leftreadinghere.11Cf. Dancy’s (2000) normative constraint, which is aweaker version of this thesis. It says that if anagentactsforthereasonthatpthenpcouldhavebeenareasontherewasforthemtoact.12SeeTableI-6forthelistofthosetowhomIattributethisview.13For instance, Kearns and Star suggest that ‘a fact counts in favor ofφing just in case this fact isevidence that one ought to φ.’ (2008, 44) And Mitova (2016) also rejects the conventionalinterpretation,althoughsheleavestheworkofdevelopinganalternativeforanothertime.IalsothinkLord (2010) requires another interpretation of the ‘counting in favour of’ relation if he is to get hisdesiredresultthatthatwhichjustifiesabeliefalsocounts infavourofanactionthatthebeliefmakesrational.
52
follow,acknowledgingthattherearesubtlevariationsbetweentheirtheories(moreonthisto
come).
InwhatfollowsIprovidesomecontextforhowIarriveatthecharacterisationthatIdo,aswell
assomeexplicationofwhatitisforanactiontobe,insomerespect,worthdoing.
3.1 ‘Countinginfavourof’
AsGjelsvik(2007)notes,therearedifferingwaysofconceivingofwhatitisto‘countinfavour
of’anactionthatarriveatseeminglythesameconclusionaboutwhenanactionisfavoured.
Tounderstandthefirstway,letusrevisitthefollowingremarks:
Iwilltaketheideaofareasonasprimitive.Anyattempttoexplainwhatitistobeareasonforsomethingseemstometoleadbacktothesameidea:aconsiderationthatcountsinfavorofit. ‘Counts in favor how?’ onemight ask. ‘By providing a reason for it’ seems to be theonlyanswer.(Scanlon1998,17)
Ifweareaskedwhatreasonsare,itishardtogiveahelpfulanswer.Factsgiveusreasons,wemight say,when they count in favourof ourhaving somebelief ordesire, or acting in someway.But‘countsinfavourof’means‘isareasonfor’.Likesomeotherfundamentalconcepts,suchas thoseof reality,necessity, and time, the conceptof a reasoncannotbeexplained inotherterms.(Parfit2001,18)
I firstdiscussedtheseremarks in§(I)4.4,whenIdrewattentiontothefactthatScanlonand
Parfit take the ‘counting in favour of’ relation and the ‘being a reason to’ relation to be
equivalent. This is something they have in common with, I think, everyone who endorses
favourism about reasons to act. But another observation that we can make about these
remarks, andonewhich, aswe shall see, separates Scanlon andParfit from someothers to
whom I attribute favourism about reasons to act, is that the former take the ‘counting in
favourof’relationtobeprimitive.Thisconstrual,(‘favouring’asaprimitiverelation)isthefirst
wayofconceivingofwhatitistocountinfavourofanaction.
ThemainalternativeisgivenitsclearestexpositionbyJohnBroome,whostatesthat‘tocount
infavourofφistoplayaparticularroleinanexplanationofwhyyououghttoφ’(2006,41).
JohnHyman(e.g.2015,133–34)alsoseemstobeclearabouttheexplanatorycharacterofthe
reason-relation,andwhilehelessdirectlyanalysesthe‘countinginfavourof’relationinthose
terms,Ithinkhispositionamountstothesame.
Others are less explicit, but, I think, more or less implicitly analyse favouring as such an
explanatoryrelation.Forinstance,someauthorscharacterise‘countinginfavourof’asmaking
good,orrightorvaluable:
Areasonforactionissomethingthatfavoursormakesvaluableanactionoftherelevantkind.(Everson2009,22emphasisadded)
53
Ifapersonhasareasonforφingthenitfollowsthatthatpersonoughttoormay(ifonlyprotanto)φ.Thismightmeanthatthereasonfavoursφing,orrecommends,permits,warrants,ordemands, etc. φing. But the question arises why a reason for φing favours, warrants, ordemandsφing? I suggested that theanswer to thisquestion is thata reason forφingmakesφingrightorappropriate(sometimesmerelyprotantorightorappropriate).(Alvarez2010,12–13emphasisadded)
Whenwe think of such reasons, we think of features that speak in favour of the action (oragainst it)… theymake actions right or wrong, sensible or unwise. (Dancy 2000, 1 emphasisadded)
Ithinkthatitisnaturaltointerpretthe‘making’relationsinvokedhere(reasonsmakeactions
valuable, right, sensible…) as explanatory relations. I find it odd to think that some x could
makesomeyanFwithout(atleastpartly)explainingwhyitisanF,soIlikewisefinditoddto
think that some reason could make some action right or valuable without (at least partly)
explaining why it was right or valuable. Thus, I take accounts that say that reasons make
actions right, valuable or good to be, albeit implicitly, of the same explanatory view as
Broome.14Iftheyaren’tthenthereisperhapsathirdwayofarrivingatwhatisultimatelythe
sameconclusion.
Whatisthatconclusion?Ithinkwecanputit,15asIdidinhopefullyun-contentiously,thus:
- For any p, p counts in favour ofA’sφing if and only ifA’s φing is, in some respect,worthdoing.
Given the presumed equivalence of ‘being a reason to’ and ‘counting in favour of’ amongst
advocatesof favourismaboutreasonstoact,wethusarriveat theview,setoutabove, that
there isareasonforanagenttoφ if,andonly if,φing is, insomerespect, ‘worthdoing’ for
thatagent.Butwhatdoesitmeantobeforanacttobe,insomerespect,worthdoing?
3.2 Whatitisforanacttobe,insomerespect,worthdoing
I say ‘in some respect’worthdoing,because the fact that there is something that counts in
favour of doing an action does not mean that it is, all things considered worth doing. For
instance, although something counts in favour of having a doughnut every day (they taste
great!),sothatitis,insomerespect,worthdoing,itisprobablynotallthingsconsideredworth
doing(thecalories!).Insuchcaseswecansaythattheactionis,insomerespect,worthdoing,
butnotallthingsconsideredworthdoing.
14To the extent that ‘in virtue of’ relation is an explanatory relation, Raz (perhaps against his ownjudgment) is socategorisable, cf. ‘reasonsare facts invirtueofwhich thoseactionsaregood in somerespect.’(Raz1999a,22emphasisadded)15AsIdidin§(I)4.2.
54
When is an act, in some respect, worth doing? It might be when it results in (or just is)
somethingthatis,insomerespect,good,right,valuable,ormerelydesired.16Forthepurposes
ofthisdiscussionwedon’tneedtodecidewhichofthoseitis–allweneednoteisthatanact
is,insomerespect,worthdoingif,asamatteroffact,itwouldresultinsomethingthatis,in
somerespect,good,right,valuable,desiredorwhathaveyou.
An important point to take from the last remark is this: whether or not an action is worth
doing(insomerespect,orallthingsconsidered)forsomeagentisanobjectivematteroffact
that is, inparticular, independentof theirperspective. Soanaction is stillworthdoing if an
agent is ignorantof the fact that it’sworthdoing. For instance: it’sworth taking a different
routehomeifthere’strafficonone’susualroute–andthat’sworthdoingevenifonethinks
one’s usual route is all clear (that it’s worth doing doesn’tmean that onewill do it, it just
meansthatifonedidit,some‘good’17wouldcomeofit).18Andanactionisn’tworthdoingjust
becauseyouthinkitis:youmightthrowawaysomemilkbecauseyouassume,falsely,thatit’s
goneoff–throwingawaythemilkisnotworthdoing,althoughyoumaythinkthatitis.Soit’s
theway theworld is thatmatters to whether or not some action is worth doing for some
agent,19notthewayshetakesittobe.
3.3 Favourismaboutreasonstoact
SofarI’vesaidthat,accordingtothisview,thereisareasontoactifandonlyifanactionis,in
somerespect,worthdoing,and I’veexplainedwhat it is foranactiontobeworthdoing.All
thoseauthors towhomIattribute favourismaboutreasonstoactwould, I think,agreewith
thoseremarks.
Wheremy characterisation of favourism about reasons to act strays into the contentious is
that it goesbeyond these remarks to claim that reasonsareactuallywhatmakes theaction
worthdoing.Thismaybeunpalatabletothosewhotakethe‘countsinfavourof’relationtobe
primitive. For those that do balk at it I am happy for them to shrug it off (along with my
16 Alvarez gives what she describes as a ‘rough’ characterisation, which, to my mind, is a neatrepresentationofmanyofthedifferentviews.Shenotesthatfortheretobeafactthatfavoursanagentdoing someaction that ‘requires theirhaving somemotivation thatwouldbe servedbyacting in theway favoured…Themotivationmaybesuch thingsasdesires,plans, long-standingprojectsorvalues.Anditmaybesomethingtheagentactuallyhas,orsomethingshewouldhaveifshereasonedproperlyfromhercurrentmotivations.’(Alvarez2016a,10)ThisiswhatImeanbysayingthatanactis‘favoured’ifitis,insomerespect,worthdoing.17Howeverwechoosetounderstand‘good’…18Cf.Inagameofchess,agoodmoveisagoodmoveregardlessofwhetherornotyou’vespottedit.19NotethataccordingtoBroome-typeconceptionsofthe‘countinginfavourof’relation,actsareworthdoingbecauseofthewaytheworldis.
55
categorisationoftheirtheories),althoughthatmeansleavingunanalysedarelationthatsome
interpretdifferently.20
WhatreallymattersisthatallofthetheoriststowhomIattributefavourismaboutreasonsto
actarecommittedtothefollowingposition,which,asIwillshow,iswhatreallycreatestrouble
forthem:
(FAV) Foranyp,p isareasonforAtoφonly ifA’sφing, is insomerespect,worthdoing.
Thisisclearlylogicallyentailedbyfavourismaboutreasonstoact.Moreover,Ithinkallthose
towhomIattributefavourismaboutreasonstoactwhomightbalkatmycharacterisationofit
(suchasParfitandScanlon)wouldnonethelessagreewith(FAV).
Tobeclear:(FAV)isweakerthanfavourismaboutreasonstoact, inthesensethatthelatter
entailstheformer,buttheformerdoesnotentailthelatter.Mypointisjustthatthat(FAV)is
enoughtocreatetroubleforfavourismaboutreasonstoact.
4 Theproblemsforfavourism
Toseewhattheproblemsfor(FAV)are, let’sreturntoSallyandhernon-existentbear.Since
thereisn’tactuallyabearchasingher,exhypothesi,runningisn’tworthdoingforSally.There
isnothingtobegained(foranyone)fromSally’srunning. Indeed, it’spossiblethatshemight
fallandhurtherself,orattracttheattentionofactualbearsbyrunning.Sonotrunningisworth
doing,butrunningisreallytonoextentworthdoing.21Thus:
(F10) Sally’srunningwasnotatallworthdoing.
This observation, combinedwith theprima facie reasonable claims set out above, results in
three distinct problems for the favourist view: The Rational Action Problem; TheDeliberate
ActionProblem;andThePsychologicalReasonProblem(forFavourism).
4.1 TheRationalActionProblem
TheRationalActionProblemisthis:itisrationalforSallytorun,soshehasareasontorun,so
there is a reason for her to run, so her running is, in some respect, worth doing, but her
runningisinnorespectworthdoing!Explicitly,thefollowingclaimsaremutuallyinconsistent:
(F1) ItwasrationalforSallytorun.
20See§(I)fn.26fordifferentinterpretationsofthe‘countinginfavourof’relation.21Youcouldmakeastoryinwhichrunningwas,insomerespect,worthdoing,ifyouliked(Sallyneedstoloseafewpounds,say),butthat’snotmystory.InmystorySallystandstogainnothing,andpotentiallylosemuch,fromrunning(and,likewise,nothingofworthaccruestoanyoneelseifSallyruns).
56
(F2) IfitisrationalforAtoφthensomepwasareasonAhadtoφ.
(F8) Foranyp,ifpwasareasonAhadtoφthenpwasareasonforAtoφ.
(FAV) Foranyp,p isareasonforAtoφonly ifA’sφing, is insomerespect,worthdoing.
(F10) Sally’srunningwasnotatallworthdoing.
4.2 TheDeliberateActionProblem
TheDeliberateActionProblemisthis:Sallyrandeliberately,sosheranforareason,sothere
wasareasonforhertorun,soherrunningis,insomerespect,worthdoing,butherrunningis
innorespectworthdoing!22Explicitly,thefollowingclaimsaremutuallyinconsistent:
(F4) Sallyrandeliberately.
(F3) IfAφsdeliberatelythenAφsforareason.
(F7) IfAφsforareasonthentherewasareason,p,forAtoφ.
(FAV) Foranyp,p isareasonforAtoφonly ifA’sφing, is insomerespect,worthdoing.
(F10) Sally’srunningwasnotatallworthdoing.
4.3 ThePsychologicalReasonProblem(forFavourism)
ThePsychologicalReasonProblem(forFavourism)isthis:Sally’sreasonforrunningwas,inter
alia,thatshebelievedthatabearwaschasingher,sothatwasareasonforhertorun,soher
runningwas, insomerespect,worthdoing,butherrunningwastonoextentworthdoing!23
Explicitly,thefollowingclaimsaremutuallyinconsistent:
(F5) Sally’sreasonforrunningwas,interalia,thatshebelievedabearwaschasingher.24
(F9) Foranyp,ifAφsforthereasonthatpthenpwasareasonforAtoφ.
(FAV) Foranyp,p isareasonforAtoφonly ifA’sφing, is insomerespect,worthdoing.
22AvariantofthisproblemisBroome’s(2013,71)‘quickobjection’tothethesisthatactingrationallyis‘respondingcorrectlytoreasons.’23It’sperhapsworthnotingthatthisisanargumentusedbythosewhoalreadyacceptfavourismaboutreasonstoacttoshowthatSally’sreasonforrunningcan’tbethatshebelievedthatabearwaschasingher (assetout in§(I)4.2).Mypoint is thatwhenwestartour investigationofwhat reasonsare fromprima facie reasonable claims about them, without assuming favourism about reasons to act, theargumentrunsintheotherdirection–againstfavourism.24RecallthatIassumethat:foranyp,A’sreasonforφingwasthatp ifandonlyifAφsforthereasonthatp(see§(I)1.5).
57
(F10) Sally’srunningwasnotatallworthdoing.
4.4 TheExperientialReasonProblem(forFavourism)
TheExperientialReasonProblem(forFavourism) is this:Sally’s reason for runningwas, inter
alia,thatsheheardabear-likesound,sothatwasareasonforhertorun,soherrunningwas,
in some respect,worthdoing, buther runningwas tonoextentworthdoing! Explicitly, the
followingclaimsaremutuallyinconsistent:
(F6) Sally’sreasonforrunningwas,interalia,thatsheheardabear-likesound.
(F9) Foranyp,ifAφsforthereasonthatpthenpwasareasonforAtoφ.
(FAV) Foranyp,p isareasonforAtoφonly ifA’sφing, is insomerespect,worthdoing.
(F10) Sally’srunningwasnotatallworthdoing.
5 Responsestotheproblemsforfavourism
How can a proponent of favourism about reasons to act respond to these problems? The
optionsarelimited.Idon’tthinkonecouldtolerablyrejecteitherthatit’srationalforSallyto
run,orthatSallyrunsdeliberately–so(F1)and(F4)areoffthemenuofpotentialresponses.
Itmightbetemptingtothinkwecouldreject(F10)andkillthreeproblemswithonerejection.
Weshould resist thisurge– todoso isultimately just tochange thenotionof ‘beingworth
doing’inawaythatamountstonomorethanarejectionoffavourismaboutreasonstoactby
anothername.25
So,(F1),(F4)and(F10)areoffthemenu.Thisbringsmetotheconclusionofmyargument:if
youwanttopreserve(FAV)thenyouhavetorejectatleastoneoftheprimafaciereasonable
25Maybeyouwant tosay thatSally’sbelief thatabear ischasingher iswhat favoursheraction (VeliMitova(2015,2016)givesthisagoodgo).Well,wecouldseehowitcouldmakerunningworthdoing.If,forinstance,Sallyhasaparticularlyoddconstitutionsuchthatifshebelievesthatabearischasingherandshedoesn’trunthenshewillhaveaheartattackorsuffersomeotherunpleasantness.Then,inthatcase, her belief that a bear is chasing herwouldmake running (in some respect)worth doing.Why?Because,giventhatshebelievesthatabearischasingher,ifshedoesn’trunshe’llhaveaheartattack.But this isn’t the situation that Sally is in. Sally is like you or I – she hears what sounds like a bearrunning,knowssafetyisnearby,sosheruns.Perhapsyouwanttosaythatsomethingdoesfavourherrunning:thefactthattheremightbeabearnearby.Butthatfactdoesn’tmakerunningworthdoing–sincethereisn’tabearnearby.Thefactthatinclosepossibleworldsabearischasingherdoesnotmakeitworthrunninginthisactualworld.AsitstandsSallyhasnothingtogainandquiteabittolosefromrunning.Runningisnotworthdoingforher,althoughitisthemostrationalthingtodo–andgiventheconceptoffavouringthatweareworkingwith,thisjustmeansthatnothingfavoursheraction.
58
claims fromeachof the followinggroups (andpossibly two fromgroup3– sinceoneof the
disjunctsisaconjunction),andacceptthecounterintuitiveconsequencesofdoingso26:
(F2) IfitisrationalforAtoφthensomepwasareasonAhadtoφ.
(F8) Foranyp,ifpwasareasonAhadtoφthenpwasareasonforAtoφ.
(F3) IfAφsdeliberatelythenAφsforareason.
(F7) IfAφsforareasonthentherewasareason,p,forAtoφ.
EITHER:
(F5) Sally’sreasonforrunningwas,interalia,thatshebelievedabearwaschasingher.And;
(F6) Sally’s reason for running was, inter alia, that she heard a bear-likesound.
OR:
(F9) Foranyp, ifAφs for thereasonthatp thenpwasa reason forA toφ.27
For reference, here is a very much non-exhaustive account of which of these prima facie
reasonableclaimssomedifferentproponentsoffavourismaboutreasonstoactreject(where:
û=rejects):
Group1 Group2 Group3
(F2) (F8) (F3) (F7) (F5)&(F6) (F9)
TheReceivedView û û û
Stout(2009),Alvarez(2010),Parfit(2011),Littlejohn(2012) û û û
Hornsby(2008)28 û û û
Schroeder(2008),Comesaña&McGrath(2014) û û û û
Dancy(2000,2014),Davis(2005),Sandis(2009) û û û û
TableII-1:Howproponentsoffavourismrespondtosomeproblemsforit
26Groups1,2and3consistoftheremainingpremisesofTheRationalActionProblem,TheDeliberateAction Problem, and The Psychological Reason Problem (for Favourism) & The Experiential ReasonProblem(forFavourism),respectively.27It’s worth noting that (F9) entails (F7) – so rejecting the former serves as a response to both TheDeliberateActionProblemandTheWrongReasonsProblem.28IthinkthatHornsby’s(2008)viewwithrespecttothetruthof(F9)haschanged.Inanearlierworkshestatesthat‘pmaybetheagent’sreason[foracting]evenwhenitisfalsethatp.’(2007,299)Ifpwerefalse it could not have been a reason for anyone to act because falsehoods don’t count in favour ofanything(thatisnottosaythatnegationsdon’tfavouranything–negationsarefactsthatfavoursomethings,butfalsehoodsaren’tfactsatall).Hornsby’searlierviewisthusmorecloselyalignedtotheviewsofSchroeder(2008)andComesaña&McGrath(2014).However inher laterworkshesuggeststhat ‘aconditionofφingforthereasonthatp,whenonebelievesthatp,isthatoneknowsthatp.’(2008,251)Ithinkher idea in this laterwork is thatonemayact fora reason though theremaybeno reason forwhichoneacts–thisissomethingthat,in§(I)1.5,Iassumedcouldnothappen.Totheextentthatthatisherview,itsnuancesarenotcapturedinmycategorisationschema.
Group1
Group2
Group3
59
Ofcourse,manyoftheproponentsoffavourismaboutreasonstoacthavegoodargumentsas
to why it is acceptable to reject these prima facie reasonable claims. For instance, Alvarez
(2010,142)defendsherrejectionof(F3)bynotingthatwhenanagentactsonafalsebeliefwe
mightwellsaythattheyactfornoreason–someonewhoisrunningtocatchatrainmightsay,
on discovering that it has been cancelled, ‘You mean I ran all this way for no reason?’29
MeanwhileDancy(2000)suggeststhatwecanexplainawayexpressionslike‘Sally’sreasonfor
runningwasthatshebelievedthatabearwaschasingher’asactuallymeaning‘Sally’sreason
forrunningwasthatabearwaschasingher,asshebelieved’–wherethe‘asshebelieved’is
meant to be understood appositionally, in amanner that only qualifieswhat is said, rather
thanchangingthemeaningofit.
I amnot seeking to refute theseargumentshere.Mypoint is not that thesearguments are
wrong or that, more generally, there is no way tomake the rejection of these prima facie
reasonableclaimsintelligibleMypointisthatrejectingtheseprimafaciereasonableclaimsisa
stancethatdemandssomeexplanation–because,indoingsoyouarerejectingsomethingthat
onthefaceofitseemsreasonable.Sothatatleastcountsinfavouroflookingforatheorythat
doesn’tcommitonetosuchrejections.
6 Conclusion
The purpose of this chapter was to show the costs involved in accepting favourism about
reasons toact. Ihaveargued that ifonewants toaccept that view,onemust reject several
primafaciereasonableclaims.Thisisapricethatsomearehappytopay;Iwouldrathernot.
29Alvarezput thisparticular example tome in adiscussionof thispoint.Bywayof further response,Parfit(2001)andAlvarez(2010)wouldbothsuggestthatweshouldsaythatbothSallyandthepersonwhosetrainwascancelledrunforanapparentreasonbutnotagenuinereason.
60
(III)
Actingforpsychologicalreasons
InwhichIshowwhatitcoststothinkthatthereasonforwhichanagentactsisalways a feature of their psychology. I show how ‘psychologism about thereasonsforwhichweact’clasheswithsomeprimafaciereasonableclaims. Inparticular,Ishowthatisinconsistentwiththeideathatweareoftenabletoactfor reasons thatmakeouractionsmorallyworthyand,moregenerally,worthdoing.
Myfriendhaswonamuch-covetedaward;IreadaboutitinanewspapersoIcallherupand
congratulateher. Itseemsnatural tosaythatmyreasonforcongratulatingherwasthatshe
hadwonanaward.Italso,Ithink,seemsnaturaltosaythatmyreasonforcongratulatingher
wasthatIreadthatshehadwonanaward.Anotherexample:ifJonathanseessomeonewhois
aloneandintrouble,hecould,undertherightcircumstances,helpherforthereasonthatshe
wasaloneandintrouble.1
Thesefairlyanodyneobservationscreatealotofdifficultyforpsychologismaboutthereasons
forwhichweact,whichsaysthatanagent’sreasonforactingcanonlyeverbeafeatureofher
psychology,andisprobablythedefactoaccountofwhatanagent’sreasonforactingis.
The purpose of this chapter is to set out a number of prima facie reasonable claims that
psychologism about the reasons for which we act must reject.2This is not meant to be a
conclusive argument against that view – it is only meant to show that accepting this view
comesatsomecost.
1 Someprimafaciereasonableclaims
1.1 Myreasonforcongratulatingmyfriend
I’ve already statedmy firstprima facie reasonable claim, but it’sworth re-iterating;when I
readinanewspaperofrecordthatmyfriendhadwonanawardandIconsequentlycallherup
tocongratulateher,itisnaturaltosaythat:
(P1) My reason for congratulatingmy friendwas, interalia, that shehadwonanaward.
Itis,Ithink,similarlynaturaltosaythat:
1ThisexampleisfromDancy(2000).2Astheformerchapterdidforfavourismaboutreasonstoact.
61
(P2) Myreasonforcongratulatingmyfriendwas,interalia,thatIreadthatshehadwonanaward.
We should add to this observation two seemingly obvious remarks, whose importance will
becomeclear(ifitisnotalready)whenweconsidertheproblemsforpsychologismaboutthe
reasonsforwhichweact:
(P3) Thefactthatmyfriendhadwonanawardisnotafeatureofmypsychology.
(P4) The fact I read that my friend had won an award is not a feature of mypsychology.
1.2 Morallyworthyactions
SupposethatJonathanisonhiswaytotheofficewhenheencounterssomeonewhoisalone
andintrouble.Takingallmorallyrelevantfeaturesofthesituationintoaccount,withoutone
thoughttoomany,withoutanyundueconsiderationsoffurtheringhisownendsorotherless
upstandingconcerns,hedulycomestoheraid.Isubmittoyouthatwhathedoesisamorally
worthyact:
(P5) Jonathan’sactofhelpingthewomanismorallyworthy.
Whatdoesthemoralworthofhisaction,andactsingeneral,comedownto?Oneseemingly
relevantconsiderationisthis: ifJonathanhadhelpedheronlybecauseheknewthatshewas
veryrichandwouldrewardhimamplyfordoingso,themoralworthinessofhisactdissipates.
AsJuliaMarkovitsnotes,‘whenwedotherightthingbecauseithappenstosuitus,orhappens
tobe inour interest,ouractionhasnomoralworth.This is intuitive.Morallyworthyactions
mustbeperformedfortherightreasons.’3(2010,203)
So,wecandotherightthingwithoutdoingsomethingmorallyworthyifwedon’tdoitforthe
rightreasons.Andwhataretherightreasons?Theyaretheonesthatmaketheactionright:
Myactionismorallyworthyifandonlyif…IperformtheactionImorallyoughttoperform,forthereasons4whyitmorallyoughttobeperformed.(Markovits2010,205)
WhenIdothemorallyrightthing,becauseitistherightthingtodo,myreasonsarethefactsthatmakethatactionright.(GarrardandMcNaughton1998,53)
I take this tobean intuitiveview,and itprovidesuswith thebasis forournextprima facie
reasonableclaim:
3Markovits’ original says ‘the right (motivating) reasons’ – I omit the parenthetical terminologicalremark only because I am trying to avoid this terminology (see §(I)4). On her account ‘motivatingreasons’are‘thereasonsforwhichanagentacts’and‘normativereasons’are‘thereasonsforanagenttoact’.4HereMarkovits’originalreads‘the(normative)reasonswhy’–Iomitittheparentheticalterminology,again,toavoidterminologicalconfusion(seefn.3).
62
(P6) A’sφingismorallyworthyonlyifAφsforareasonthatmakesφingright.5
So,whatJonathandoesismorallyworthyonlygiventhathedoesitforareasonthatmakesit
right.
1.3 Whatmakesitright?
SowhatmakesitrightforJonathantohelpher?Whyisitthathelpingheristherightthingto
do?AsDancynotes,‘itisbecausesheisintroublethatIoughttohelpher,notbecauseIthink
sheisintrouble.’(2000,52)
Dancy’spointisthatitistheobjectivefeaturesoftheworldthatdeterminewhetherornotan
actisthemorallyrightthingtodo,andnotfeaturesoftheagent’spsychology.Itisrightforme
torecyclebecauseitwillhelptheenvironment–andthatissoevenifIdon’tknowthatitwill
helptheenvironment.Likewise, it is right for Jonathantocometothewoman’saidbecause
sheisintrouble,andthatwouldbesoevenifhehadnoideathatshewasintrouble.6
Of course, sometimes features of our psychology might enter into the fray as moral
considerations: youmight say that regardlessofwhetherornot she is in trouble,given that
Jonathanthinkssheis,therightthingforhimtodoistotryandhelpher.Thatis,youmightsay
thatnothelpingsomeonewhoyoubelievetobeintroubleiswrong,regardlessofwhetheror
nottheyareintrouble.AndinthatcaseitisafeatureofJonathan’spsychologythatmakeshis
actionright.Onthisaccounthisactionisdoublyright:bothbecausehelpingherwillsaveher
fromtroubleandbecause,ifhedoesso,hewon’tbeneglectingsomeonehebelievestobein
trouble.
However, it’s also quite possible that Jonathan’s act is made right only by the objective
featuresofthesituation.Evenifyouthinkthatjustbelievingthatsheisintroublecouldmake
helpingher theright thingtodo (regardlessofwhetherornotshe’s introuble), it’spossible
thatgiventherestofwhathebelieves,believingthatsheisintroubledoesn’tmakeitrightfor
himtotrytohelpher,althoughthefactthatsheisintroublecontinuestodoso.Forinstance,
supposethatwhilehebelievesthatsheisintroublehealso(falsely)believesthatifhewereto
trytohelpheritwouldonlyworsenhersituation–inthatcasehisbeliefthatsheisintrouble
5This is entailed byMarkovits’ bi-conditional. I use the weaker claim because it is sufficient for mypurposes.6I’m not doing an analysis of what it is for an action to be ‘right’ and certainly not of what ‘ought’means.Evenifthereisanambiguitybetweenobjectiveandsubjective‘ought’claims,Ithinkthereisaclear and commonplace sense of ‘right’ actions that is independent of the agent’s perspective – andthat’sthesenseIamworkingwith–anditis,importantly,inthissenseofbeing‘right’thatIthink(P6)isanintuitiveclaim.Ifitweren’tthenitwould,forinstance,beunclearhowonecoulddotherightthingforthewrongreasons,orthewrongthingfortherightreasons.
63
doesnotmakeitrightforhimtotrytohelpher(becausehethinksthatbyhelpingheritwould
make herworse off). However, since he could, in fact, help herwithoutworsening her lot,
helpingherisstill,insomeobvioussense,therightthingtodo.
WhileIdon’tsupposethatJonathanisinthesituationsodescribed(inparticular,hedoesthink
thathecouldimproveherlotbyhelpingher),sinceIwanttoremovecomplicatingfactors,I’m
going to suppose that Jonathan is ina situationsuch that it isonlyobjective featuresof the
circumstancethatmakehisactionright.Itakeitasagiventhatthatisatleastpossibleevenif
itisalsopossiblethatfeaturesofhispsychologycouldmakehisactionright.
To what end, all this convoluted reasoning? It is tomake several points: objective (that is,
worldly,notpsychological)featuresofasituationcanmakeanactionright;andwhilstfeatures
ofanagent’spsychologymayalsobeabletomakeanactionright, theydon’tneedto–the
reasonswhyit isrightforsomeagenttodosomeactioncouldhavenothingtodowiththeir
psychology,and I’msaying that that isactually thecase in theexampleof Jonathanand the
troubledwoman,asIconstructit.Thus,wecan,exhypothesi,makethefollowingprimafacie
reasonableclaim:
(P7) NofeaturesofJonathan’spsychologymakehelpingthewomanright.
1.4 Actingforreasonsthatmakeitworthacting
IfmyfriendwinsanawardandIcongratulateheritwillletherknowthatI’vethoughtofher
andampleasedforher–andshe’llgetsomejoyfromthat(andotherthingsbesides).Maybe
I’llalsogetsomejoyfromittoo.Inthesensediscussedin§(II)3.2,congratulatingmyfriendis
somethingthatis,forme,worthdoing.
Whatmakescongratulatingmy friendanact that is, forme,worthdoing? It seemsthat it is
things like this: the fact that she won an award, the fact that it would please her to be
congratulated(giventhatshehadthewontheaward),andsoon.Thatis,factsabouttheway
theworldismakemyactworthdoing(includingfactsaboutmyfriend’spsychology).EvenifI
didn’tknowthesethings,ifIhadnoideaaboutherhavingwontheaward,congratulatingher
wouldstillbeworthdoing–Ijustwouldn’tdoit.Solet’ssaythis:
(P8) Thefactthatmyfriendwonanawardmakescongratulatingherworthdoing.
Now,here’sarelevantprimafaciereasonableclaimaboutreasonsthatIwanttoputforward:
64
(P9) Foranyp,ifpmakesA’sφingworthdoingthenAcouldφforthereasonthatp.7
A note on ‘could’: what is themodal concept we are working with here? I think there are
intuitivegrounds foraquite restrictiveone,however, all that I need for thepurposeof this
argument is logical possibility: that is, if p makes A’s φing worth doing then it is logically
possiblethatAcouldφforthereasonthatp.
Thisseems,tome,likeaclaimthatishardtodeny,butifyouneedsomepersuasionconsider
this:ifyouaren’tdoingsomethingforreasonsthatmakeitworthdoing,whatreasonsareyou
doingitfor?Thewrongones?It justseemsoddtometothinkthattherecouldbeanaction
that isworthdoingbutwhichonecouldneverdoforreasonsthatmakeitworthdoing.It is,
seemingly, onlywhen you do something for reasons that actuallymake itworth doing that
you,‘dotherightactionfortherightreason.’(Lord2008,2)Indeed,theveryideaofacting‘for
the right reasons’ seems to depend upon the logical possibility (and probably something
strongerthanthat)ofactingforreasonsthatmakeone’sactionworthdoing.Thefalsityof(P9)
would implythat therecouldbesomeactionsthatareworthdoingthatonecouldneverdo
‘fortherightreasons’.Iwouldsuggestthatthatis,atleast,aprimafacieimplausibleview,so
that(P9)is,atleast,primafaciereasonable.
2 Psychologismaboutthereasonsforwhichweact
RecallthefollowingclaimaboutreasonsfromTableI-4:
Psychologismaboutthereasonsforwhichweact:Foranyp,pisareasonforwhichAφ’difandonlyifpisafeatureofA’spsychologythatrationalisesφingandexplains(intherightway)whyAφ’d.
Davidson’s ‘Actions, Reasons andCauses’ (2001a) is probably theprogenitor of this view. In
addition,Ithinkthatthisclaim(orsomethingsufficientlysimilartoit)isalsoadvocatedbyTurri
(2009), Gibbons (2010) andMitova (2015, 2016) (although they don’t articulate it in these
terms). This claim is also perhaps most commonly associated with what I have called the
‘ReceivedView’(ofwhichtheformerarenotadvocates(seeTableI-6)).
7ThisiscloselyrelatedtoBernardWilliams’claimthat:‘Iftherearereasonsforaction,itmustbethatpeople sometimes act for those reasons, and if they do, their reasons must figure in some correctexplanationoftheiraction.’(Williams1981,102)(SeealsoDancy2000,101;Smith2004,175;Hornsby2007,301;Raz2009,194;Hieronymi2011,415;WayandWhiting2016,214).IusethisversionbecauseWilliams(andotherswhoexpressthisclaim)takefavourismaboutreasonstoact(seepreviouschapter)for granted – but someonewho rejects that view can satisfyWilliams’s claimwithout difficulty. (P9)entailsWilliams’ claim (given favourism about reasons to act), while providing a claim that onewhorejects the latterwill still alsohave to reject inorder toendorsepsychologismabout the reasons forwhichweact(see§3.2).
65
Why would anyone hold this view? I think that most philosophers who endorse it do so
becausethey,eitherimplicitlyorexplicitly,thinkthatanagent’sreasonforactingmustmake
their action rational and that only features of an agent’s psychology canmake their actions
rational.However,thefocusofthisdiscussionisontheproblemsthatarisefrompsychologism
aboutthereasonsforwhichweact,soIleaveasidethereasonswhyonemightendorseitfora
laterdiscussion.8
ItispossiblethatsomeofthosetowhomIhaveattributedpsychologismaboutthereasonsfor
whichweact(seeTable I-6)mightbalkatmyexactwordingof it.9Nonetheless, I thinkallof
thosetheoristsshareacommitmenttothefollowing,whichisentailedbypsychologismabout
thereasonsforwhichweact,10andwhichiswhatreallycreatestroubleforit:
(PSY) Foranyp,ifAφsforthereasonthatpthenpisafeatureofA’spsychology.11
3 Theproblemsforpsychologism
I present three problems for (PSY): The Moral Worthiness Problem; The Right Reasons
Problem;andTheNon-PsychologicalReasonProblem.
3.1 TheMoralWorthinessProblem
The Moral Worthiness Problem is this: Jonathan’s act of helping the woman was morally
worthy,sohediditforreasonsthatmadeitright,sofeaturesofhispsychologymadeitright,
but no featuresof his psychologymade it right! Explicitly, the following claims aremutually
inconsistent:
(P5) Jonathan’sactofhelpingthewomanismorallyworthy.
(P6) A’sφingismorallyworthyonlyifAφsforareasonthatmakesφingright.
(PSY) Foranyp,ifAφsforthereasonthatpthenpisafeatureofA’spsychology.
(P7) NofeaturesofJonathan’spsychologymakehelpingthewomanright.
8In§(VIII),Iwilldiscuss,atlength,themotivatingargumentforpsychologism.9Inthesamewaythat,say,ScanlonandParfitmighthavebalkedatmycharacterisationoftheirviewsas‘favourismaboutreasonstoact’(see§(II)3.3).10Forthesakeofclarity:(PSY)isweakerthanpsychologismaboutthereasonsforwhichweact, inthesensethatthelatterentailstheformer,buttheformerdoesnotentailthelatter.Mypointisthat(PSY)iswhatcausespsychologismaboutthereasonsforwhichweacttofacetheproblemsitdoes.11Cf. ‘Psychologism… is the claim that the reasons for which we act are psychological states ofourselves.’(Dancy2000,98).
66
3.2 TheRightReasonsProblem
The Right Reasons Problem is this: the fact that my friend has won an award makes
congratulating her worth doing, so it’s logically possible forme to congratulate her for the
reason that shehaswonanaward, so the fact thatmy friendhaswonanawardmustbea
featureofmypsychology,butitisn’t!Explicitly,thefollowingclaimsaremutuallyinconsistent:
(P8) Thefactthatmyfriendwonanawardmakescongratulatingherworthdoing.
(P9) Foranyp,ifpmakesA’sφingworthdoingthenAcouldφforthereasonthatp.
(PSY) Foranyp,ifAφsforthereasonthatpthenpisafeatureofA’spsychology.
(P3) Thefactthatmyfriendhadwonanawardisnotafeatureofmypsychology.
3.3 TheNon-PsychologicalReasonProblem
TheNon-Psychological Reason Problem is this:my reason for congratulatingmy friendwas,
interalia,thatshehadwonanaward,sothatmusthavebeenafeatureofmypsychology,but
itisn’t!Explicitly,thefollowingclaimsaremutuallyinconsistent:
(P1) My reason for congratulatingmy friendwas, interalia, that shehadwonanaward.
(PSY) Foranyp,ifAφsforthereasonthatpthenpisafeatureofA’spsychology.
(P3) Thefactthatmyfriendhadwonanawardisnotafeatureofmypsychology.
3.4 TheExperientialReasonProblem(forPsychologism)
TheExperientialReasonProblem(forPsychologism) is this:my reason forcongratulatingmy
friendwas, interalia, that Ireadthatshehadwonanawardinthenewspaper,sothatmust
havebeenafeatureofmypsychology,butitisn’t!Explicitly,thefollowingclaimsaremutually
inconsistent:
(P2) Myreasonforcongratulatingmyfriendwas,interalia,thatIreadthatshehadwonanaward.
(PSY) Foranyp,ifAφsforthereasonthatpthenpisafeatureofA’spsychology.
(P4) The fact I read that my friend had won an award is not a feature of mypsychology.
67
4 Responsestotheproblemsforpsychologism
Whatcanaproponentofpsychologismaboutthereasonsforwhichweactdo?Itseemstome
thereislittletonochoiceinthematter.(P3)and(P4)areundeniable:ofcoursefactsaboutmy
friendorwhatI’vereadarenotfeaturesofmypsychology.
We constructed the Jonathan example so as to ensure the truth of (P7) (the claim that no
features of Jonathan’s psychology made his act right), so unless one wants to change the
conceptof ‘makingright’, toadifferentonetothatwhich Iamusing,onecannotreject(P7)
(and if onewere to change the concept, thatwouldobviously justbeanew (perhapsmore
solvable) problem, it would not be a solution to the problem I am posing). Similar remarks
countagainstrejecting(P8).
What about (P5) (the claim that Jonathan’s act ismorallyworthy)? Endorsing some formof
moralanti-realismcouldallowonetoclaimthatnoactsaremorallyworthy,sothatJonathan’s
actisn’teither,therebyrejecting(P5).Thisavoidshavingtoreject(P6),butonlybytrivialising
it – it is trueonlybecause theantecedent isnever satisfied– so I think that a strategy that
rejects(P5)isatleastasprimafacieimplausibleasjustrejecting(P6).
So, leaving aside the possibility of rejecting (P5), there are no choices for the proponent of
psychologismaboutthereasonsforwhichweact;theymustrejectallofthefollowingprima
faciereasonableclaims:
(P1) My reason for congratulatingmy friendwas, interalia, that shehadwonanaward.12
(P2) Myreasonforcongratulatingmyfriendwas,interalia,thatIreadthatshehadwonanaward.13
(P6) A’sφingismorallyworthyonlyifAφsforareasonthatmakesφingright.14
(P9) Foranyp,ifpmakesA’sφingworthdoingthenAcouldφforthereasonthatp.15
Of course, one can perhaps put forward good arguments as to why these claims are
nonethelessfalse,orwhyitisnonethelessacceptabletorejectthem.Forinstance,onemight
arguethat(P9)isfalse,byprovidingcasesinwhichoneseeminglycannotdosomeactionfora
12Rejecting(P1)solvesTheNon-PsychologicalReasonProblem.13Rejecting(P2)solvesTheExperientialReasonProblem(forPsychologism).14Rejecting(P6)solvesTheMoralWorthinessProblem.15Rejecting(P9)solvesTheRightReasonsProblem.
68
reason that makes it worth doing.16 Alternatively, perhaps one could construct counter-
examplesto(P6).17
My point, again, is not that one cannot argue against these claims, or that the arguments
againsttheseclaimsarewrong,ratheritisjustthattheseclaimsareprimafaciereasonableso
thatrejectingthemisastancethatdemandssomeexplanation–youarerejectingsomething
thatonthefaceofitseemsreasonable.So,again,thatatleastcountsinfavouroflookingfora
theorythatdoesn’trejectsuchprimafaciereasonableclaims.
5 Conclusion
Aswith theprevious chapter’sdiscussionof favourismabout reasons toact, thepurposeof
thischapterwasnot toargueagainstpsychologismabout the reasons forwhichweact,but
onlytoshowthecostofacceptingit,whichisthatonemustreject(P1),(P6)and(P9),which
are,asIhaveargued,allprimafaciereasonable.
16Schroeder (2007, 33) gives the followingexample: The fact that there is a surprisepartywaiting athomeforhimmakesgoinghomeearlyworthdoingforNate.However,Natecannotgohomeearlyforthereasonthatthereisasurprisepartywaiting(becauseitwon’tbeasurprisepartyifheisawareofit,andhecan’tgohomeforthereasonthatthereisasurprisepartywaitingforhimifheisn’tawarethatthereis).17Forinstance,onemightthinkthatdiscoveringthecureforcancerexclusivelyforthereasonthatonewillbeadmired forhavingdoneso isnonethelessmorallyworthy. Iwoulddisagree,as, I think,manyotherswould.Ultimately, I think,whetherornotonerejects itdependsuponwhetherornotone isaconsequentialist about moral worthiness – noting that one can be a consequentialist about moralrightnesswithoutbeingaconsequentialistaboutmoralworthiness(cf.Mill:‘themotivehasnothingtodowiththemoralityoftheaction,thoughmuchwiththeworthoftheagent.’(1863,29)).
69
(IV)
Actingforwhatyoubelieve
InwhichIshowwhatitcoststothinkthatareasonforwhichanagentacts isthe content of thebelief theyactedon. I showhow ‘deliberativismabout thereasons for which we act’ clashes with some prima facie reasonable claimsaboutthefactivityofreasons,theexplanatorypowerofthereasonsforwhichweact,thefactivityofexplanationandwhatanagent’sreasonsforactingareinGettiercases.Isetoutwhichclaimstheproponentofthisviewmustchoosebetweenrejecting.
Thereisanaccountofthereasonsforwhichanagentactsthataimstoreconciletheideathat
anagentwhoacts intentionally,deliberatelyandpurposefullyalsoactsforareasonwiththe
ideathatthereasonsforwhichtheyactareoftennotfeaturesoftheirpsychology.According
tothisaccount,whichiswhatIhavecalled‘deliberativismaboutthereasonsforwhichweact’,
theconsiderationsinlightofwhichanagentacts,whichiswhattheybelievedwhentheyacted
ontheirbelief,arethereasonsforwhichtheyact.
Theproblemsforthisview,asIshallshow,arisefromtheprimafaciereasonableclaimsthat
‘being a reason’ is a seemingly factive property, that one’s reasons for acting explain one’s
actions and that even when the considerations in light of which one acts are true, we are
sometimesreluctanttocallthemtheagent’sreasonsforacting.
1 Someprimafaciereasonableclaimsaboutreasons
1.1 Thefactivityofreasons
IfSally is runningbecauseshemistakenly thinks thatabear ischasingher,wedon’t tendto
say, ‘herreasonforrunning is thatabear ischasingher,eventhoughone isn’t’.Onereason
thatwedon’t tend to say it is, I submit, that theexpressionsoundsodd.1AsAlvarezputs it,
thereisan‘airofparadox’aboutit:
‘OthellokillsDesdemonaforthereasonthatDesdemonahasbeenunfaithfultohim,althoughshehasnotbeenunfaithfultohim’soundsonlymarginally lessparadoxicalthan‘OthellokillsDesdemonabecauseofthefactthatDesdemonahasbeenunfaithfultohim,althoughshehasnotbeenunfaithfultohim’.It isnotthatonecannotgivemeaningfulinterpretationstotheseexpressions: wemayhearthemasconveyingwhattheagenthim-orherselfwouldhavesaidifaskedabouttheirreason...Allthesame,thefactthattheseexpressionshaveanairofparadox
1Indeed,thisclaimwasmetwithlaughterwhenIputittoaseminaraudience.
70
and require this special interpretation are arguably explained by the thought that both ‘thereasonthat’and‘thefactthat’arefactiveoperators.(Alvarez2016b,8)2
I thinkAlvarez’sreading iscorrect: it isnotthatwecan’t findanymeaning intheexpression
‘Sally’sreasonforrunningwasthatabearwaschasingher,eventhoughonewasn’t’–it’sthat
theexpressionsoundsodd,andwehavetore-interpretittomakesenseofit.Moreover,the
factthatwedon’tsaysuchthingsshouldtellussomethingabouttheapplicabilityoftheterm
tosuchcases.
Isuggest,therefore,thattheclaimthat‘…istheagent’sreasonforacting’isafactivepredicate
isprimafaciereasonable.Thus:
(D1) Foranyp,ifpwasA’sreasonforφingthenpisthecase.
1.2 Theexplanatorypowerofreasons
Wecanexplainanagent’sactionbygiving their reason foracting.Moreover, it is seemingly
naturaltothinkthatwhenwesoexplainanagent’saction, it istheagent’sreasonforacting
itself that explains their action. When I say that Sally’s reason for running was that she
believedthatabearwaschasingheritstronglyseemstosuggest(ifnotdirectlyimplies)that
Sallyranbecauseshebelievedthatabearwaschasingher.Likewise,ifIsaythatmyreasonfor
congratulatingmyfriendwasthatshehadwonanaward itsuggeststhat Icongratulatedmy
friendbecauseshehadwonanaward.
Theideathatanagent’sreasonforactingalwaysexplainstheiractionis,asLilianO’Briennotes
‘verywidelyshared’(2015,282).Forinstance:
Iactinlightofthosereasons.TheyarethereasonswhyIdowhatIdo.(Dancy2000,103)
Whenanagentactsfora(specific)reasonthatveryreasonisalsotheexplanation(oratleastpartoftheexplanation)ofwhyshedidwhatshedid.(Heuer2004,45)
Takingsomethingasone’sreason,inactingonit,istakingitasanexplanatoryreason.(Setiya2007,36)
A fairly standardway of linking reasons for action and explanations of action… is thatwhensomeoneactsforareasonthentheirreasonforactingthatwayexplainstheiractingthatway.(Stout2009,57emphasisadded)
When there is a reason for which an agent acted then that reason explains (features in anexplanationof)thataction.(Raz2011,14)
2For others who take such remarks to sound odd see Unger (1978, 208), Scanlon (2014, 36), Dancy(2008a,267)andComesaña&McGrath(2014,75)–the lattertwoareparticularlyworthnotingsincetheyadvanceadeliberativeaccountandnonethelessrecognizetheoddnessofsuchexpressions.
71
I suggest that it is sowidelyheldpreciselybecause itmatchesour intuitions.Theclaim that
somethingwasone’sreasonforactingseemstobesubstitutablefortheclaimthatoneacted
becauseofit.Thus,wehavethesecond,primafaciereasonableclaimforthisdiscussion:
(D2) Foranyp,ifpwasA’sreasonforφingthenpexplainswhyAφ’d.
1.3 Thefactivityofexplanation
Consider the following statements: I took my umbrella because it was raining, although it
wasn’training.Thereasonwhythewindowbrokewasthatabrickstruck it,althoughabrick
didn’t strike it. That it’s unfair explainswhy it’swrong, although it’s not unfair. All of these
statementssoundstrangetothepointofunintelligibility.Whatmakesthemsoundstrangeis
thatexplanatoryrelations(whethercausalornon-causal)areseeminglyfactiverelations(and
that is sowhether theyarepickedoutby ‘because’, ‘explains’or ‘reasonwhy’)– something
cannotexplainunlessitistrue.
I take this to be such an obviously prima facie reasonable claim that I won’t defend its
intuitivenessfurther:
(D3) Foranyp,ifpexplainsanythingthenpisthecase.
1.4 Gettiercases
EdmundGettier(1963)introducedanow-familiarsortofcharactertoepistemology:someone
whohasajustified,truebeliefthatfallsshortofknowledge.Gettiercharactersarerelevantto
thisdiscussionbecause,whenanagent’s justifiedbelief is truebyhappyaccident,what the
agent believes (despite its truth) seemingly does not explain their action, as Hornsby
demonstrates:
Edmund…believesthattheiceinthemiddleofthepondisdangerouslythin,havingbeentoldsobyanormallyreliablefriend,and…accordinglykeepstotheedge.ButEdmund’sfrienddidn’twantEdmundtoskateinthemiddleofthepond(nevermindwhy),sothathehadtoldEdmundthat the ice therewasthindespitehavingnoviewaboutwhetherornot itactuallywasthin.Edmund,then,didnotkeeptotheedgebecausethe ice inthemiddlewasthin.Supposenowthat, as it happened, the ice in themiddle of the pondwas thin. Thismakes no difference.Edmundstilldidn’tkeepto theedgebecausethe icewas thin.The fact that the icewas thindoesnotexplainEdmund’sacting,eventhoughEdmunddidbelievethatitwasthin,andeventhough the fact that itwas thin actuallywas a reason for him to stay at the edge. (Hornsby2007,251emphasisadded)
I think Hornsby’s claim is intuitive: there is no sense in which Edmund’s action was the
consequenceoftheicebeingthin,and,indeed,theicecouldhavebeenthickinthemiddleof
the pond and Edmund would have stayed at the edge just the same. And given those
observations,itseemshardtoseehowitcouldbetruethatEdmundstayedattheedgeofthe
lakebecausetheicewasthin.So,Isuggestthatthefollowingclaimisprimafaciereasonable:
72
(D4) The fact that the ice was thin does not explain why Edmund stayed by theedgeofthelake.
2 Deliberativismaboutthereasonsforwhichweact
RecallthefollowingclaimaboutreasonsfromTableI-4:
Deliberativismaboutthereasonsforwhichweact:Foranyp,pisareasonforwhichAφ’difandonlyifpisaconsiderationinlightofwhichAφs.
Whiledeliberativismaboutthereasonsforwhichweactischaracterisedbythisbi-conditional,
itmayhelpthereaderwhenwecometotheproblemswiththeaccountifweseparateoutthe
left-to-right and right-to-left readings (noting that the deliberative account is committed to
both).Thus:
(DEL1) Foranyp,ifpisaconsiderationinlightofwhichAφsthenpisA’sreasonforφing.
(DEL2) Foranyp,ifpisA’sreasonforφingthenpisaconsiderationinlightofwhichAφs.3
Itiseasyenoughtosaythatsomethingis‘aconsiderationinlightofwhichoneacts’,perhaps,
butwhat does itmean?4 JonathanDancy,who is probably the progenitor of contemporary
deliberativismaboutthereasonsforwhichweact,providesthefollowinginsight:
Thereasonsforwhichweactaretheconsiderations inthe lightofwhichwedowhatwedo.Thesearethefeatureswhichwetaketotellinfavourofsoacting;theywillfigureprominentlyinourdeliberation.(Dancy2006,123)
So, a condition on something’s being a consideration in light of which one acts is that one
should have taken it to ‘favour’ doing that action. Dancy, and other proponents of
deliberativism,understandthefavouringrelationintheconventionalwaythatIhavealready
discussed(see§(I)4.4and§(II)3.1)–as‘making,insomerespect,worthdoing’.So,acondition
onsomething’sbeingaconsiderationinlightofwhichoneactsisthatonemusttakethatthing
tomakeone’saction,insomerespect,worthdoing.
But this isnotall there is tobeingaconsideration in lightofwhichoneacts;onecouldtake
somethingtomakeanaction,insomerespect,worthdoingandactuallydothatactionwithout
thatthingbeingtheconsiderationinlightofwhichonedidit.Forinstance,Icouldtakethefact
that it’spleasantoutsidetomakegoingoutsideworthdoingand Icouldactuallygooutside,
3It’sworthnotingthatmanywhodonotsubscribetodeliberativismaboutthereasonsforwhichweactnonethelesstake(DEL2)tobetrue(e.g.Stout2009;Alvarez2010;Parfit2011;Hyman2015).4 These are what Scanlon (1998) calls ‘operative reasons’, what Olson and Svensson (2005) call‘deliberativereasons’andwhatDancy(2000)andSchroeder(2008)call‘motivatingreasons’.
73
buttheconsiderationinlightofwhichIdosomightbe,say,thatIneedtogetlunch(Iwould
stillhavegoneoutsideiftheweatherhadbeendreadful).
Whatmakessomethingaconsiderationinlightofwhichoneactsisnotmerelythatonetakes
ittofavourone’saction,butthatoneactsonit.Whatthatmeansmaybealittleopaque,butI
thinkthatwecansay,bywayofexplication,thatoneactsonsomethingthatonetooktomake
one’sactionworthdoingifoneactsasaresultoftakingittomakeone’sactionworthdoing.5
Whatdoesthisallmean?Well,whenIseerainoutside,Ibelievethatit’srainingandIbelieve
thatmyumbrellawillkeepmedry.Ireasonthus:it’sraining,myumbrellawillkeepmedry,I
wanttostaydry(orjudgeitgoodorwhathaveyou),andsoIhadbettertakemyumbrella.I
taketheseconsiderationstofavourtakingmyumbrellaandItookmyumbrellaasaresultof
taking those considerations to favour doing so. It is these two conditions (taking some
considerationtofavouracting,andactingasaresultoftakingittofavourdoingso)thatmake
someconsiderationaconsiderationinlightofwhichanagentacts.Somuch,then,forwhatit
istobeaconsiderationinlightofwhichoneacts.
3 WhatSallyandEdmundtooktofavouracting
3.1 Sally’sconsiderations
WhatweretheconsiderationsinlightofwhichSallyran?Thatis,whatwasitthatshetookto
favour running (that is, to make running, in some respect, worth doing)?Well, Sally heard
something that sounded likeabear,andher thoughtprocess,albeitaquickone,musthave
beensomethinglike:‘Abear’scoming!Thesafe-houseisnearby–Icanmakeit. Ihadbetter
run!’ So, amongst the considerations that Sally took to favour runningwas that a bearwas
chasing her. Of course, a bearwasn’t chasing her, but that doesn’t prevent it from being a
considerationinlightofwhichsheacted,anymorethanthefactthatabearwasn’tchasingher
preventedherfrombelievingthatonewas.
Inorderforonetotakesomethingtofavoursomeaction,forittobeaconsiderationinlightof
which one acted, it is, as I have noted, enough that one believes it, believes that it favours
one’saction(inthesenseofmakingit,insomerespect,worthdoing)andactsonthosebeliefs.
Inparticular,itisnotnecessarythatoneknowit,oreventhatitbethecase,becausethebelief
operator,unlikeknowledge,isnotfactive.
5The natural interpretation of ‘as a result’ here is an explanatory one – but I want to avoid such acommitment since some (e.g. Dancy 2000) don’t think the ‘taking’ is explanatory, but is merely an‘enablingcondition’fortheaction.
74
This is a noteworthy upshot of the deliberative account: an agent can act in light of a
considerationthatwasfalse.Andthis iswhatSallydoes:Sally takes thatabearwaschasing
hertofavourheraction,andactsinthelightofthatconsideration.
(D5) AconsiderationinlightofwhichSallyranwasthatabearwaschasingher.
Was the fact that shebelieved thatabearwaschasingheraconsideration in lightofwhich
Sallyran? Itwasnot.Sallydoesnotthinktoherself: ‘Hmmm, Ibelievethatabear ischasing
me.ThatIbelievethatabearischasingmemakesrunningworthdoing(regardlessofwhether
ornotabear ischasingme…),so Ihadbetterdo it.’ ItwouldbeoddforSally toreasonthis
way, because it’s only in unusual circumstances thatwe take considerations aboutour own
psychologytofavourouractions.
Theseremarksrelatetoremarksmadein§(I)4.2,thatareperhapsworthrevisiting.Recallthat
Sambelievesthatthesecurityservicesaretryingtoreadhermind.Shetakesthatthesecurity
servicesare trying to readhermind to favourwearinga foilhat (she takes it tobe, in some
respect,worth doing). However, knowing that she has a delusional disorder, she also takes
thatshebelievesthatthesecurityservicesaretryingtoreadhermindtofavourgoingtoseea
doctor.Samdeliberatesaboutthewayshetakestheworldtobeaswellasthewayshetakes
hermindtobe.
SallyisnotlikeSam–whenSallydecidestorunit’snotbecauseshehasdeliberatedabouther
ownpsychology–whatSallytakestofavourheractionisthewayworldis,accordingtoher.If
Sally thought thatmerely believing that a bearwas chasing herwouldmake runningworth
doingregardlessofwhetherornotabearwaschasingher,thenitcouldbeaconsiderationin
light of which she runs. But, ex hypothesi, that isn’t what Sally thinks, what Sally takes to
favourrunningisthatabearischasingher(eventhoughnobearisactuallychasingher)–Sally
is just mistaken about what makes running worth doing, because nothing makes it worth
doing.
ThepointisthatweshouldnotinferfromthefactthatSallyrunsbecauseshebelievesthata
bear is chasing her that what she deliberates about is her mental states rather than their
propositionalcontents.Inshort:
(D6) ThatshebelievedthatabearwaschasingherwasnotaconsiderationinlightofwhichSallyran.
WhataboutthefactthatSallyheardabear-likesound?Isthataconsiderationinlightofwhich
sheran?MaybeyouwanttosaythatSally’sreasoninggoeslikethis‘Thatsoundedlikeabear!
75
Theremust be a bear coming! I had better run!’ Her reasoningmight go like that. But that
doesn’tmakehearingabear-likesoundsomethingshetooktofavourrunning.
Why not? Because she doesn’t think that hearing a bear-like sound makes running worth
doing.Sallydoesn’tthink:‘ifIhearabear-likesoundIshouldrun,regardlessofwhetherornot
a bear is chasing me.’ Hearing a bear-like sound is what makes her believe that a bear is
chasing her, but it’s that a bear is chasing her that she takes to favour running. That is, if
hearingabear-likesoundplaysanyroleinherdeliberationitisinhelpingherdecidewhatto
believe;butonceshehassettledonbelievingthatabearischasingheritiswhatshebelieves
thatshetakestofavourheraction,notwhatshetooktofavourbelievingit.So:
(D7) Thatsheheardabear-likesoundwasnotaconsiderationinlightofwhichSallyran.
3.2 Edmund’sconsiderations
WhatweretheconsiderationsinlightofwhichEdmundstayedbytheedgeofthepond?That
is,whatwasitthathetooktofavourstayingbytheedgeofthepondandonthebasisofwhich
hedidso?
Well,histhoughtprocessmighthavegone(ifsomewhatelaborately)likethis:‘Theiceisthinin
themiddle.IfIskatethereitmightcrack,Imightfallthrough.Thatwouldbedreadful,perhaps
fatal.I’dbetterjuststayattheside.’Assumingthatthis(orsomethinginitsvicinity)ishowhis
reasoning went, the things that Edmund took to favour his action, and which made the
difference towhat he did,were things like that the icewas thin, that skating on thin ice is
dangerousandsoon.Thus,wecansay:
(D8) AconsiderationinlightofwhichEdmundstayedbytheedgewasthattheicewasthin.
4 Theproblemsfordeliberativism
Theproblemsfordeliberativismaboutthereasonsforwhichweactwillnodoubtbeobvious
fromtheremarksoftheprevioussections.Nonetheless,itisworthbeingexplicitaboutthem.
What follows are four distinct problems for the deliberative account: The False Reasons
Problem; The Explanatory Reasons Problem; The Deliberative Gettier Problem; and The
PsychologicalReasonProblem(forDeliberativism).
76
4.1 TheFalseReasonsProblem
TheFalseReasonsProblemisthis:Aconsiderationin lightofwhichSallyranwasthatabear
waschasingher,soitwasherreasonforrunning,soabearwaschasingher,butnobearwas
chasingher!Explicitly,thefollowingclaimsaremutuallyinconsistent:
(D5) AconsiderationinlightofwhichSallyranwasthatabearwaschasingher.
(DEL1) Foranyp,ifpisaconsiderationinlightofwhichAφsthenpisA’sreasonforφing.
(D1) Foranyp,ifpwasA’sreasonforφingthenpisthecase.
(D9) ItisnotthecasethatabearwaschasingSally.
4.2 TheExplanatoryReasonsProblem
TheExplanatoryReasonsProblemisthis:AconsiderationinlightofwhichSallyranwasthata
bearwaschasingher,so itwasherreasonforrunning,so itexplainswhysheran,soabear
was chasing her, but no bear was chasing her! Explicitly, the following claims aremutually
inconsistent:
(D5) AconsiderationinlightofwhichSallyranwasthatabearwaschasingher.
(DEL1) Foranyp,ifpisaconsiderationinlightofwhichAφsthenpisA’sreasonforφing.
(D2) Foranyp,ifpwasA’sreasonforφingthenpexplainswhyAφ’d.
(D3) Foranyp,ifpexplainsanythingthenpisthecase.
(D9) ItisnotthecasethatabearwaschasingSally.
4.3 TheDeliberativeGettierProblem
TheDeliberativeGettierProblem is this:Aconsideration in lightofwhichEdmundstayedby
theedgeofthepondwasthattheicewasthininthemiddle,sothatwashisreasonforstaying
bytheedge,soitexplainswhyhestayedbytheedge,butitdoesn’texplainwhyhestayedby
theedge!Explicitly,thefollowingclaimsaremutuallyinconsistent:
77
(D8) AconsiderationinlightofwhichEdmundstayedbytheedgewasthattheicewasthin.
(DEL1) Foranyp,ifpisaconsiderationinlightofwhichAφsthenpisA’sreasonforφing.
(D2) Foranyp,ifpwasA’sreasonforφingthenpexplainswhyAφ’d.
(D4) The fact that the ice was thin does not explain why Edmund stayed by theedgeofthelake.
4.4 ThePsychologicalReasonProblem(forDeliberativism)
ThePsychologicalReasonProblem(forDeliberativism) is this:Sally’s reason for runningwas,
inter alia, that she believed a bearwas chasing her, so thatwas a consideration in light of
whichsheran,butitwasn’t!6Explicitly,thefollowingclaimsaremutuallyinconsistent:
(F5) Sally’sreasonforrunningwas,interalia,thatshebelievedabearwaschasingher.7
(DEL2) Foranyp,ifpisA’sreasonforφingthenpisaconsiderationinlightofwhichAφs.
(D6) ThatshebelievedthatabearwaschasingherwasnotaconsiderationinlightofwhichSallyran.
4.5 TheExperientialReasonProblem(forDeliberativism)
The Experiential Reason Problem (for Deliberativism) is this: Sally’s reason for runningwas,
interalia,thatsheheadabear-likesound,sothatwasaconsiderationinlightofwhichsheran,
butitwasn’t!Explicitly,thefollowingclaimsaremutuallyinconsistent:
(F6) Sally’sreasonforrunningwas,interalia,thatsheheardabear-likesound.
(DEL2) Foranyp,ifpisA’sreasonforφingthenpisaconsiderationinlightofwhichAφs.
(D7) Thatsheheardabear-likesoundwasnotaconsiderationinlightofwhichSallyran.
5 Responsestotheproblemsfordeliberativism
What can the proponent of this account do? Rejecting (D9) is not an option: a bearwasn’t
chasingSally–thatmuchweknowforsure.Rejectinganyof(D5),(D6),(D7)or(D8)(theclaims6Thisargument isoften (e.g.Dancy2000,124–25;Alvarez2016b,9)usedbyproponentsof (DEL2) toargueagainsttheideathatSally’sreasonforrunningwasthatshebelievedthatabearwaschasingher.My point is that it is prima facie reasonable to suggest that Sally’s reason for running was that shebelievedthatabearwaschasingher–soifone’stheoryofreasonscommitsyoutorejectingit,itisatsomecostthatitdoesso.7Thisisfrom§(II)1.3,wherewefirstconsideredthisexample.
78
abouttheconsiderationsinlightofwhichSallyandEdmunddidordidnotact) isalsonotan
option–thetruthof(D5)to(D8)justfollowsfromwhatit istobeaconsiderationinlightof
whichoneacts(andtheconstructionoftheexamples).Ofcourseonecouldchangethenotion
of consideration that we are working with, but that would be to change the account, it
wouldn’tsolvetheproblemfordeliberativismaboutthereasonsforwhichweact.
So,theproponentofdeliberativismaboutthereasonsforwhichweactmustreject:
(F5) Sally’sreasonforrunningwas,interalia,thatshebelievedabearwaschasingher.8
(F6) Sally’sreasonforrunningwas,interalia,thatsheheardabear-likesound.9
(D1) Foranyp,ifpwasA’sreasonforφingthenpisthecase.10
And,EITHER:
(D3) Foranyp,ifpexplainsanythingthenpisthecase.11And;
(D4) The fact that the ice was thin does not explain why Edmund stayed by theedgeofthelake.12
OR:
(D2) Foranyp,ifpwasA’sreasonforφingthenpexplainswhyAφ’d.13
Asever,onecan,perhaps,constructcompellingargumentsastowhytheseclaimsare,inspite
of theirprima facie reasonableness, nonetheless false. For instance,Dancy(2000) suggested
thatrejecting(D3),andinsistingthatthereissuchathingasnon-factiveexplanation,mightbe
theappropriateresponsetoTheExplanatoryReasonsProblem.Thisviewhasprovedtobeless
than compelling; indeed, the claim that explanation could be non-factive is apparently so
unpalatableathesisthatevenDancyhasnowabandonedit,yieldingtowhathedescribesas‘a
barrageofcriticism’.14
Therejectionof(D2) isnowthemorefavouredapproachamongstdeliberativists.15Themain
strategy for doing so appears to be this: given that an agent’s reason for acting is an
intentional object (qua what the agent believes), it is not the sort of thing that can do
8Rejecting(F5)solvesThePsychologicalReasonProblem(forDeliberativism).9Rejecting(F6)solvesTheExperientialReasonProblem(forDeliberativism).10Rejecting(D1)solvesTheFalseReasonsProblem.11Rejecting(D3)solvesTheExplanatoryReasonsProblem.12Rejecting(D4)solvesTheDeliberativeGettierProblem.13Rejecting(D2)solvesbothTheExplanatoryReasonsProblemandTheDeliberativeGettierProblem.14SeeDancy(2014).Although,Comesaña&McGrath(2014)appeartohavepickedupthenon-factivebaton.15(E.g.Stoutland1998;Davis2003,2005;Sandis2013;Dancy2014)
79
explaining, so (D2) can’tbe true. Theproblem, it seems tome, is that this lineof argument
reasons to its conclusion by taking (DEL1) as given, when that is precisely what is in
contention.16
Nonetheless,aswiththepreviousdiscussions,what isat issuehere isnotwhetherornotan
argumentcanbegiventhatmakesrejecting(D2)(and(F5),(F6)and(D1))tolerable,mypointis
just that it counts against deliberativism about the reasons for which we act that such an
argument needs to be given in the first place. If your theory has some counter-intuitive
consequencesitscounter-intuitivenessdoesn’tmakeitwrong,butitcertainlycountsagainstit
whencomparedtoamoreintuitivetheory.
6 Conclusion
This chapter has shown that the deliberative account must reject a number of prima facie
reasonable claims, as previous chapters did for favourism about reasons to act and
psychologismaboutthereasonsforwhichweact.Thenextchapterconsiderswhetherornot
thedifficulties facedby these threeaccountsshouldpersuadeus that thereare justseveral,
irreconcilableconceptsofreasonatplay.
16That is, it concludes thatanagent’s reason foractingcan’texplain theiractionbyassuming it isanintentionalobject,butthisbegsthequestion,whichis,inpart,aboutwhetherornottheagent’sreasonforactingisanintentionalobject.
80
(V)
Onthepluralityofreasons
InwhichIexplainwhatapluralisttheoryofreasonsisandwhy‘goingplural’isnotapanacea.Isuggestthatagivenreasonexpressioncouldhavemorethanone sense, and I show how we can accommodate theories of reasons thatacceptthatidea,i.e.pluralisttheoriesofreasons,inourcategorisationschema.I discuss some examples of pluralist theories from the literature. I show howpluralist theories can solve some of the problems discussed in the previouschapters. I explainwhypluralism is not, however, enough, and I suggest thatourinvestigationshouldgobeyondfavourism,psychologismanddeliberativism.
Thepreviousthreechapterssetoutsomeofthemainproblemsforthemostpopularclaims
from each of the three families of claims about reasons. In doing so they painted a bleak
picture of contemporary theories of reasons; as Dancy notes (paraphrasing Aristotle) the
theories ‘leaveone saying things thatnobodywould sayunlessdefendinga theory.’ (2008a,
267)Sowhatistobedone?Onepossiblesolution,whichIwishtoconsidernowonlysothat
wemaysetitaside,istothinkthattherearedifferentsensesofagivenreasonexpression.
UpuntilnowallofthetheoriesIhaveconsideredhavebeenunivocal;thatis,theyhave(asI
noted in §(I)1) all assumed that a given reason expression always picks out reasons of the
same kind. Homonyms, I said, are the exception and not the rule. But what if reason
expressions are homonyms of some sort, picking out kinds of things that are confusingly
similar, but nonetheless distinct? Perhaps, that is, there is a sense inwhich Sally runs for a
reason and a sense inwhich shedoesn’t run for a reason (or runs for no reason). Similarly,
perhaps ifthere isnomilkathomebut Ibelievethatthere is,thenthere isasense inwhich
thereisnoreasonformetobuymilkandasenseinwhichthereisareasonformetobuymilk.
Perhaps, the same reason expression can have different senses; perhaps, that is, the same
reasonexpressioncanbeusedtopickoutdifferentkindsofreason.
A theory that admits that a single reason expression can have different senses is apluralist
theoryofreasons.Thepurposeofthischapteristoexplainwhatpluralisttheoriesofreasons
are,whyonewouldadoptpluralism,andtoshowthateven ifweadoptapluralisttheoryof
reasons, we should look for a new account of the reason-relation, beyond favourism,
deliberativismandpsychologism.
81
1 Thesenseofanexpression
ForexpositionalpurposesIwanttostressthedistinctionswearenowworkingwith:thereare
(i) different reason expressions, (ii) different kinds of reason they pick out and, now (iii)
differentsensesofareasonexpression.
In§(I)1,Idiscussed,atlength,thepossibilitythatdifferentreasonexpressionsmightpickout
differentkindsofreason.However,Iassumedthateachreasonexpressionpicksoutonlyone
kind of reason.Now, I am allowing that the same reason expressionmay pick outdifferent
kindsof reason–and to theextent that it does,we can say that the reasonexpressionhas
differentsenses.Thatis,asinglereasonexpressionhasdifferentsensesifandonlyifitcanbe
usedtopickoutdifferentkindsofreason(andtwokindsofreasonaredifferentifandonlyif
theconditionsforbeingareasonofeachkinddiffer).
Afurther,important,clarificationistonotethatsayingthatareasonexpressionhasdifferent
sensesisnotthesameassayingthatitpicksouta‘disjunctive’kindofreason.If,forinstance,
‘areasonthereistoact’justpicksoutonekindofreason,whichhappenstobedisjunctive(in
thesensethattheconditionsforbeingareasonofthatkindaredisjunctive1),thentherecould
notbeasenseinwhichsomethingisandasenseinwhichitisn’tareasontoact–thereisjust
one sense of that reason expression, even though the conditions for its application are
disjunctive.Incontrastifasinglereasonexpressioncanbeusedtopickouttwodifferentkinds
ofreason,therearetwosensestothatexpression,andthis issoevenifbothofthekindsof
reasonpickedoutarenon-disjunctive.2
2 Expandingthecategorisationschema
Toaccountforthepossibilityofapluralityofsensesforanygivenreasonexpression,weneed
toexpandthecategorisationschemasetoutinTableI-3(and,indeed,expandingtheschema
inthiswaycanbeusedtofurtherexplicatewhatwemeanbytherebeingdifferentsenses).1I haven’t considered any disjunctive conditions for being a reason (I amnot convinced that anyoneholdswhatIwouldcalladisjunctivetheory).However,wewouldhaveadisjunctiveclaimaboutreasonstoact if,for instance,wesaid: ‘p isareasonforAtoφ ifandonlyifeitherpmakesA’sφing, insomerespect,worth doing or A takesp tomake A’s φing, in some respect,worth doing.’ Thiswould be adisjunctive, but nonetheless univocal, account of the expression ‘a reason there is to act’ (it isdisjunctivebetweenthefavouristanddeliberativistconditions).2Intheeventthatthisisnotclear,considerthefollowing:thereisasenseoftheword‘bat’accordingtowhichthatwhichahitterinabaseballgameusesisa‘bat’,andasenseinwhichitisn’ta‘bat’(it’snotawingedmammal). The expression ‘a bat’ has two senses, each picking out a non-disjunctive kind ofthing.Incontrast‘(sporting)bat∨(animal)bat’isadisjunctiveexpression(youcouldsaythatitpicksouta‘disjunctivekindofthing’ifyoubelieveindisjunctivekinds)thathasonlyonesense–it’snotthecasethatsomethingcouldbeboth‘(sporting)bat∨(animal)bat’andnot‘(sporting)bat∨(animal)bat’(thatis, if something is either a sportingbator ananimalbat then is a ‘(sporting)bat∨ (animal)bat’).Mypointisjustthatbeingdisjunctiveisunrelatedtohavingseveralsenses.
82
TableV-1 isacategorisationschemathatallowsfortwosensesofagivenreasonexpression
andtherebyprovidesawaytorepresentbothunivocalandpluralist3theoriesofreasons.
Reasonexpression SenseA SenseB
Foranyp,pisareasonforAtoφ… Claim Claim
Foranyp,pisareasonforA’sφing… Claim Claim
Foranyp,pisareasonAhastoφ… Claim Claim
Foranyp,pisA’sreasonforφing… Claim Claim
TableV-1:Acategorisationschemathataccommodatespluralisttheoriesofreasons
Recall that a reason expression only picks out different kinds of reason if the conditions
betweendifferentsensesoftheexpressiondiffer(iftheydon’tthenthe‘different’sensesboth
pickoutthesamekindsofreason,inwhichcasethereisreallyonlyonesense).So,aunivocal
theory is represented inTableV-1byproviding the same conditionsunderboth senses. For
instance, the ‘Received View’, which is a univocal theory that is classified in the original
schemaas(F,F,F,P),wouldbeasfollowsinthisnewschema:
Reasonexpression SenseA SenseB
Reasonstherearetoact Favourism Favourism
Reasonsforacting Favourism Favourism
Reasonsonehastoact Favourism Favourism
Reasonsforwhichoneacts Psychologism Psychologism
TableV-2:The,univocal,'ReceivedView'representedinthenewschema
Thereisonlyonesensetoeachexpressioninthe‘ReceivedView’,hencetheclaimsabouteach
reasonundereach ‘sense’ in the categorisation schemaare the same.Wecanenrichour4-
tuple descriptions to represent the possibility of theories with multiple senses of a given
reasonexpressionbyintroducinga‘/’todenotealternatesenses.So:(F,F,F,P)≡(F/F,F/F,F/F,
P/P).
Now, in contrast to thisunivocal theory, apluralist theoryof reasons isany theory that, for
some reason expression, makes a different claim under each sense. Here is an example
pluralisttheoryrepresentedusingthisschema:
3Oratleast‘dual’sensetheories–Isupposeit’spossiblethatanexpressioncouldhavemorethantwosenses,butIdon’tconsiderthathere.
83
Reasonexpression SenseA SenseB
Reasonstherearetoact Favourism Deliberativism
Reasonsforacting Favourism Favourism
Reasonsonehastoact Favourism Favourism
Reasonsforwhichoneacts Favourism Favourism
TableV-3:Anexamplepluralisttheory
The4-tupledescriptionofthistheoryis(F/D,F/F,F/F,F/F).Thisisnot,tomyknowledge,one
that anyone advocates – I use it only to indicate what a pluralist theory looks like in this
schema.Thistheoryispluralistwithrespecttoreasonstherearetoact;ittakes‘areasonthere
is to act’ to pick out either a fact that makes the act, in some respect, worth doing, or
somethingthattheagenttooktomaketheact,insomerespect,worthdoing.Butitisunivocal
withrespecttoalltheotherreasonexpressions.
InwhatfollowsIwanttobrieflyrepresent(whatItaketobe)thetwomain‘pluralist’theories
ofreasonsthathavebeenconsideredintheliteraturetodate.
3 Favourist/Deliberativist(F/D)pluralism
3.1 Objectiveandsubjectivereasons
AnincreasinglycommonresponsetocaseslikeSally’smistakeaboutthebearormyignorance
aboutmy lackofmilk is todistinguishbetweenobjectiveandsubjectivekindsofreason(e.g.
Stoutland2007;Schroeder2007;Markovits2011;Vogelstein2012;Whiting2014).Anobjective
reason issomethingthat (inmyparlance)makesone’saction, insomerespect,worthdoing,
whereasasubjectivereasonissomethingthat(again,inmyparlance)theagenttooktomake
theiraction,insomerespect,worthdoing.4
So, to give some examples, what Sally takes to make running worth doing (that a bear is
chasingher) isasubjective reason forher to runbutnotanobjective reason forher to run.
And, in contrast, the fact that I am out of milk is an objective reason for me to buymore
(because it makes it, in some respect, worth doing) but not a subjective reason (because I
believethatIhaveplenty–Idon’ttakeanythingtomakebuyingmilk,insomerespect,worth
doing).If,ontheotherhand,itisrainingandIbelievethatitisrainingthenthefactthatitis
raining is both an objective reason and a subjective reason for me to take my umbrella
4An alternative vernacular for the subjective/objective reasons distinction is talk of first-person andthird-personreasons(respectively).
84
(becauseitbothmakestakinganumbrella,insomerespect,worthdoingandIbelievethatit
does).5
It should perhaps be clear that, given these definitions, an objective reason is the kind of
reason that favourists take reason expressions to pick out (i.e. the sort of thing thatmakes
actions, in some respect, worth doing), and a subjective reason is the kind of reason that
deliberativists take reason expressions to pick out (i.e. the sort of thing that agents take to
maketheiractions,insomerespect,worthdoing).
3.2 Apluralisttheory
Sofarthedistinctionbetweenobjectiveandsubjectivereasonsismerelyterminological.One
couldmakethisdistinctionandyetretainaunivocaltheoryofreasonsifoneweretosaythat
no single reason expression can be used to pick out either objective or subjective reasons.6
However, someexplicitly invoke thedistinctionbetweenobjectiveandsubjective reasonsso
as tooffer apluralist theoryof reasons. For instance,Vogelstein saysofParfit’s(2011)well-
known‘snake’example7:
Thereseemstobeasenseinwhichthereisareasonforyoutorunaway(sinceyoubelievethatrunningawaywillsaveyourlife),andasenseinwhichthereisnoreasonforyoutorunaway(sincenogoodwillcomeofit).Thatis,thereisasubjectivereason,butnoobjectivereason,foryoutorunaway.Likewise,thereisasenseinwhichthereisareasonforyoutostandstill(sinceitwillsaveyour life),andasense inwhichthere isnoreasonforyoutostandstill (sinceyoubelievenothingtosuggestthatanygoodwillcomeofit).Thatis,thereisanobjectivereason,butnosubjectivereason,foryoutostandstill.(Vogelstein2012,241)
Vogelsteinissayingisthattherearetwosensestobeingareasonthereistoact.Onesense,he
suggests,correspondstoobjectivereasons,theothercorrespondingtosubjectivereasons.So,
sincehethinksoneofthereasonexpressionshastwosenses,heisofferingwhatIhavecalled
a‘pluralist’theoryofreasons.Andaccordingtohispluralisttheoryofreasonsonesenseofthe
expression,‘areasonthereistoact’isfavouristandtheothersenseisdeliberativist.
5If one assumes that intentional objects are propositions and that true propositions are facts –otherwise the ontology of subjective and objective reasons is different, so one and the same thingcannotbebothanobjectiveandasubjectivereason.Thatbeingso,inthisexampleIwouldstillhaveanobjective and a subjective reason, they would just be different things on account of the ontologicaldifferencebetweenthesekindsofreason.6Forinstance,IthinkthatSchroeder(2008)invokestheobjectiveandsubjectivereasondistinctiononlytosaythattheexpression‘areasonthereistoact’picksoutobjectivereasons,whereastheexpression‘areasononehastoact’picksoutsubjectivereasons(thisis,Ithink,alsoDancy’s(2012)interpretationof Schroeder’s view), so his theory is not pluralist, it just distinguishes between the kinds of reasonpicked out by these expressions (which are normally taken to be co-extensive (see §(I)1)). For theclassificationofSchroeder’sviewseeTableI-6.7Here, for reference, is Vogelstein’s version of this example: ‘While walking in a desert, you haveangeredapoisonous snake. Youbelieve that runningawaywill save your life, andbelievenothing tosuggestotherwise.Asitturnsout,however,youmuststandstillinordertosaveyourlife,asthissnakewillattackmovingtargets.’(Vogelstein2012,241)
85
Vogelstein is not alone. Although they don’t talk in terms of ‘objective’ and ‘subjective’
reasons,bothHyman(2011)andLocke(2015)offerapluralistaccountoftheexpression‘the
agent’sreasonforacting,’ofpreciselythiskind.Theybothhold(inmyparlance)thatthereisa
sense of ‘the agent’s reason for acting’ that picks out a consideration in light of which the
agentacted(that is,asubjective reasonthattheyactedon),andadifferentsensethatpicks
out a fact thatmakes their action,all things considered,worth doing, and explains it in the
rightway(that is,anobjectivereasonthatexplainstheiractionintherightway).8Thus,both
Hyman and Locke think that one of the senses of being a reason for which an agent acts
correspondstofavourismaboutthereasonsforwhichweactandanothersensecorresponds
todeliberativismaboutthereasonsforwhichweact.
Sotherearepluralisttheoriesthatmixfavourismanddeliberativism.Idon’twant(orneed)to
categorisethemall,so,forthesakeofargumentlet’sjustcharacteriseapluralisttheoryofthis
kindaspluralistabouteveryreasonexpression(i.e.(F/D,F/D,F/D,F/D)):9
Reasonexpression SenseA SenseB
Reasonstherearetoact Favourism Deliberativism
Reasonsforacting Favourism Deliberativism
Reasonsonehastoact Favourism Deliberativism
Reasonsforwhichoneacts Favourism Deliberativism
TableV-4:PureF/Dpluralism
4 Favourist/Psychologist(F/P)pluralism
Michael Smith (1987,1994) is typically taken toholdwhat Ihave called the ‘ReceivedView’
(see Table V-2), which is a combination of, inter alia, favourism about reasons to act and
psychologismaboutthereasonsforwhichweact.IthinkthatthisisamisreadingofSmith.In
what follows I want, briefly, to make the case that Smith argues for a pluralist theory of
reasons.
4.1 ThemisinterpretationofMichaelSmith
Firstly, although Smith’s definition of ‘normative reasons’ in these works is loose,10we can
plausibly treatwhat he refers to as ‘normative reasons’ aswhatwe have already called an
8Although it isperhapsworthnoting that theydiffer in theirviewsofwhat it takes forsucha fact toexplainanactionintherightway(Hymanthinksitisknowledge,Lockethinksitisanexplanatorychain).9Sometheoriesmaynotbepluralistabouteveryexpression,inthewaythatthisoneis–butthatisn’tveryimportanthere.10(SeeSmith1987,39)
86
‘objective reason’ (that is, something thatmakes an action, in some respect, worth doing),
which is, you will recall, what favourists take reasons to be. Smith’s usage of the term
‘normativereason’isthusconsistentwithitscontemporaryusage.11
In contrast, Smith’s usage of the term ‘motivating reason’ is at odds with contemporary
usage.12 Smith argues thatmotivating reasons are all psychological states of the agent that
rationalise their action (that is,motivating reasons are the sorts of thing that psychologism
takesreasonstobe),andthisisbecausehedefinesmotivatingreasonssuchthat:
Thedistinctivefeatureofamotivatingreasontoφisthatinvirtueofhavingsuchareasonanagent is inastatethat ispotentiallyexplanatoryofhisφing. (Notethe‘potentially’.Anagentmay thereforehaveamotivating reason toφwithout that reason'sbeingoverriding.) (Smith1987,38emphasisinoriginal)
It is typical to interpret Smith’s remarks aboutmotivating reasons to be about ‘the agent’s
reason foracting’,andhence to takeSmith tobearguing foraunivocalpsychologismabout
theagent’sreasonforacting13–this,Ithink,isamistake.Forone,thefactthathetalksabout
motivating reasons being only potentially explanatory should already tell us that he isn’t
talkingabouttheagent’sreasonforacting–theagent’sreasonforactingissomethingthatis
generallytakentobe(andSmithcertainlytakesittobe)actuallyexplanatory(see§(IV)1.2).
Furthermore, thefact thatSmithtalksaboutmotivatingreasonstoφ shouldbe indicativeof
thefactthatthekindofreasonheistalkingaboutissuchthatreasonsofthatkindare(inthe
senseintroducedin§(I)1) independentoftheactionsforwhichtheyarereasons.Thatis,the
‘to’ prepositionmakes clear that something could be amotivating reason to do something
even ifonedoesnotdo it.Now,sinceweknow14that reasonsof thekindpickedoutbythe
expression‘theagent’sreasonforacting’,aredependentontheactionstheyarereasonsfor,15
whenSmithrefersto‘motivatingreasonstoφ’hecan’tbetalkingabouttheagent’sreasonfor
acting.
Thus, contrary to the typical interpretation,when Smith talks about ‘motivating reasons’ he
isn’tnecessarilytalkingaboutthosereasonspickedoutbytheexpression‘theagent’sreason
foracting’.So,whenhesaysthatmotivatingreasonsarefeaturesoftheagent’spsychology,he
islikewisenotnecessarilyadvocatingpsychologismaboutthereasonsforwhichweact.Ithink
thatinsayingwhathedoes,Smithisn’tmakingaclaimaboutanyparticularreasonexpression
–heisjustdefiningtheterm‘motivatingreason’.11See§(I)4.1foradiscussionofthecontemporaryusageof‘normative’and‘motivating’reasons.12ThisapointthatbothDarwall(2003,442–43)andSetiya(2007,30)make.13Whichis,indeed,whyheisassociatedwiththeReceivedView.14See§(I)1.5.15Inthesensethatitsreason-hooddependsontheoccurrenceoftheactionforwhichitisareason.
87
TherightwaytounderstandSmith’susageoftheterms‘normative’and‘motivating’reasons
is, I suggest, as different senses of any given reason expression. For instance, I take such a
pluralistconceptiontobethemostnaturalwaytoconstruethefollowingremarks:
TheclaimthatAhasareasontoφisambiguous.ItmaybeaclaimaboutamotivatingreasonthatAhasoraclaimaboutanormativereasonthatAhas.(Smith1987,38emphasisinoriginal)
Smith is saying that theexpression ‘a reasonAhas toφ’ isambiguousbetween twosenses:
being a normative reason (which is the favourist kind of reason, on Smith’s definition) and
beingamotivatingreason(whichisapsychologistickindofreason,onSmith’sdefinition).That
is, Smith is recommending a pluralist account of that expression, and therefore, a pluralist
theoryofreasons.Moreover,Ithink,Smith’spluralismextendstoallthereasonexpressions.16
4.2 Anotherkindofpluralism
Thathavingbeensaid,Smith’sactualtheory is largely irrelevanttoourprimaryconcern.The
remarks above aremainly intendedas context (if somewhatpolemical); all thatweneed to
take from this discussion is that the materials for another pluralist theory of reasons are
alreadyoutthere–onewhichtakesallreasonexpressionstohavetwosenses,onefavourist
and the other psychologist (i.e. (F/P, F/P, F/P, F/P)).We can represent such a theory in our
revisedschemaasfollows:
Reasonexpression SenseA SenseB
Reasonstherearetoact Favourism Psychologism
Reasonsforacting Favourism Psychologism
Reasonsonehastoact Favourism Psychologism
Reasonsforwhichoneacts Favourism Psychologism
TableV-5:PureF/Ppluralism
5 Whybeapluralist?
Wehave considered twodistinctpluralist theoriesof reasons,which I have called ‘pureF/D
pluralism’and‘pureF/Ppluralism’.Butwhywouldyouwanttobeapluralistinthefirstplace?
16Considerhisuseof theexpression ‘theagent’snormative reason for acting.’ (Smith1994,131–32).Consider also the following: ‘The distinction is that between psychological states that teleologicallyexplain[i.e.motivatingreasons]andconsiderationsthatjustify[i.e.normativereasons].Theimportanceofmakingthisdistinctioninthiswaybecomesclearwhenweaskwhetherallactionsmustbedoneforreasons. For though this question gets answered resoundingly in the affirmative when reasons areunderstoodtobemotivatingreasons…thequestiongetsansweredjustasresoundinglyinthenegativewhenreasonsareunderstoodtobenormativereasons.’(Smith2004,174–75)
88
5.1 Pluralismrespectsthe‘twosenses’intuition
Well,ifitstrikesyou(asitstrikesme)thatitistruethatthereisasenseinwhichSallyrunsfor
areasonandasenseinwhichshedoesn’t,orthatthereisasenseinwhichIhaveareasonto
buymilkandasenseinwhichIdon’t,thenyouhavealreadysetthestageforapluraltheoryof
reasons.Indeed,itseemstome,ifyouwantyourtheoryofreasonstorespectthis‘twosenses’
intuition (which you might, if it is an intuition you share) then you have to adopt a plural
theoryofreasons.
5.2 Pluralismmaysolvetheproblemsthatunivocaltheoriesface
Asecondappealing featureofpluralist theoriesof reasons is that they seemtoprovideone
withameansof solvingmanyof theproblems facedby theunivocalclaimsdiscussed in the
previouschapters.
Forinstance,considerTheDeliberateActionProblemdiscussedin§(II)4.2,whichconsistedof
thefollowingsetofmutuallyinconsistentclaims:
TheDeliberateActionProblem
(F4) Sallyrandeliberately.
(F3) IfAφsdeliberatelythenAφsforareason.
(F7) IfAφsforareasonthentherewasareason,p,forAtoφ.
(FAV) Foranyp,p isareasonforAtoφonly ifA’sφing, is insomerespect,worthdoing.
(F10) Sally’srunningwasnotatallworthdoing.
Accordingtoapluralisttheoryofreasons,thisproblemneedsre-formulatinginamannerthat
brings to lightwhich sense of each reason expression the claim is being made about. So,
denotingthedifferentsensesofeachreasonexpressionas‘reasonA’(favourist)and‘reasonB’,
(eitherdeliberativistorpsychologist –dependingonone’s theory),we can re-formulate this
problemasthefollowingsetofmutuallyinconsistentclaims:
89
Are-formulationofTheDeliberateActionProblem17
(F4) Sallyrandeliberately.
(F3’) IfAφsdeliberatelythenAφsforareasonB.
(UNI) AφsforareasonBifandonlyifAφsforareasonA.
(F7’) IfAφsforareasonAthentherewasareasonA,p,forAtoφ.
(FAV’) Foranyp,p isareasonAforAtoφonlyifA’sφing,isinsomerespect,worthdoing.
(F10) Sally’srunningwasnotatallworthdoing.
By adding (UNI), this re-formulation makes a premise of the problem that is implicit in its
original formulation,explicit: that ‘the reason forwhichoneacts’ isaunivocalexpression.A
pluralisttheoryofreasonsthensolvesTheDeliberateActionProblembyrejecting(UNI),which
thusavoidstheneedtorejecttheotherprimafaciereasonableclaims.
And, indeed, pluralist theories of reasons can solve most of the problems considered in
previous chapters in a similarmanner. However, they don’t solve all of them, and pluralist
theoriesfacenewproblemsoftheirown,asthenextsectionswillshow.
6 Pluralismisnopanacea
SinceIsharethe‘twosenses’intuition,Ithinkthatoureventualtheoryofreasonsoughttobe
pluralist.However,asIwillargueinthissection,justadoptingapluralisttheoryofreasonsis
notenoughtosolvetheproblemsconsideredinthepreviouschapters.Firstly, ‘conventional’
pluralist theories (i.e. thosemade up of only favourist, psychologist or deliberativist claims)
cannotsolveallof theproblemsconsidered inpreviouschapters.Andsecondly, thepluralist
solutiontoanygivenproblemreliesontherebeinganimplicitunivocalityassumptioninthat
problem – but since it is not at all clear that all of the problems considered in previous
chapters implicitly include such an assumption, it is not at all clear that pluralism really
providesuswithasolution.
17Thisisonlymeanttobeanindicativeexampleofhowthepluralistmightrespond;inparticular,thereare other ways to formulate this problem that change which reason expression the univocalityassumption concerns. For example, compare (UNI) with (UNI*) in the following, alternative re-formulationofTheDeliberateActionProblem:
(F4) Sallyrandeliberately.(F3’) IfAφsdeliberatelythenAφsforareasonB(F7*) IfAφsforareasonBthentherewasareasonB,p,forAtoφ.(UNI*) Foranyp,pisareasonBforAtoφifandonlyifpisareasonAforAtoφ.(FAV’) Foranyp,pisareasonAforAtoφonlyifA’sφing,isinsomerespect,worthdoing.(F10) Sally’srunningwasnotatallworthdoing.
90
6.1 Conventionalpluralismcannotsolvealltheproblems
There is a problem that a pluralist theory that only consists of favourist, psychologist or
deliberativist claims,18i.e. a conventional pluralist theory, cannot solve – The Experiential
ReasonsProblem.
Recallthatfavourism,psychologismanddeliberativismallfacesomeformofTheExperiential
ReasonsProblembecausenoneofthemarecompatiblewiththeprimafaciereasonableclaims
that(i)Sally’shearingabear-likesoundcouldbeherreasonforrunning;or(ii)thatmyreading
thatmyfriendhadwonawardcouldbemyreasonforcongratulatingher.Thatbeingso,any
conventionalpluralisttheorywillalsofaceTheExperientialReasonsProblem.
Furthermore,itisworthnotingthatsincebothfavourismanddeliberativismfacesomeformof
The Psychological Reasons Problem, any pluralist theory that only consists of favourist or
deliberativist claims (such as pure F/D pluralism), will also face The Psychological Reasons
Problem.
The point of these remarks is this: even if we adopt a (conventional) pluralist theory of
reasons,thatstillwillnothelpussolvealltheproblemsconsideredinthepreviouschapters.If
wearetodothat,weneedanewfamilyofclaimsaboutreasons.
6.2 Justbecauseyoucoulddoesn’tmeanyoucan
Secondly, we should be sceptical about the pluralist’s approach to solving the problems
considered; as Dancy puts it: ‘one cannot resolve philosophical puzzlement in this way by
multiplicationofsenses.’(2011,351)
OnewaytointerpretDancy’sremarkistheinsistencethatyoucan’tsolvetheseproblemsby
justpostulatingdifferentsensesthataren’tactually ‘there’.However,asDustinLockerightly
pointsout,thisisnotwhatthepluralistintends:
Iamnotsuggestingthatwecanresolvephilosophicalpuzzlementbymultiplicationofsenses.Rather, I am claiming that senses are alreadymultiple—the phrase ‘S’s reason [for acting]’alreadyhastwodistinctsenses.(Locke2015,218)
Tore-interpretLocke’sargumentinmyownterms:ifyouhavethe‘twosenses’intuitionit is
that intuition thatmakesyou think that reasonsareplural,which then forms thebasis fora
pluraltheoryofreasons.Thatis, it’snotthatyouaremultiplyingsensesinordertosolvethe
problems,it’sthat(atleastifyouhavethe‘twosenses’intuition)thesensesjustaremultiple,
andonceyouacknowledge that,manyof theproblems that aunivocal theory faces can fall
18Asopposedtotheclaimsofsomenewfamily,notyetconsidered.
91
away. Ifthe‘twosenses’ intuitionisright,then,thepluralistcaninsistthatourphilosophical
puzzlement has arisen only because we haven’t recognised that the senses already are
multiple.
However, a more nuanced interpretation of Dancy’s objection is harder for the pluralist to
avoid.IthinkthatDancycanreasonablybeinterpretedashavingmeantthatevenifthesenses
aremultiple, the fact that invokingmultiple sensesof the relevant reasonexpressionscould
solvemanyoftheproblemsconsideredinthepreviouschaptersshouldnotmakeusthinkthat
theycanbesolvedthatway(i.e.thatthatisthecorrectsolutiontothem).
Itmightbethattheprimafaciereasonableclaimsinanygivenproblemarereallyrestrictedto
asinglesenseofthereasonexpressionsinvolved–inwhichcasetheassumptionofunivocality
doesnowork,andtheproblemreturns.That is, for instance,TheDeliberateActionProblem
mayactuallybelikethis:
Anotherre-formulationofTheDeliberateActionProblem
(F4) Sallyrandeliberately.
(F3*) IfAφsdeliberatelythenAφsforareasonA.
(UNI) AφsforareasonBifandonlyifAφsforareasonA.
(F7’) IfAφsforareasonAthentherewasareasonA,p,forAtoφ.
(FAV’) Foranyp,p isareasonAforAtoφonlyifA’sφing,isinsomerespect,worthdoing.
(F10) Sally’srunningwasnotatallworthdoing.
Inthisconstrual,becausealltheprimafaciereasonableclaimsareallaboutthesamesenseof
reason, ‘reasonA’, theunivocalityassumptiondoesnowork.So,apluralist theoryof reasons
wouldthusstillhavetofindsomeclaim,inadditionto(UNI),toreject–theproblemreturns.
Apluralistsolutiontoanygivenproblemreliesonthatproblembeingtheresultofaconflation
of different senses of the same reason expression – that is, it relies on a univocality
assumptionbeingapartofeveryproblem.Thedifficultyforpluralistsolutionstotheproblems
considered is that it seemsquiteplausible thatat leastsomeof theproblemsconsidereddo
notresultfromaconflationofsenses,and,inthatcasepluralismisnohelpinsolvingthem.
So,again,itseemsthatifwedowanttosolvealloftheproblemsconsideredintheprevious
chapters,weneedanewfamilyofclaimsaboutreasons.
92
7 Achallengeforpluralism
It is generally19agreed that whenever we give an agent’s reason for acting we explain the
agent’s action in a way that makes them seem rational.20For instance, saying that Sally’s
reasonforrunningwasthatshebelievedthatabearwaschasingherexplainsthefactthatshe
raninawaythatmakesherseemrationalforrunning;and,likewise,sayingthatmyreasonfor
congratulatingmyfriendwasthatshehadwonanawardexplainswhyIcongratulatedherina
waythatmakesmeseemrationalforhavingdoneso.
Thisobservationisofsomerelevancetopluralismbecauseitisseeminglynotrestrictedtoany
particularsenseofthe ‘agent’sreasonforacting’expression.That is, itseemsasthough,we
mayaddafurtherprimafaciereasonableclaimtothosealreadyconsidered:
(S1) Whenever we give an agent’s reason for acting,whatever the sense of theexpression used, we explain their action in a way that makes them seemrational.
Whyisthisclaimproblematicforpluralism?Well,recallthefollowing:
(F5) Sally’sreasonforrunningwas,interalia,thatshebelievedabearwaschasingher.
(P1) My reason for congratulatingmy friendwas, interalia, that shehadwonanaward.
Only pure F/P pluralism21can accommodate the truth of both (F5) and (P1); it does so by
insisting that a different sense of the expression, ‘the agent’s reason for acting’, is being
invoked in each case (psychologistic in the former, favourist in the latter). This view is
problematicbecause, if itwere true, itwouldbehard toseewhyour respective reasons for
actingcanbothbecitedinanexplanationofouractionsthatmakesuseachseemrational.
Consider:accordingtopureF/Ppluralism,therelationsbetween(i)Sally’sbeliefandheraction
and (ii) the fact thatmy friend won an award andmy action are different. And yet, giving
eitherexplainsourrespectiveactions inawaythatmakesusseemrational.SothepureF/P
pluralist is seemingly forced to say that in spite of the reason-relations being different, by
incredible coincidence, both Sally’s andmy reason for acting end up standing in a common
relationtoourrespectiveactions–therelationinvirtueofwhichgivingtheagent’sreasonfor
actingexplainstheiraction inawaythatmakesthemseemrational.PureF/Ppluralismthus
19(E.g.Dancy2000,8;Stout2009,53;Gibbons2010,343;Broome2013,47)20This is not to assume that it is the agent’s reason thatdoes theexplaining: one canhold this viewwithoutholdingthatitistheagent’sreasonforactingthatexplainstheiraction,as,forinstance,Dancy(2014)does.21Orless‘pure’variantsofit,suchas:(F/F,F/F,F/F,F/P).
93
facesanunenviabledilemma:eitheritacceptsthishighlyimplausiblecoincidence22oritrejects
thisnewprimafaciereasonableclaim.23
ThisisaparticularproblemforF/Ppluralism,butitisalsoaconstraintonanypluralisttheory
ofreasons.That is,anypluralisttheoryofreasonswillhavetoprovidesomeaccountofhow
(S1)couldbetrue,orsuffertheconsequencesofrejectingit. Iwillreturntothispoint inthe
finalchapter,whenIsetoutmyownpluralisttheoryofreasons.
8 Conclusion
Ihavesuggestedthatwhilethe‘twosenses’intuitionmightprovideuswithsomemotivation
foradoptingapluralisttheoryofreasons,wecannotrelyonpluralismtosolvetheproblems
considered in thepreviouschapters. Instead, Ihaveargued,weneedanewfamilyofclaims
aboutreasons.Theaimoftheremainingchaptersistoadvanceandthendefendsuchafamily.
22The‘highlyimplausibleco-incidence’beingthatgivingeithersenseoftheexpression,independently,explainstheagent’sactioninawaythatmakesthemseemrational.23I.e.(S1).
94
(VI)
Anewfamilyofclaimsaboutreasons
InwhichIsetoutanewfamilyofclaimsaboutreasons,andintroducethemajorchallenge to it. I define ‘pro tanto rational’ actions as actions that an agenttakestobe,insomerespect,worthdoing.Isetoutanewfamilyofclaimsaboutreasons,explanatory rationalism,which says thatall practical reasonsexplainwhy theactions forwhich theyare reasonsarepro tanto rational. I introducethe major challenge for explanatory rationalism, The Explanatory ExclusionProblem, which argues that only features of an agent’s psychology couldexplaineitherwhytheydosomethingorwhyitwasrationalforthemtodoit.Isetouttheprogramfortheforthcomingchapters.
In§(I), Inotedthatmosttheoriesofreasonssubscribetoone,ifnotseveralofthefollowing
claims: favourism about reasons to act, psychologism about the reasons for which we act
and/or deliberativism about the reasons for which we act (see Table I-6). In §§(II)-(IV), I
showedthateachoftheseaccountsisinconsistentwithseveralprimafaciereasonableclaims.
In§(V),Iarguedthat,evenifthesensesofanygivenreasonexpressionareplural,pluralismis
nopanacea: that is,wecannot just relyonthepluralityofsensesas thewaytomakethose
prima facie reasonable claims consistent with our theory of reasons. In short, I argued, we
mustlookbeyondfavourism,psychologismanddeliberativism:weneedanewfamilyofclaims
aboutreasons.
In this chapter I introduce a new family of claims about reasons: explanatory rationalism.
According to explanatory rationalism, the fundamental reason-relation is that of explaining
whyanaction isprotanto rational;which is tosaythatallpractical reasonsexplainwhythe
actionsforwhichtheyarereasonsareprotantorational.Now,sinceexplanatoryrationalism
rejectsfavourismaboutreasonstoact,psychologismaboutthereasonsforwhichweactand
deliberativismaboutthereasonsforwhichweact, itcansolvealloftheproblemssetout in
§§(II)-(IV). My aim, in the chapters that follow, is to set out and defend explanatory
rationalism.
InthischapterIdefinewhatitisforanactiontobeprotantorational,Idescribeexplanatory
rationalismandIsetoutthechallengesforit.
1 Protantorationalaction
Fevzi iswaiting toboarda flight to Japan. It’searly sohehad tomisshismorning swim.He
looks forlornlyoutof thewindow, yearning to go swimming.He couldabandonhis trip and
95
havehisswimbuthethinks itwouldbebetter tostayandboardthe flight.Nonetheless,he
stillthinksthatthereissomethingtobesaidforgoingswimming.WecansaythisaboutFevzi:
hethinksthatswimmingis,insomerespect,worthdoing,buthethinksthatboardingtheflight
isallthingsconsideredworthdoing.
Incontrast,Fevziseesnothingofworthinaimlesslywanderingaroundtheairport.Ifhewent
forawanderhe’dmisstheflightwithout,atleastbyhislights,anygoodcomingofit;hethinks
thatwanderingaimlesslyis,innorespect,worthdoing.
There are clear differences between these three actions (flying, swimming,wandering) that
are,Isubmit,ofrelevancetotheirrationalstanding.First,Iwillintroducesometerminologyto
characterisethosedifferencesandthenIwillarguethatwhatdifferentiatesthethreeactions
is relevant to their rational standing. In particular, I will suggest that even though it is not
rationalforFevzitogoswimming,itisnonethelessmorerationalforhimtogoswimmingthan
it is for him to wander aimlessly around the airport, and that we should have some
terminologythatreflectsthat.
1.1 Twokindsofrationalaction
TherationalthingforFevzitodoistoboardhisflight.Ifheweretogoswimmingwewouldsay
thatheactedirrationallysince,byhisownlights,itwasnotallthingsconsideredworthdoing.
Doingonethingwhenyoubelievesomethingelsetobeallthingsconsideredmoreworthdoing
isnotarationalthingtodo.
If we leave aside the question as to what makes an action worth doing,1I think we can
(hopefullyuncontroversially)characteriseafamiliarsenseofrationalactionasfollows:
Assumption It is rational forA toφ if and only if A takesφing to be, all thingsconsidered,worthdoing.2
This is an assumption, it is not a definition or an analysis of what it is for an action to be
rational,norisitaclaimaboutwhyanactionisrational(wewillcometothatshortly).Thisis
meanttobeafairlyblandassumptionaboutwhatobtainswhenitisrationalforsomeagentto
act; it leaves unspecified all of the details that would actually furnish us with a theory of
1Dependingonone’stheoryofrationalityor‘motivation’,sayingthatFevzitakesgettingontheplanetobe, insomerespect,worthdoingcouldbecashedoutbysayingthathebelievesthatbyboardingtheplanehewillgotoJapanandeitherthathewantstogotoJapanorthathejudgesthatgoingtoJapanwould be good, or right, or something of the sort. Nothing that I have to say ismeant to express acommitmenttoeitheroftheseviews.See§(II)3.2forrelatedcaveats.2For instance, this is, I think, in keeping with Parfit’s (2011, 34) characterisation of rational action.Comparealso:‘anagentisshowntobeactingrationallyif,aswemightputit,heisshowntobetryingtodowhatthereisgoodreasontodo,evenifasamatteroffactheisquitemistakenonthatfront.’(Dancy2004,33)
96
rationality. I state it here only so that wemay better understand the concept of pro tanto
rationality,whichIwillintroduceforthwith.
What separatesFevzi’sboarding the flight frombothhisgoing swimmingandhiswandering
aimlesslyaroundtheairport isthat it isrationalforhimtodotheformer,butnotthe latter.
However it isnonetheless common todistinguishacts likeFevzi’sgoing swimming fromacts
likehiswanderingaimlessly,inspiteofthefactthattheyarebothirrational.Whileitwouldbe
plainly irrationalofFevzi todothe latter, ifheweretogoswimmingthatwouldbeatypical
exampleofanakraticaction,orso-called‘weaknessofwill’.
Isuggestthatwhatseparatesakraticactsfrom‘plainly’irrationalonesisofrelevancetotheir
rationalstanding.IfFevziweretogoswimmingthatwouldbemarkedlymorerationalthanif
heweretowanderaimlesslyaroundtheairport;thatis,therewouldbemarkedlymorebyway
ofrationalintelligibilityinhisactionifheswamthanifhewandered–andthatissoevenifwe
concedethatgoingswimmingisnonethelessnotarationalthingtodo.
Now,ifwhatseparatesakraticbehaviourfromwhatIhavecalled‘plainlyirrational’behaviour
is of relevance to the rational standing of those actions, then, I suggest, we need an
intermediateconceptofrationalitythatallowsustorecognisetheformerassomehowmore
rationalthantheother.Tothatend,letusdefine‘protantorationalactions’thus:
Definition It isprotantorational forA toφ ifandonly ifA takesφing tobe, insomerespect,worthdoing.3
This isadefinition,notaclaim.4Ihaveusedtheterm‘rational’because(forthereasons just
outlined)Ibelievethatanaction’sbeingprotantorationalisofsomerelevancetotheaction’s
rationalstanding(aprotantorationalaction,is,Isubmit,inafamiliarsenseofthewordmore
rationalthananactionthatisnotevenprotantorational).Ifthereaderbalksatterminology
that says that it is to any extent rational for Fevzi to go swimming, thenplease feel free to
substitute some less objectionable term in its place (mutatis mutandis throughout this
discussion).5
Lastly, inordertomakethedistinctionbetweenrationalactionandprotanto rationalaction
clear,Isuggestthatwecanrefertotheformerasallthingsconsideredrationalaction,noting3DanielWhiting (2014,5)uses the terminologyofpro tanto rationalactionequivalently.Parfit (2011,34)alsoindicatesthedistinctionIhavesuggestedwhenhedistinguishesbetweenactionsthatare‘lessthanfullyrational’andactionsthatare‘irrational’.4Thatis,Iamjustdefiningtheuseofthetechnical‘protantorational’predicatehere.5Ibelievethataprotanto rationalaction isanactionthatonSmith’s(1987)terminologytheagent ismotivatedtodo(note:thestateofbeingmotivatedisdefeasible,onSmith’saccount),soIcouldhavetalkedintermsof ‘motivation’.However, Ihaveavoidedtheterminologyof ‘motivation’as Ithinkthedangersofmisunderstandingareevenmorepronouncedthere.
97
thatthisisthetypicalunderstandingof‘rational’action.So,forthesakeofclarity,hereishow
thetwoconceptsapplytoFevzi’spossibleactions:
Action Allthingsconsideredrational? Protantorational?
Boardinghisflight ü ü6
Goingswimming û ü
Wanderingaimlessly û û
TableVI-1:ThewaysinwhichFevzi’sactionsare(oraren’t)rational
Sayingthatgoingswimmingisprotantorationalwhereaswanderingaimlesslyisnotevenpro
tanto rationalallowsustorecognisethatthere isadifference intherationalstandingofthe
twoactionswithout impingingon the fact thatboarding the flight is theonly really rational
thingforFevzitodo.
2 ExplanatoryRationalism
ThefamilyofreasonclaimsIwanttointroducesaysthatreasonsofanykindexplainwhythe
actions for which they are reasons are pro tanto rational. Because this family of claims
emphasises theexplanatory characterof the reason-relation, andbecausea reasonexplains
why an action is rational, Iwill call this family of claims ‘explanatory rationalism’.Using the
schemadevelopedin§(I),wecanrepresentexplanatoryrationalismasfollows:
Reasonexpression Explanatoryrationalism
Foranyp,pisareasonforAtoφ… …ifandonlyifpexplainswhyitisprotantorationalforAtoφ.
Foranyp,pisareasonforA’sφing… …ifandonlyifpexplainswhyitisprotantorationalforAtoφandpmakesA’sφing,insomerespect,worthdoing.
Foranyp,pisareasonAhastoφ… …ifandonlyifpexplainswhyitisprotantorationalforAtoφ.
Foranyp,pisA’sreasonforφing… …ifandonlyifpexplainswhyitisprotantorationalforAtoφandexplains(intherightway)whyAφ’d.
TableVI-2:Explanatoryrationalism
According to explanatory rationalism, there are three kinds of reason: the expressions ‘a
reason forA toφ’ and ‘a reasonAhas toφ’pickoutonekind, theexpression ‘a reason for
φing’picksoutanother7and‘A’sreasonforφing’picksoutafinalkind.
6Anyall thingsconsideredrationalaction isautomaticallyaprotantorationalactionsince ifanagenttakes an action to beall things consideredworth doing they certainly take it to be, in some respect,worthdoing.
98
InlaterchaptersIwillarguethatexplanatoryrationalismavoidsalloftheproblemssetoutin
§§(II)-(V) and that, indeed, explanatory rationalismcharacterises thede facto senseofeach
reasonexpression.However, in linewiththetwosenses intuition, Iwill suggest thatthere is
anothersensetoeachreasonexpression,whichisfavourist.Beforewegetthere,though,Iwill
need todemonstratehow it is that explanatory rationalism couldbe true in the faceof the
majorobjectiontoit,whichIwillcall‘TheExplanatoryExclusionProblem’.
3 Aproblemforexplanatoryrationalism
Recall the following example from § (III):My friend haswon amuch-coveted award; I read
aboutitinanewspapersoIcallheruptocongratulateher.Isuggestedthatitwasprimafacie
reasonabletoclaimthatmyreasonsforcongratulatingmyfriendwere,interalia,thatshewon
anawardandthatIreadthatshehadwonanawardinthenewspaper.
Now,according toexplanatory rationalism,anagent’s reason foractingbothexplainswhy it
wasprotantorationalforthemtodowhattheydidandexplainswhytheydidit.Therefore,if
explanatory rationalism is to be consistent with these prima facie reasonable claims about
whatmyreasonsforcongratulatingmyfriendwere(asIintendittobe),thefollowingmustbe
true:
(R1) Icongratulatedmyfriendbecauseshehadwonanaward.
(R2) IcongratulatedmyfriendbecauseIreadthatshehadwonanaward.
(R3) Itwaspro tanto rational forme to congratulatemy friend because she hadwonanaward.
(R4) ItwasprotantorationalformetocongratulatemyfriendbecauseIreadthatshehadwonanaward.
There isa familiarargumentagainstthe ideathat factsabouttheexternalworldcanexplain
whyweact(which(R1)supposes),whichproceeds,broadlyasfollows:if,forinstance,Ibelieve
thatitisrainingwhenitisn’t,Iwillstilltakemyumbrella–becauseIbelievethatitisraining.
However,giventhatIneedtobelievethatitisraininginordertotakemyumbrella8,theneven
ifItakemyumbrellawhenitisraining,ImuststilltakemyumbrellabecauseIbelievethatitis
raining.But,ifmybeliefthatitisrainingcanexplainmyactionwhetheritistrueorfalse,then
what explanatory work can the fact that it is raining do? None, the argument concludes –
whichmustmeanthatonlyfeaturesofanagent’spsychologycanexplaintheiractions–that
7As a result, something couldbe a reason for an agent toφwithoutbeing a reason for theirφing. Idiscussthispointfurtherin§(XVI)A.3.8AssumingIseenothingelseofworthintakingmyumbrella.
99
is, (R1) is not literally true. The same argument applies to (R2), and, as I will show, can be
generalisedalsototheexplanationofwhyitis(protanto)rationalforanagenttoact;thatis,it
canbeusedtoshowthat(R3)and(R4)arealsofalse.
Thisargument,whichIcall‘TheExplanatoryExclusionProblem’,isthemotivatingargumentfor
psychologism(andpsychologismaboutthereasonsforwhichweact,inparticular)becauseit
insists that only features of an agent’s psychology can explain their actions, and therefore
(given that an agent’s reason for actingmust explain their action), that only features of an
agent’spsychologycouldbeamongsttheirreasonsforacting.
4 Anoutlineofwhatfollows
To the extent that the argument considered in the previous section is right, explanatory
rationalism cannot solve the problems considered in §§(II)-(IV) – indeed, to the extent that
that argument is right, explanatory rationalism just collapses into psychologism. The rest of
thisdiscussionisthus,forthemostpart,aresponsetotheargumentoftheprevioussection.
In §(VII), Imake some assumptions about the structural principles and logical properties of
explanatory relations, which I will use throughout my discussion. In §(VIII), I use this
framework to provide a formal construal of The Explanatory Exclusion Problem. In § (IX), I
show how the Problem also precludes perceptual experiences from explaining why we act,
thereby counting against (R2); and, further, how it applies to the explanation of why it is
rationaltoact,therebyalsocountingagainst(R3)and(R4).
Thede facto response toTheExplanatoryExclusionProblem is toaccept theconclusionand
say that thepurported explanans in (R1), i.e. the fact thatmy friend hadwon an award, is
merelyellipticalfortherealexplanans,whichisthefactthatIbelievedthatmyfriendhadwon
anaward.Becausethis isaclaimabouthowit isthatanormativereason9todosomeaction
could explain why someone does it, I call this ‘the elliptical theory of normative reason
explanation’.
An alternative and increasingly popular theory, which seeks to preserve the bona fide
explanatory role of normative reasons in action explanation, rejects the conclusion of The
ExplanatoryExclusionProblemandsaysthatthefactthatmyfriendwonanawardexplainsmy
actiondirectly.Thisisthedirecttheoryofnormativereasonexplanation.
In§(X), I setoutboth theellipticaland thedirect theoriesofnormative reasonexplanation,
andIarguethattheyareeachflawedinwaysthatshouldmakeuslookforanalternative.9Ire-habilitatethisterminologyin§(X)1,understandingnormativereasonsasthingsthatmakeactions,insomerespect,worthdoing;withoutassociatingthemwithanyparticularreasonexpression.
100
In §§(XI) & (XII), I develop that alternative: the indirect theory of normative reason
explanation. This theory argues that a normative reason explains an agent’s action by
explainingthefeaturesoftheagent’spsychologythat,inturn,explaintheiraction.
My argument for the indirect theory proceeds in two stages: first, in §(XI), I argue thatwe
should reject the conclusion of The Explanatory Exclusion Problembecause it is based on a
false principle of explanation, which I call ‘the exclusion principle’. The exclusion principle
requires thatonly themostproximal explanationsof someexplananda explain it;but this is
mistaken – most of the explanations we are interested in are, to some extent, distal
explanations.
Then, in§(XII), I showhow that insighthelps inform theaccountofhownormative reasons
explain actions. I argue that normative reasons are distal explanations of our actions; they
explainthosefeaturesofourpsychologythat,inturn,explainouractions.Ithenshowhowthe
indirecttheorycanbeusedtoshowthatboth(R1)and(R2)aretrue.
In §(XIII), I suggest that the same reasoning accounts for the truth of (R3) and (R4). For
instance,IarguethatthefactthatIreadthatmyfriendhadwonanawardexplainswhyit is
protantorationalformetocongratulateherbecauseitexplainswhyIbelievedthatshehad
wonanaward,which, in turn,explainswhy itwaspro tanto rational forme tocongratulate
her.
However,Inote,thetransitivityofexplanationfailsonsomeoccasions,inparticularwhenthe
explanatory chain is adeviant causal chain. So,weneed someaccountofwhy it is that the
explanatorychainuptothefactthatitisprotantorationalforanagenttoactistransitiveifit
isn’tdeviant,butisn’ttransitiveifitisdeviant.In§§(XIV)&(XV),Iprovidesuchanaccount.
First, in §(XIV), I introduce themystery relation. I argue that themystery relation is a non-
causal,transitive,explanatoryrelationthatrelates:thebeliefthatptosomejustificationforit
when that belief is justified; the belief that p to the fact that p when the belief that p is
knowledgeable; a justification for the belief thatp to the fact thatpwhen that justification
affords theopportunity for knowledge; and an action to somebelief that explainswhy that
actionisprotantorationalwhenthatactionisdoneintentionally.
In §(XV), I argue that the mystery relation is transitive with the non-causal explanatory
relationinvolvedinexplainingwhysomeactionisrational,whereasmerelycausalrelationsare
not. This allowsme to distinguish between deviant cases (which lack the required chain of
mystery relations) and non-deviant cases (which don’t). This leads me to argue: (i) that
becauseIknowthatmyfriendhaswonanaward,thefactthatshehaswonanawardexplains
101
whyitisprotantorationalformetocongratulateher(whichis(R3));and(ii)thatbecausemy
beliefthatshehadwonanawardisbasedonthefactthatIreadthatshehadwonanaward,
thefactthatIreadthatshehadwonanawardalsoexplainswhyitisprotantorationalforme
tocongratulateher(whichis(R4)).
Finally,in(XVI),Irevisitexplanatoryrationalismandshowhowitavoidstheproblemsfacedby
othertheories.Iconcludebysettingoutmypreferredtheoryofreasons,newpluralism,which
holds that explanatory rationalism tells us one sense of what it is to be a reason, whilst
favourismtellsustheother.Ithenshowhownewpluralismeasilyrespondstothechallenge
forpluralisttheoriesintroducedinthepreviouschapters.
102
(VII)
Weneedtotalkaboutexplanation
In which Imake some assumptions about explanation. I say what Imean by‘explains’ and I state that I will talk as though explananda are facts andexplanantiaarepropositions(whetherornottheyare).Idistinguishtwosortsofexplanatory relation, ‘fully explains’ and ‘partially explains’, where a fullexplanation is sufficient for the truthof the fact that it explainsandapartialexplanation is an element (or subset) of a full explanation, and Imake someassumptions about the logical properties of these relations. Lastly, I say thatsome fact is ‘overexplained’ just in case thereare twogenuinelydifferent fullexplanationsofthatfact.
InthischapterImakesomeassumptionsaboutthestructuralprinciplesandlogicalproperties
of explanatory relations. This iswith a view to having a technical frameworkwithwhich to
moreformallycharacteriseTheExplanatoryExclusionProblem,whichwillbediscussedinthe
next chapter. While I think that the assumptions below are intuitive and hopefully
uncontroversial,1IthinkthatneitherthethrustofTheExplanatoryExclusionProblemnorthe
effectivenessofmysolutiontoitdependsonformalisingexplanatoryrelationsinthismanner.
1 WhatdoImeanby‘explains’?
HereiswhatBroomehastosayaboutthemanymeaningsof‘explains’:
‘Explain’incommonusagehasvarioussenses.Inoneofthem,Darwinexplainedwhyevolutionoccurs. In another, The Origin of Species explains why evolution occurs. In a third, naturalselectionexplainswhyitoccurs.(Broome2013,48)
I,likeBroome,wishtosticktohisthirdsenseof‘explains’inthisdiscussion.Thatis,whenItalk
ofsomethingexplainingsomethingelse,thatwhichexplainsismeanttobetheexplanansand
notadescriptionofitortheonewhodescribesit.Similarly,whenItalkofsomethingbeingan
explanation2ofsomethingelse,Imeantosaythatitisanexplanansofthatthing.3
This‘explains’relation,sounderstood,is,Isuggest,thesamerelationastheonepickedoutby
the‘because’or‘reasonwhy’expressions,soIshallhereaftertakethemtobeequivalent.
1ItakeseveraloftheseassumptionsfromBroome(2013).2That said, in § (X), when I come to talk of normative reason explanation, I will mean somethingdifferent(andmoreakintothesecondsense).3Kim suggests that theremay be a fourth kind of ‘explains’ relation: ‘The explanans relation relatespropositions or statements; the explanatory relation relates events or facts in the world. Theexplanatory relation is an objective relation among events that, as we might say, “ground” theexplanansrelation,andconstitutesits“objectivecorrelate.”’(Kim1988,226)The‘explains’relationsasIwilluseitisthuswhatKimcalls‘theexplanansrelation’,andnotablynottheobjectiverelationthathetakestounderpinit(whathecalls‘theexplanatoryrelation’).
103
2 Ontologicalassumptions
2.1 Ontheontologyofexplananda
ThroughoutthisdiscussionIwillassume,purelyforexpositionalconvenience,thatexplananda
arefacts.Oneconsequenceofthisassumptionisthat,for instance,onecanexplainwhyit is
raining if it is raining, but one cannot explainwhy it is raining if it is not raining. A second
consequenceisthatwhenItalkofsomethingexplaining,forinstance,anactionorabeliefthis
isnotmeanttoimplyanythingabouttheontologyofthoseexplananda.
2.2 Ontheontologyofexplanantia
Throughout this discussion I will mostly treat explanantia as propositions (or sets of
propositions).4Importantly, I do not assume, in my exposition that explanantia are true
propositions: this is because I want to be able to recognise the possibility of non-factive
theorieswithinmydiscussion–althoughIwillultimatelydismissthem,becauseIdothinkthat
explanationisfactive.5
Explanatoryrationalismimpliesthatallpracticalreasonsstandinexplanatoryrelationstothe
rationalityoftheactionsforwhichtheyarereasons.Thus,whateversortofthingexplanantia
are,sotooarereasons.Thismeans,therefore,that Iwillbetreatingreasonsaspropositions
also.
However, although I will talk as though explanantia and reasons are propositions, I am not
arguing that theyare.My theory isneitheraboutwhatexplanantiaarenor is it aboutwhat
reasonsare; it isabouttherelation inwhichreasonsstandtotheactionsforwhichtheyare
reasons.So,ifyourpreferredtheoryofexplanationsaysthatmentalstatesorstatesofaffairs
canbeexplanantiathenitiscompatiblewithmytheorythatreasonscouldbementalstatesor
statesofaffairsalso;mypoint isnotthatreasonsarepropositions,mypoint isonlythat it is
the ontology of the relata of explanatory relations that determines the ontology of reasons
(becauseIthinkthereason-relationisultimatelyanexplanatoryrelation).6
4Again,whileImightoccasionallytalkabout(forinstance)beliefsexplainingthings,thatshouldnotbereadasimplyingthatitisthebelief(quamentalstate)doingtheexplaining,asopposedtothefactthattheagenthasthatbelief.5 Given the assumption that explanation is factive (i.e. all explanantia are true), and that truepropositions are facts, my treating explanantia as propositions, amounts to Broome’s (2013, 48)conventionoftreatingthemasfacts.6Thisisnotatrivialcaveat:somearestronglyoftheviewthatmentalstates,andnotfactsaboutthem,areanagent’sreasons(e.g.Turri2009),whilstothersareseeminglyoftheviewthatitisonlystatesofaffairs(andnotfacts)thathavethemetaphysical‘oomph’neededtoexplain(e.g.Dancy2000).
104
3 Fullexplanationandpartialexplanation
SupposethatJoanne’sroofleaksandthatitrainedlastnightandthathercarpetisnowwet.7
When I say that Joanne’s carpet is wet because her roof leaks and she says that it is wet
becauseitrainedlastnight,althoughweeachcitedifferentfactsbywayofexplanationofwhy
thecarpetiswet, itseemsclearthat,asBroomeputsit, ‘ourexplanationsarenotrivals,and
wewould not feelwewere contradicting each other.’(2013, 49) Furthermore, althoughwe
eachsaythatthefactsweciteexplainwhythecarpetiswet,neitherfactiswhatwemightcall
‘thewholestory’ofwhythecarpetiswet.
It seems natural to think that the reason why the explanations that Joanne and I give are
neither rivalsnor, individually, thewholestoryofwhythecarpet iswet, isbecausetheyare
eachpartof,whatBroomecalls,‘onebigexplanation’–wherethatonebigexplanationisthe
whole story ofwhy the carpet iswet. Supplementing this idea, Broome suggests thatwhen
givinganexplanationwetypicallypick justsomepartofthisbigexplanationandthat ‘which
partwepickoutwilldependonourcontext:ourbackgroundknowledge,ourinterestsinthe
matterandsoon.’(2013,49)
I want to further regiment Broome’s suggestion. I suggest that we call this ‘one big
explanation’a ‘fullexplanation’,and theelements (or subsets)of it ‘partialexplanations’, so
that all explanatory relations are either full or partial. In the following sections I will make
some assumptions about the logical properties of the ‘fully explains’ and ‘partially explains’
relations.8
3.1 Fullexplanations
Wecouldthinkoffullexplanationsascomplexpropositionsorsetsofpropositions;foreaseof
expositionIwillusethelatter,without,indoingso,intendinganyclaimabouttheontologyof
explanantia.
What differentiates a full explanation from a partial explanation, I suggest, is that a full
explanation is sufficient for the truth of itsexplanandum, so that a set of propositions fully
explainssomefactonlyifitissufficientforthetruthofthatfact.Thatis:
7ThisisBroome’s(2013,48)example.8It is worth noting that Ruben (2004) makes much use of the distinction between full and partialexplanation, which is, in many respects, similar to mine (though I am perhaps more prescriptive).Schnieder (2011) draws the same distinction between what he calls ‘ complete’ and ‘incomplete’explanations. See also Raz’s (2009, 185–86) discussion of a ‘complete reason why’ for analogousremarks.
105
Assumption Foranypropositionp,someset,Δ,fullyexplainsthefactthatponlyifΔentails9p.10
However, just entailing the truth of some proposition is obviously not sufficient for fully
explaining it (hencetheabove isnotabi-conditional).For instance, the fact that it is raining
entails the fact that it is raining, but the fact that it is raining does not explain (fully or
otherwise)thefactthatitisraining.
Tosubstantiatethispoint,wecanassumethatthe‘fullyexplains’relationis11:
- Irreflexive–Nothingfullyexplainsitself;
- Asymmetric–Ifpfullyexplainsqthenqdoesnotalsoexplainp;and
- Non-monotonic–IfΔisafullexplanationofthefactthatp,thenaddingsomearbitrarypropositiontoΔdoesnotentailthatthesetsocreatedisalsoafullexplanationofthefactthatp.
Whiletheargumentofsubsequentchaptersdoesnotdependonexplanatoryrelationshaving
theseproperties, ifat least someof themseemplausible then that shouldmake it clear the
extenttowhichmereentailment fallsshortofexplanation(asentailment isareflexive,non-
symmetric,monotonicrelation).
Assuming that full explanations are non-monotonic means that adding some arbitrary
propositionintowhatisalreadyafullexplanationofsomefactdoesnotgiveyouafurtherfull
explanation. I want to strengthen this assumption by requiring that full explanations never
containsuperfluousparts;thatis,asWedgwoodputsit:
Theexplanans…mustnotcontainanyirrelevantelementsthatcouldbestrippedawaywithoutmakingitanylesssufficienttoproducetheexplanandum.(Wedgwood2002,363)
Thisisastrengtheningofnon-monotonicitybecauseitrequiresnotjustthatyoucannotaddan
arbitrary proposition into a full explanation and still say that that enlarged set is a full
9Inwhatwaydoesasetofpropositionsthatfullyexplainssomeotherfact ‘entail’ it? Isuggestthat itlogically entails it, so that full explanations necessitate their explananda. This may mean that a fullexplanationofthefactthatpmay includefacts (suchas factsaboutphysical laws)thatareseeminglyextremelyperipheral (thoughnot irrelevant)tothequestion ‘why is it thecasethatp?’However, it isimportant tonotethat I’mnotsayingthateverything ina fullexplanation is thesortof thingthatwewouldsay‘explains’theexplanandum–soonecouldacceptthatsomefactisapartofafullexplanationwithoutacceptingthatitisthesortofthingthatwewouldsay‘explains’theiraction(thismightbewhatwewouldcallan‘enablingcondition’).Seefurtherremarksinthenextsection.10Cf.Schniedersaysthatacompleteexplanationisanexplanation‘whoseexplanansissufficientfortheexplanandum.’(2011,450)11Explanatory relations are commonly assumed to have these properties – see, for instance, Rosen’s(2010)remarksaboutexplanation(ingeneral)inhisdiscussionofgroundingexplanations.Althoughitisworth noting that there is some dissent on whether or not explanatory relations respect theseproperties(seee.g.Ruben2004).
106
explanation, but that each element of the full explanation should be, in some sense,
‘necessary’ to it. This property is, in other areas, known as ‘minimality’.12I characterise it
formallyasfollows:
MINIMALITY Foranypropositionp,someset,Δ,fullyexplainsthefactthatponlyifthereisnoΓsuchthatΓfullyexplainsthefactthatpandΓisapropersubsetofΔ.13
Lastly,itisworthnotingthatwhileafullexplanationissufficientforitsexplanandum,atleast
somefullexplanationsarenotnecessaryfortheirexplananda.Forinstance,supposethatthe
factthatI’vejustbeentothegymtogetherwiththefactthatIamalwaystiredafterI’vebeen
to the gym fully explains why I’m tired. The fact that it fully explains my tiredness in this
instance clearlydoesnotmean thatwhenever I am tired it is because I’vebeen to thegym
etc.,whichistosaythatatleastsomefullexplanationsarenotnecessaryfortheirexplananda.
3.2 Partialexplanation
We can now define the concept of ‘partial explanation’ in terms of the concept of ‘full
explanation’,asfollows:anyelementofafullexplanationofsomefactisapartialexplanation
ofthatfact.Thatis:
Definition Foranypropositions,pandq,ppartiallyexplainsthefactthatqifandonlyifthereisaset,Δ,suchthatΔfullyexplainsthefactthatq,andpisanelementofΔ.
Forthesakeofcompleteness,itisworthnotingthatasetcanalsobeapartialexplanation(ifit
isasubsetofafullexplanation):
Definition Foranyproposition,p,andset,Γ,ΓpartiallyexplainsthefactthatpifandonlyifthereisaΔsuchthatΔfullyexplainsthefactthatpandΓisasubsetofΔ.
Somehouse-keeping:Firstly,justbecausesomeproposition(orset)isapartialexplanationof
somefact,weneednotsaythatitexplainsthatfact.Thatis,whilewemightsaythatthefact
thatJoannehadcarpetundertheholeinherroofisapartofthefullexplanationofwhyher
carpetiswet(it’sisanecessarypartofthesufficientcondition)wemightnotwanttosaythat
thatfactexplainswhyhercarpetiswet.LikeBroome,Isuggestthatwhatdetermineswhether
12See,for instance,Audi’s(2012b,699)characterisationofminimality inthecontextofgrounding.Myformalizationisatranspositionofhis.13 See, for instance, Raz’s (2009, 185–86) remarks about the importance of non-redundancy toexplanations.
107
or notwe say that somepartial explanationexplains someother is amatter of background
knowledgeandotherfeaturesofthecontext.14
Secondly, however, I will assume that if something explains some fact then it is a partial
explanationofthatfact(notethisisnotmerelyrestrictedtowhatwesayexplainsthefact,but
whatactuallyexplainsit).AlthoughIstruggletoseehow,ifoneacceptsthebasicstructureof
fullandpartialexplanationsetouthere,onecoulddenythis,itisperhapsstillworthstressing:
Assumption Foranypropositions,pandq, ifpexplainsthefactthatq thenp isapartialexplanationofthefactthatq.
Thirdly,the‘partiallyexplains’relation,sounderstood,isacontingentrelation(unlikethe‘fully
explains’relation);thefactthatitrainedlastnightonlypartiallyexplainswhyJoanne’scarpet
is wet given other facts about the way the world is. However, even though that fact only
partially explains why the carpet is wet given other facts, the ‘partially explains’ relation is
nonethelessstillbetweenthatfactandthefactthatthecarpetiswet.
Fourthly,andmoretrivially,itfollowsfromtheabovethatallfullexplanationsarealsopartial
explanations(sinceanyfullexplanationisasubsetofafullexplanation).
Lastly,Iwillassume,forthesakeofcompleteness,thatthe‘partiallyexplains’relationis(like
the‘fullyexplains’relation),irreflexive,asymmetricandnon-monotonic.
3.3 Explainingwhythecarpetiswet
Returning to our example, then: the full explanation of the fact that Joanne’s carpet iswet
includes, interalia, factssuchasthefactthat itrainedlastnight,thefactthatherroof leaks
and the fact that there is carpet beneath her leaky roof. That full explanation entails that
Joanne’scarpetiswet.
Eachof themembersof that full explanation (i.e. the fact that it rained lastnight, etc.) is a
partial explanation of why Joanne’s carpet is wet. However, contextual and pragmatic
considerationswilldeterminewhetherornotwesay,ofanygivenpartialexplanation,thatit
explainswhyhercarpetiswet.15
14Thisisasuggestionaboutthepracticeofsayingthatonethingexplainsanother–Imakenocommentonwhetherornotallpartialexplanationsexplainactionsinsomenon-contextrelativesense.15Cf. ‘A partial explanation may be good relative to one set of circumstances, but bad relative toanother,inwhichinterests,beliefs,orwhateverdiffer.’(Ruben2004,22)
108
4 Overdeterminationandoverexplanation
Havingthuscharacterisedfullandpartialexplanation,Inowwanttoofferacharacterisationof
overexplanation in terms of these concepts, as it is of some relevance to The Explanatory
ExclusionProblem.
4.1 Overdetermination
Somethingissaidtobeoverdeterminedifandonlyiftherearetwoseparatesetsofconditions
that are each individually sufficient for it to obtain and those conditions determine that it
obtain.16Differentkindsofdetermination relationyielddifferent sortsofoverdetermination.
Causal overdetermination (and whether or not it is possible) is probably the most hotly
debatedkindofoverdetermination–hereisarepresentativecharacterisation:
Suppose that a certain event, in virtue of its mental property, causes a physical event. Thecausal closureof thephysical domain says that thisphysical eventmust alsohaveaphysicalcause.Wemayassume that thisphysical cause, in virtueof itsphysical property, causes thephysical event… Could it be that the mental cause and the physical cause are each anindependent sufficient cause of the physical effect? The suggestion then is that the physicaleffect isoverdetermined.So if thephysical causehadn'toccurred, themental causeby itselfwouldhavecausedtheeffect.(Kim1993,280–81)
Theprincipleissomethinglikethis:somethingiscausallyoverdeterminedifyoucouldtakeone
of its causes away and it would still obtain (or occur, or exist or what have you). A classic
example: twovandalseach throw rocks that simultaneously strikeandbreakawindow.The
breaking of thewindow is seemingly causally overdetermined because either rock-throwing
wouldhavebeensufficienttobreakthewindow.It’sbeyondthescopeofthisdiscussiontoget
intowhatthatmeansforthecausalstatusofeitherrock-throwing.
4.2 Overexplanation
HereishowIproposetocharacteriseoverexplanation:
Definition For any propositionp, the fact thatp is overexplained if and only ifthereare(atleast)twogenuinelydifferentfullexplanationsofthefactthatp.
Is there a difference between overexplanation and overdetermination? If you are an
explanatory realist (so that explanatory relations 17 are underpinned by ontological
16I stress the latterconjunctas therebeingmerely twosetsofconditions thatentail somefact isnotsufficient for the fact to be over-determined (if it were then, arguably, everything would beoverdetermined, given the reflexivity of the entailment relation, and the claim that everything isdeterminedinsomemoremetaphysicallysignificantsensethanentailment).17OfthekindIhaveinmind(seefn.3ofthischapterforclarification).
109
determinationrelations18)andyouholdthatexplanationsaregenuinelydifferentonly if they
are independent,19then youwill probably think that overdetermination andoverexplanation
arethesamething.
However,sinceIwanttoallowforthepossibilityofrejectingeitherexplanatoryrealismorthat
two explanations are genuinely different only if they are independent, I distinguish
overexplanation from overdetermination. Distinguishing them in this manner does not
preclude thepossibilityof therebeing the same; thatdependsonlyonhowonedefines the
definiensofoverexplanation.
Whatdoesitmeanfortwofullexplanationstobegenuinelydifferent?Isuggestitisthatthey
shouldexplaintheexplanandumindifferentways.Whycallthem‘genuinelydifferent’andnot
merely ‘different’explanations?Because Iwanttoallowforthepossibilitythatnon-identical
fullexplanations(i.e.‘different’)fullexplanationsmaynonethelessexplainsomeexplanandum
inthesameway(i.e.withoutbeing‘genuinelydifferent’).
4.3 Benignoverexplanation
It is widely believed that genuine causal overdetermination is rare (if it is even possible).20
Assuming that causal determination relations underpin causal explanatory relations, causal
overexplanationispresumablyequallyrare.
However,bonafidecasesofnon-causaloverexplanationabound.Forinstance,recallthatIsaid
thatswimmingwillbothhelpmesleepbetterand improvemymood.Swimming is then, for
me, in some respect, worth doing partly because it will help me sleep better and partly
because it will improve my mood. The explanatory relations involved here are, I suggest,
clearlynotcausal.
Nowconsiderthis:swimmingwouldstillbe, insomerespect,worthdoingevenif itwouldn’t
helpmesleepbetterbecauseitwouldstillimprovemymood.Conversely,ifitweren’tthecase
thatswimmingwouldimprovemymood,wecouldstillsaythatitwas,insomerespect,worth
doingbecauseitwouldhelpmesleepbetter.
18Cf.: ‘According to “explanatory realism,”when something is correctly invoked as an explanation ofanotherthing,theexplanatoryrelationmustbegroundedinsomeobjectiverelationofdependenceordeterminationholdingfortheexplanansandtheexplanandum.’(Kim1993,xii)19Thatis,inwhateversenseyoutakeanoverdeterminingcause/factortobeindependent.20Kim (1993, 280) describes the idea that there could be systematic causal overdetermination as‘absurd’.
110
Incontrast,ifithadn’trainedlastnightthen,ceterisparibus,thefactthatJoanne’sroofleaks
wouldnotcontinuetopartiallyexplainthefactthathercarpetwaswet(hercarpetwouldnot
havebeenwet,ifherroofdidn’tleak).
Whatdifferentiates these twocases is seemingly this:while the fact that it rained lastnight
andthefactthatherroofleaksarepartofthesamefullexplanationofwhyhercarpetiswet,
the fact that swimmingwould helpme sleep better and the fact that itwould improvemy
moodarepartoftwogenuinelydifferentfullexplanationsofwhyswimmingwouldbe,insome
respect,worthdoing–theyexplaintheexplanandum indifferentways.Thus,sincethereare
twogenuinelydifferentfullexplanationsofwhyswimmingis,insomerespect,worthdoing,it
is overexplained. Whereas, in contrast, since there are not two genuinely different
explanationsofwhyJoanne’scarpetiswet,thefactthathercarpetiswetisnotoverexplained.
The characterisation of overexplanation hangs on what it is for two explanations to be
genuinelydifferent. I havegivenwhat I take tobeanunambiguousexamplehere,but Iwill
returntowhatmakesexplanationsgenuinelydifferentinthenextchapter(see§(VIII)3.2).
5 Summary
I have said that there are two kinds of explanatory relation: full and partial, and I’vemade
some assumptions about the logical properties of each. I then provided an analysis of
overexplanation in termsof fullexplanation. In thenextchapter Iwillput theseconcepts to
work in a characterisation of the main challenge to my theory: The Explanatory Exclusion
Problem.
111
(VIII)
TheExplanatoryExclusionProblem
InwhichIsetoutTheExplanatoryExclusionProblem,whichis,insomeformoranother, the motivating argument for psychologistic theories of reasons. Iprovide a formal construal of the Problem, showing how it results from twoseemingly trivial claims aboutwhat explains an agent’s actionwhen they actfromerrorandfromignorancetogetherwithfiveseeminglyplausibleprinciplesofexplanation.IshowhowtheProblemimpliesthatIdidnotcongratulatemyfriendbecauseshehadwonanaward,butonlybecauseIthoughtshehad.
Myfriendhaswonamuch-covetedaward;IreadaboutitinanewspapersoIcallherupand
congratulate her. Did I congratulate her because she had won an award, or just because I
thoughtshehad?
There is a well-established response to this question that proceeds, broadly, along the
followinglines:ifshehadn’twonanawardbutIhadbelievedthatshehad,thenIwouldstill
havecongratulatedher,and Iwouldhavecongratulatedherbecause Ibelievedthatshehad
wonanaward.Conversely, if Ihadn’tbelievedthatshehadwontheawardtheneven ifshe
hadwonit Iwouldnothavecongratulatedher.So, if Ihadn’tbelievedthatshehadwonthe
award,thefactthatshehadwontheawardcouldnothaveexplainedwhyIcongratulatedher
(sinceIwouldn’thave).
So,itseemsasthoughwhatmatterstotheexplanationofwhyIcongratulatedmyfriendisthe
factthatIbelievedthatshehadwonanawardandnotthefactthatshehadwonanaward;
that is, I did not really congratulate my friend because she had won an award, but only
becauseIthoughtshehad.
Anargumentalongtheselinesiswhattypicallymotivatestheviewthatfactsabouttheworld
cannotexplainanagent’sactionand that, therefore,anagent’s reason foractingmustbea
featureoftheirpsychology.1Indeed,thislineofreasoningisthemotivatingargumentforwhat
Icalled‘psychologismaboutthereasonsforwhichweact’.2So,forthosewhowanttoreject
thatformofpsychologism(asIdo),thisargumentistheonetobeat.
InwhatfollowsIofferaformalconstrualoftheargumentasanargumentaboutwhatexplains
anagent’saction(i.e.asanargumentthatisonlyindirectlyaboutwhattheirreasonforacting
is).Iwillusetheconceptsintroducedinthepreviouschaptertoformallycharacterisethethree
1Sinceanagent’sreasonforactingalwaysexplainstheiraction(see§(IV)1.2).2SeeTableI-4.
112
componentsofwhat Iwillcall ‘TheExplanatoryExclusionProblem’,whichare:theargument
fromfalsebelief3;theargumentfromimpotentfacts4;andaprincipleofexplanationthatIcall
‘the exclusion principle’5. In particular, on my construal, the conclusion of The Explanatory
ExclusionProblemcanbearrivedatfromtwoseeminglytrivialclaimsaboutwhatexplainsan
agent’s action when they act in error or ignorance, together with five seemingly plausible
principlesofexplanation.
MyintentinformalisingTheExplanatoryExclusionProblem isthree-fold:firstly,doingsowill
helpmedemonstrate,in§(IX),thattheProblemappliesalsotothefactsonwhichanagent’s
beliefsarebased6andtotheexplanationofwhyit isrationalforsomeonetodosomething.7
Secondly,formalisingtheProblemwillhelpmetodiscriminatemoreeasilybetweendifferent
responsestoit.And,thirdly,aformalconstrualallowsmetoidentifymorepreciselywhereThe
ExplanatoryExclusionProblemgoeswrong.
I should note that other construals of this argument are possible, andmine is by nomeans
definitive (though I hope it is illuminating). However, I do not think one can construe the
overallprobleminawaythatmakesitimmunetomyeventualresponse;thatis,Idonotthink
thatmyresponsetotheoverallproblemhangsonformalisingitinthewaythatIdo.
Lastly, we should also be clear that my discussion here is strictly about what explains an
agent’saction,andnotwhattheirreasonforactingis.TheExplanatoryExclusionProblemonly
bearsuponwhatanagent’sreasonforactingcouldbetotheextentthatweassumethatan
agent’s reason for acting always explains their action (although this is a widely held
assumption-see§(IV)1.2).
1 Anoverview
Recall that I said that if explanatory rationalism is to be consistent with the prima facie
reasonableclaimssetoutin§§(II)-(IV),thefollowingmustbetrue:
(R1) Icongratulatedmyfriendbecauseshehadwonanaward.
TheExplanatoryExclusionProblemprovidesthefollowingargumentagainstthisclaim:
3ThisisStout’s(1996,2009)nameforthisargument.4Stout(1996)discusseswhathecalls,‘TheArgumentfromtheImpotenceUnrepresentedFacts’;whilehisargument is in somerespects similar tomine, thepremisesandconclusionsofourargumentsaresufficientlydifferentastomakethemdifferentarguments.5This principle share’s some similarities with Kim’s principle of causal exclusion, but, as I shall note,differsfromitinparticularrespectsthatshallturnouttobecriticaltothisdiscussion.6Andnotmerelyfactsabouttheexternalworld.7Andnotmerelytotheexplanationofwhytheydoit.
113
TheExplanatoryExclusionProblemfor(R1)
Premise1 ThereisafullexplanationofwhyIcongratulatedmyfriendsuchthatthe fact that she had won an award is neither a part of that fullexplanationnorisitpartofagenuinelydifferentexplanation.8
EXCLUSION For any propositions,pandq, if there is a full explanation ofwhyqsuchthatp isneitherapartofthatfullexplanationnor is itpartofagenuinelydifferentexplanation9,thenpdoesnotpartiallyexplainq.
Conclusion1 The fact that my friend had won an award does not explain why Icongratulatedher.
The focus of this chapter is on setting out the argument for Premise1 and for EXCLUSION. In
§(IX) I will show how, by altering Premise1, The Explanatory Exclusion Problem can also
provideargumentsagainst(R2),(R3)and(R4).
1.1 TheargumentforPremise1
Premise1, as I will demonstrate, follows from the conclusions of two other arguments: the
argumentfromfalsebeliefandtheargumentfromimpotentfacts.
TheconclusionoftheargumentfromfalsebeliefisthatwecangiveafullexplanationofwhyI
congratulatedmy friendwithoutmentioning the fact that shewonanaward; so longaswe
notethatIbelievedthatshehadwonanaward.Ishowhowthisconclusioncanbearrivedat
fromaseeminglytrivialclaimaboutwhatwouldhaveexplainedmyactionhadmybeliefbeen
falsetogetherwiththreeseeminglyplausibleprinciplesofexplanation.
Theconclusionoftheargumentfromimpotentfactsisthatthefactthatmyfriendhadwonan
awardisnotpartofagenuinelydifferentexplanationofwhyIcongratulatedherfromthefact
that I believed that she had. I show how this conclusion can be arrived inferred from (i) a
seemingly trivial claimaboutwhat theexplanatorypowerof the fact thatmy friendwonan
awardwouldhavebeenif Ihadn’tbelievedthatshehad;togetherwith(ii)anotherplausible
principleofexplanation.
Premise1canthenbeinferredfromtheconclusionsofthesetwoarguments.
1.2 TheargumentforEXCLUSION
TheargumentforEXCLUSIONismorestraightforward.Inshort: ifsomefactisnotpartofafull
explanation of some explanandum and it is not part of a genuinely different explanation of
8That is,agenuinelydifferentexplanationofwhy I congratulatedmy friend… Iomit thisqualificationthroughout,forbrevity.9That is, a genuinely different explanation ofwhyq…Again, I omit this qualification throughout, forbrevity.
114
thatexplanandum,thenseemingly,bythelawoftheexcludedmiddle,itisnotpartofanyfull
explanationofthatexplanandum,whichmeansthatitdoesnotexplainit.
1.3 What’snext
InwhatfollowsIsetouttheargumentfromfalsebeliefandtheargumentfromimpotentfacts.
I then show the conclusions of these two arguments yield Premise1. I then set out the
argumentforEXCLUSION.IthenshowhowPremise1,togetherwithEXCLUSION,yieldsConclusion
1(whichisthedenialof(R1)).
Forreference,thefigurebelowprovidesanoverviewofthestructureofthisdiscussion.
FigureVIII-1:TheargumentforTheExplanatoryExclusionProblem
2 TheArgumentfromFalseBelief
Dancyprovidesaconcisesummaryoftheargumentfromfalsebelief,asitrealtestoanagent’s
reasonforacting:
[Consider]thecasewherethingsarenotastheagentconceivesthemtobe.Surely, insuchacase,wecannot say thathis reason foractingashedidwas thatp.Wehave to say thathisreason foractingwas thathebelieved thatp.Accepting this for thecasewhere the relevantbeliefisfalse,then,wemightstillhopethat‘thatp’canindeedbetheexplanationoftheactionwhereitisthecasethatp,butthatwhereitisnotthecasethatptheexplanationcanonlybe‘thathebelievedthatp’.But,asBernardWilliamsputsit,thetrue-falsedistinctionshouldnotbe allowed to affect the form of the relevant explanation. Supposing, therefore, that ourexplanation should take the same form whether it is or is not the case that p, and havingalreadyacceptedthatthecorrectexplanationincaseswhereitisnotthecasethatpis‘thathebelieved thatp’,wearedriven to say the samewhere the relevantbelief is true rather thanfalse.(Dancy2000,121)
AdaptingDancy’sargumenttomypresentconcernaboutwhatexplainsanagent’saction(and
notjusttheirreasonforacting),themainconclusionofthislineofreasoningseemstobethat
wecanalwaysexplainanagent’sactionintermsoftheirbeliefs,withoutreferencetothetruth
Premise1.AFACTIVITY
ENDURANCESUFFICIENCY
Premise1.B
DIFFERENCE
Premise1
Conclusion1 (§3)TheArgumentfromImpotentFacts
(§2)TheArgumentfrom
FalseBelief
EXCLUSION
(§§4-6)TheExplanatoryExclusionProblem
Conclusion1.A
Conclusion1.B
115
orfalsityofthatbelief.Thisconclusionmayseemobvious,however,sincesomedenyit,10itis
worthbeingexplicitabouttheargumentforit.
In what follows I provide a formal construal of this argument. I take my construal to be
plausibleand, Ihope, informative;however,other construalsareavailable (e.g. Stout2009).
The conclusion of the argument from false belief, in the case of my friend’s award, is as
follows:
Conclusion1a There is a full explanation of why I congratulated my friend thatincludes the fact that I believed that shehadwonan awardbut notthefactthatshehadwonanaward.
ThisconclusionisthefirstpartoftheargumentforPremise1. In§3, Ipresenttheargument
fromimpotentfacts,whichisthesecondpartoftheargumentforPremise1.
2.1 Actingonfalsebeliefs
Theargument from falsebelief starts, predictably,withanobservationaboutwhathappens
whenanagentactsona falsebelief.Consider:even ifmybeliefhadbeenfalse, Iwouldstill
havecongratulatedmyfriend,andIwouldhavecongratulatedherbecauseIbelievedthatshe
hadwonanaward.11Now,tosaythatIwouldhavecongratulatedherbecauseIbelievedthat
shehadwonanawardistosaythattheformerpartiallyexplainsthelatter.
Moreover, I suggest, when I congratulate her because of my false belief that she won an
award, the fact that she did not win an award is not part of any explanation of why I
congratulatedher.12
Thismuchshouldbeundeniable.Thus:
Premise1a If,ceterisparibus,myfriendhadnotwonanaward(butIstillbelievedthat shehad13), then (i) the fact that Ibelieved that shehadwonanawardwouldhavepartiallyexplainedwhyIcongratulatedher;and(ii)the fact that she had not won an award would not have partiallyexplainedwhyIcongratulatedher.
2.2 TheFactivityPrinciple
Recallingthediscussionof§(IV)1.3,wherewenotedthatthefactivityofexplanationisaprima
faciereasonableandwidelyheldprincipleofexplanation,letusjustassumethefollowing:
10Seethediscussionofexclusivedisjunctivisttheoriesin§(X)A.4.11Compare:whydidSallyrun?Becauseshethoughtabearwaschasingher.12Perhaps, one could craft a weird example in which it was, but that is not my example. Likewise,compare:thefactthatabearwasn’tchasingSallydoesnotexplainwhysheran.Therelevanceofthisqualificationwillbecomeclearasthediscussionproceeds.13Thisremarkisparentheticalbecauseitisalreadyimpliedbytheceterisparibuscondition.
116
FACTIVITY Foranypropositionspandq,ifppartiallyexplainsthefactthatqthenpisthecase.
ThismeansthatwhenSallyrunsbecauseshemistakenlybelievesthatabearischasingherthe
falseproposition that a bearwas chasingher doesnot explainwhy she ran. Likewise, ifmy
belief thatmy friendhadwonanawardhadbeen false, the falseproposition thatmy friend
hadwonanawardwouldnotexplainwhyIcongratulatedher.
2.3 Twomoreprinciplesofexplanation
Considerthefollowing,much-discussedremark:
Thedifferencebetweenfalseandtruebeliefsontheagent’spartcannotaltertheformoftheexplanationwhichwillbeappropriatetohisaction.(Williams1981,102)
Here iswhat I take to be an implication ofwhatWilliams is saying, inmy terminology: the
samefullexplanationofone’sactionisavailablewhetherone’sbeliefistrueorfalse.14While
thisclaimmayseemobvious,therearethosewhodenyit,andwhodosofordifferentreasons,
soitisworthconsideringitinmoredetail.
In what follows, I want to show how this claim can be motivated by two principles of
explanation, the endurance principle and the sufficiency principle, that are both general (i.e.
applybeyondtheexplanationofaction)andplausible.
2.3.1 TheEndurancePrinciple
Thefollowingremark,Isuggest,reliesontheenduranceprinciple:
Whenanagentactsonfalsebeliefs,wecannotexplaintheactionintermsofthefactsbutonlyintermsofthosebeliefs–thereisonlyaninternalistexplanationoftheiraction.Butevenwhenthebeliefsaretruethatsameinternalistexplanationworks.(Stout1996,24emphasisadded)
Stoutsuggests thatsincewecanexplainanagent’sactions in termsof the factsaboutwhat
theybelievedwhentheirbeliefwasfalse,wecanlikewiseexplaintheiractionintermsofthe
factsaboutwhattheybelievedwhentheirbeliefistrue.Inwhichcase,giventhefactthatSally
believedthatabearwaschasingher(partially)explainswhysheranwhenherbeliefwasfalse,
if,ceterisparibus,herbeliefhadbeentrue,thefactthatshebelievedthatabearwaschasing
herwouldstillhave(partially)explainedwhysheran.Butwhyshouldthisbethecase?Why
doesitfollowthatwhatexplainsinthefalsebeliefcasealsoexplainsinthetruebeliefcase?
Let’s startbynoticing this:whenSally runsbecause shemistakenly thought thatabearwas
chasingher,weknow, fromFACTIVITY, that the falseproposition thatabearwaschasingher
14Williams’sremark,Ithink,goesfurtherthanthis–Ithinkthat,giventhatexplanationisfactive,thereisareadingofhisremarkonwhichitjustistheconclusionoftheargumentfromfalsebelief.
117
(i.e. the content of Sally’s belief) couldnot partially explainwhy Sally ran. Furthermore, the
factthatabearwasnotchasingSallyisalsonotpartofanyfullexplanationofwhysheran–
and,inparticular,itisnotpartofthesamefullexplanationasthefactthatshebelievedthata
bearwas chasing her.15So, there seems to be a clear sense inwhich, when Sally’s belief is
false, the proposition that a bear is chasing her is irrelevant to the explanation ofwhy she
ran.16
Nowconsiderwhatwouldhavehappenedif,ceterisparibus,Sally’sbeliefhadbeentruerather
thanfalse?Thiswouldhavehappened:apropositionthatwasirrelevanttotheexplanationof
heraction(i.e.thatabearwaschasingher)wouldhavegonefrombeingfalsetobeingtrue.
But why would that proposition’s suddenly becoming true affect the pre-existing partial
explanationrelationsif itwasirrelevanttothosepartialexplanationswhenitwasfalse?That
is,whywouldSally’sbeliefthatabearwaschasingherstopexplainingheractionjustbecause
afalsepropositionthatwasirrelevanttothatexplanationsuddenlybecametrue.17Seemingly,
itwouldnot.
To the extent that this line of reasoning is persuasive, I suggest that that is just because it
accords with a more general principle of explanation – namely that if, ceteris paribus, a
propositionthatisirrelevanttosomeexplanationofsomeexplanandum(i.e.neitheritnorits
negation is part of that full explanation of that explanandum) when it is false suddenly
becomes true, then that does not stop anything that partially explained that explanandum
when that proposition was false from continuing to explain it when it is true. That is, the
partialexplanationrelationsbetweenfactsendurewhenthetruth-valueofapropositionthat
isirrelevanttothemchanges.Thus:
15Again,onecouldcraftaweirdexampleinwhichitwas,butthatisnotmyexample.16This does notmean that the proposition is irrelevant in all senses of relevance – for instance, it iscertainly something she took to make running worth doing; the point is just that from a particularexplanatory perspective, it is seemingly irrelevant. It is also worth stressing this: just because theproposition thatabear ischasingher is irrelevantwhen it is false,doesnotmeanthat it is irrelevantwhen it is true – a proposition is irrelevant to some partial explanation of some fact only in someparticularinstanceandonlyinsofarasneitheritnoritsnegationarepartofthesamefullexplanationofthatfactasthatpartialexplanation.17Consider:evenifthefactthatabearwaschasingherisrelevanttotheexplanationofheractionwhentruethatdoesnotshowthatitsbecomingtruewoulddestroypre-existingpartialexplanationrelations,despiteitspriorirrelevance.
118
ENDURANCE For any propositionsp, qand r, the following holds: Suppose thatqpartiallyexplainsthefactthatrwhenitisnotthecasethatp.Supposefurtherthatneitherpnornotpispartofthesameexplanationofrasq. Then, if, ceteris paribus, it were the case that p, q would stillpartiallyexplainthefactthatr.18
Althoughthisclaimislong-winded,Itaketheprincipletobeultimatelyintuitive.However, in
the event that it is not clear, the next section discusses some examples and a failed
counterexample.
2.3.2 Someexamplesoftheenduranceprinciple
ThereaderwhoisalreadycomfortablewithENDURANCEmayskiptheseexamples.
Example 1: An uncontroversial example: Suppose that a teacher is performing a science
experiment for her students, although one of the students, Nathan, is absent. She heats a
metalrodanditexpands.Thefactthattherodwasheatedpartiallyexplainswhyitexpanded,
and the false proposition thatNathan is present is irrelevant to that explanation of why it
expanded.19Moreover, ceteris paribus, had Nathan been present, the fact that the rodwas
heatedwouldstillpartiallyexplainwhyitexpanded.
Example 2:Of course,Nathan’sbeingpresent is rarely likely to feature inanexplanationof
whytherodexpanded,sohereisanotherexample,inwhichthepropositionthatisirrelevant
whenfalse,explainswhentrue.Considerthecaseinwhichswimmingwillhelpmesleepbetter
butwon’t improvemymood.Wealreadyestablished20that the fact thatswimmingwillhelp
me sleepbetterpartially explainswhy swimming is, forme, in some respect,worthdoing.21
Moreover, Isuggest,neitherthefalsepropositionthatswimmingwill improvemymood,nor
thefactthatitwon’timprovemymoodexplainwhyitis,insomerespect,worthdoing–sothe
propositionthatswimmingwill improvemymood is irrelevanttothefullexplanationofwhy
swimmingis,insomerespect,worthdoing.
However,if,ceterisparibus,swimmingwouldimprovemymoodthenitwouldpartiallyexplain
whyswimmingwas, insomerespect,worthdoing.But justbecause,onceitobtains,thefact
that swimmingwould improvemymoodstartsexplainingwhyswimmingwouldbe, in some
respect, worth doing, doesn’t mean that the fact that swimming will help me sleep better
18Moreformally:Foranypropositionsp,qandr,if,whenitisnotthecasethatp,qpartiallyexplainsthefact thatrthen,providedthatneitherpnornotp ispartof thesameexplanationofrasq, if,ceterisparibus,itwerethecasethatp,thenqwouldstillpartiallyexplainthefactthatr.19Neitheritnoritsnegationispartofthefullexplanation(ofwhichthefactthattherodwasheatedispart)ofwhytherodexpanded.20See§(VII)4.3.21GiventhatIwanttosleepbetter,orjudgeitgoodandwhathaveyou.
119
stopsexplainingwhyitis,insomerespect,worthdoing.Allthathappensnowisthattheyboth
explain.
Example3:Awould-becounterexample:supposethatTomthrowsarockatawindowandit
breaks, while Susie stands and watches. The fact that Tom threw a rock at the window
(partially) explainswhy it broke. Seemingly, neither the false proposition that Susie threwa
rockatthewindownorthefactthatshedidn’tthrowarockexplainswhythewindowbroke:
thepropositionthatSusiethrewarockisirrelevanttothefullexplanationofwhythewindow
broke.
However,supposethatSusiehadthrownarock,andshehadthrownitbeforeTom,andher
rock had broken thewindow. Inwhich case Tom’s rock-throwingwould seemingly cease to
explainwhythewindowbroke(hisrockwouldhavesailedthroughtheemptyspacewherethe
windowusedtobe).So,contratheenduranceprinciple,whenanirrelevantpropositiongoes
fromtruetofalsethatcanstopsomethingfrompartiallyexplainingtheexplanandum.
This counterexample fails because it violates the ‘ceteris paribus’ condition in ENDURANCE. In
makingitthecasethatSusiethrewarockatthewindowanditbrokewearenotchangingthe
truth-valueofonly irrelevantpropositions,butof relevantpropositionsaswell.For instance,
the fact that the window was intact before Tom’s rock hit it is a part of the same full
explanationofwhythewindowbrokeasthefactthatTomthrewtherock.22Soofcoursethe
fact that Tom threw the rock stops explainingwhy thewindowbroke – by adding in Sally’s
rock-throwing we have taken away an element of the full explanation that Tom’s rock-
throwingwaspartof,soeverythinginitstopsexplaining.
Incontrast,supposewehonourthe ‘ceterisparibus’conditionandchangeonlythatwhich is
not part of the same full explanation of thewindow’s breaking as Tom’s rock throwing. So,
let’s suppose that Susie’s rock harmlessly bounces off the window. In this circumstance it
shouldstillbeclear(giventhatallotherthingsareequal)thatTom’srockbreaksthewindow
andTom’srock-throwingexplainswhythewindowbroke.ENDURANCEperseveres.
2.3.3 TheSufficiencyPrinciple
Now: another principle of explanation – the sufficiency principle. Recall that I interpreted
Williams’sremarkastheclaimthatthesamefullexplanationofanagent’sactionisavailable
whether their belief is true or false. We need more than just ENDURANCE to reach that
conclusion.
22It is anecessarypartof that full explanationbecause if thewindowhadn’tbeen intact, itwouldn’thavebeenbrokenbyTom’srock.
120
Consider:thereisasetofpartialexplanationsofwhySallyran(whenherbeliefwasfalse)that
isafullexplanationofwhysheran.WhatENDURANCEtellsusisthatthatwhichpartiallyexplains
her actionwhen her belief is falsewould also partially explain if, ceteris paribus,her belief
were true. So, the same setofpartial explanations that fully explainedher actionwhenher
beliefwasfalsewouldstillbeasetofpartialexplanationsofheractionifherbeliefweretrue.
However,weneed toadd some further requirement to guarantee that that set is still a full
explanationwhenherbelief is true(asopposedtobeingmerelyasetofpartialexplanations
i.e.anincompletefullexplanation).Whatisthatrequirement?
Itisthis:whateversufficestoexplainanagent’sactionwhenanagent’sbeliefisfalselikewise
suffices to explain it when their belief is true. And, again, I suggest that this is just a
consequenceof an intuitively plausible, andmore general principleof explanation –namely
thatifsomesetofpartialexplanationssufficestoexplain(i.e.isafullexplanationof)somefact
in some situation, then whenever those partial explanations all explain that fact, they will
sufficetoexplainit.Thus:
SUFFICIENCY Foranypropositionq,andanyset,Δ, ifΔ isa fullexplanationof thefact thatq in somecircumstance, then, inanycircumstance inwhichall theelementsofΔpartiallyexplain the fact thatq,Δ fullyexplainsthefactthatq.23
2.3.4 CombiningtheEndurancePrincipleandtheSufficiencyPrinciple
Theenduranceprincipleandthesufficiencyprincipleprovideuswiththeconclusionthatthe
same full explanation of an agent’s action is available whether their belief is true or false.
How?Byensuringthat justchangingthetruth-valueofsomepropositionthat isoutsideofa
full explanation of some explanandum (i.e. which is irrelevant to the explanation of that
explanandum)cannotaffectwhetherornotthat fullexplanation isavailable.So, ifyouthink
thatneitherthepropositionthatabearischasingher,northefactthatabearisnotchasing
her, arepartsof the full explanationofwhySally ran (whenherbeliefwas false), then that
samefullexplanationwillbeavailable,ceterisparibus,regardlessofwhetherornotabear is
chasingher.24
23ThismayseemclosetoMINIMALITY,butitisaverydistinctclaim.However,theconjunctionofthetwoyieldswhatwemightcallcounterfactualminimality,whichcanbedefinedasfollows:
C-MINIMALITY Foranypropositionp,someset,Δ,fullyexplainsthefactthatponlyifthereisno Γ such that Γ is a proper subset of Δ and if, ceteris paribus, Δ had notexistedbutΓhad,thenΓwouldfullyexplainthefactthatp.
Isuspectthatan intuitivecommitmenttoC-MINIMALITY,resultingfroman implicitcommitmenttobothMINIMALITYandSUFFICIENCYasprinciplesofexplanation,helpsmotivatestheargumentfromfalsebelief.24Notethatnon-factivists(e.g.Dancy2000),whothinkthatexplanationcanbenon-factive,wouldsaythatthatabearischasingher(quathecontentofherbelief)ispartofthefullexplanationofwhySally
121
This may seem elaborate, but distinguishing these two general explanatory principles is
essentialnotonlytodiscriminatingbetweendifferentrejectionsofTheExplanatoryExclusion
Problem,25butalsotoclearlydemonstratingwhyitiswrong.
2.4 Concludingtheargumentfromfalsebelief
Here,then,ismyconstrualoftheargumentfromfalsebelief:if,ceterisparibus,myfriendhad
notwonanawardthenthefactthatIbelievedthatshehadwonanawardwouldhavebeena
partofafullexplanationofwhyIcongratulatedher(fromPremise1a).However,neitherthe
fact that she had not won an award nor the false proposition that she had won an award
wouldhavebeenpartofthatexplanation(fromPremise1aandFACTIVITY).
Weshouldmaketwoobservationsfromtheseremarksaboutwhatwouldhavebeenthecase
if my belief had been false: firstly, there would have been a full explanation of why I
congratulatedmyfriendthatwouldhaveincludedthefactthatIbelievedthatshehadwonan
award but not the (false) proposition that she had won an award. Secondly, the (false)
propositionthatshehadwonanawardwouldhavebeenirrelevanttotheexplanationofmy
action.
From the secondobservation,wecan conclude (fromENDURANCE) thatwhateverwouldhave
partially explainedmyaction ifmybeliefhadbeen falsemust alsopartially explain itwhen,
ceterisparibus,mybeliefistrue.So,sincealltheelementsofwhatwouldhavefullyexplained
myactionhadmybeliefbeenfalsealso(partially)explainmyactionwhenmybeliefistrue,we
caninfer(fromSUFFICIENCY)thatthatsetofpartialexplanationsmustlikewisefullyexplainmy
actionwhenmybeliefistrue.
Now recall the first observation: had my belief been false there would have been a full
explanation of my action that included the fact that I believed that my friend had won an
awardbutnotthe(false)propositionthatshehadwonanaward.Butgiventhatthesamefull
explanationisavailablewhethermybeliefistrueorfalse,thatmustmeanthatevenwhenmy
beliefistrue,thereisafullexplanationofmyactionthatincludesthefactthatIbelievedthat
myfriendhadwonanawardbutnotthefactthatshehadwonanaward.
ran.This,however,wouldnotstopthemfromagreeingwiththeclaimthatthesamefullexplanationofSally’sactionisavailablewhetherherbelief istrueorfalse(indeed,thedesiretoagreewithWilliams’claimisapartofwhatpersuadesDancytoadoptnon-factivism).25In particular: those whom I call ‘exclusive disjunctivists’ (e.g. Collins 1997; Stoutland 1998) rejectENDURANCE,whereasthosewhomIcall ‘supplementarists’(possiblyAlvarez2010)rejectSUFFICIENCY.SeetheAppendixto§(X)forfurtherdiscussion.
122
Formally:
TheArgumentFromFalseBelief
Premise1a If,ceterisparibus,myfriendhadnotwonanaward(butIstillbelievedthat she had), then (i) the fact that I believed that she hadwon anawardwouldhavepartiallyexplainedwhyIcongratulatedher;and(ii)the fact that she had not won an award would not have partiallyexplainedwhyIcongratulatedher.
FACTIVITY Foranypropositionspandq,ifppartiallyexplainsthefactthatqthenpisthecase.
ENDURANCE For any propositionsp, qand r, the following holds: Suppose thatqpartiallyexplainsthefactthatrwhenitisnotthecasethatp.Supposefurtherthatneitherpnornotpispartofthesameexplanationofrasq. Then, if, ceteris paribus, it were the case that p, q would stillpartiallyexplainthefactthatr.
SUFFICIENCY Foranypropositionq,andanyset,Δ, ifΔ isa fullexplanationof thefact thatq in somecircumstance, then, inanycircumstance inwhichall theelementsofΔpartiallyexplain the fact thatq,Δ fullyexplainsthefactthatq.
Conclusion1a There is a full explanation of why I congratulated my friend thatincludes the fact that I believed that shehadwonan awardbut notthefactthatshehadwonanaward.
Theconclusionof theargument from falsebeliefprovides the firstpartof theargument for
Premise1. The argument from impotent facts, which is the subject of the next section,
providesthesecondpartofthatargument.
3 TheArgumentfromImpotentFacts
Theargumentfromimpotentfacts is,mercifully,simpler. It is, insomesense,summarised in
thefollowingremark:
Whenevertheagentactsinlightofthefactthatp[i.e.becausep],theagentmusttakeitthatp, and I understand this sort of ‘taking it that’ as aweak formof belief… The psychologisedexplanationoftheactionistobeunderstoodasthesameexplanationasthenon-psychologisedone.(Dancy2000,126)
Dancymakes twokeyobservations: the first is that if theagent loses theirbelief thatp, the
possibilityofexplainingintermsofthefactthatpdisappears.Thesecondisthatthefactthatp
isnotpartofa(genuinely)differentexplanationoftheagent’sactionfromthefactthatthey
believedthatp.
TheargumentfromimpotentfactsshowshowDancy’sfirstobservation,togetherwithanother
generalprincipleofexplanation,entailshissecond.Appliedtothecaseofmyfriend’saward,it
runs as follows: the fact thatmy friendhadwon an awardwould not have explainedwhy I
123
congratulatedherifIhadn’tbelievedthatshehadwonanaward(because, interalia,Iwould
nothavecongratulatedher).Thatbeingso,theexplanatorypowerofthefactthatmyfriend
hadwonanawarddependsonmybelievingthatshehad.Butonepropositioncannotbepart
of a genuinely different explanation from another if the explanatory power of the former
dependsonthetruthofthelatter.Therefore:
Conclusion1b The fact that my friend won an award is not part of a genuinelydifferent explanationofwhy I congratulatedher from the fact that Ibelievedthatshehadwonanaward.
Thefollowingsectionssetoutthisargumentinmoredetail.
3.1 Impotence
IfIhadn’tbelievedthatmyfriendhadwonanawardIwouldnothavecongratulatedher(I’m
notasarcasticsort26).So,intheeventthatmyfriendhadwonanawardandIhadnotbelieved
that she had, the fact that my friend had won an award would not explain why I had
congratulatedher.Thus:
Premise1b If,ceterisparibus,Ihadnotbelievedthatmyfriendhadwonanaward(thoughshehad27)thenthefactthatshehadwonanawardwouldnothavepartiallyexplainedwhyIcongratulatedher.
Thiscaseisclear:hadInotbelievedthatmyfriendhadwonanawardthenthefactthatshe
hadwonanawardwouldnothaveexplainedwhyIcongratulatedherbecauseIwouldn’thave
congratulatedher.However,evenifIhadcongratulatedher,itwouldnothavebeenbecause
shewonanaward (absentsomeweirdcircumstances).For instance, suppose that Ibelieved
thatshehadjustgotanewjob,Imighthavestillcongratulatedherthen,butevensoitwould
bewrongtosaythatIcongratulatedherbecauseshehadwonanaward.
Thepointisthat,intheabsenceofanyweirdness28,somethingthatonewouldordinarilytake
tomakeone’sactionworthdoingcouldnotexplainwhyonedid itunlessonebelievedit.As
othersnote:
26And, ex hypothesi, I didn’t take anything else tomake congratulating her, in some respect, worthdoing.27Thisremarkisparentheticalbecauseitisalreadyimpliedbytheceterisparibuscondition.28Hornsbygivesthefollowingexampleofsuchweirdness:‘ConsiderGeorgewhoisquiteignorantoftheconditionoftheice…ItmightbethatGeorgeissociable,andskatesattheedgebecausethatiswheretheotherskatersare;anditmightthenbetruethatheskatesattheedgebecausetheiceinthemiddleisthin(thereisatwo-stepexplanationofGeorgesskatingtherewhichadducesthethinnessoftheice).’(Hornsby 2007, 296) The point is that I’m not saying that a fact cannever explain an agent’s actionunlesstheagentbelievesit.WhatIamestablishingisthatineitheroftheexamplesgiven(andwearekept‘in’thoseexamplesbytheceterisparibusclauses)–itdoesnotexplainmyactionunlessIbelieveit.
124
Afactcannotbeareasonthatexplainsone’sactionunlessthepersonisawareof it.(Alvarez2016a,30)
IfIactinthelightofthefactthatIammarried[i.e.becauseIammarried],ImustbelievethatIam.(Dancy2000,126)
So, the explanatory power of that which an agent believes typically depends upon their
believingit.
3.2 TheDifferencePrinciple
When is one proposition part of a genuinely different explanation fromanother? Recall the
caseof overexplanation considered in §(VII)4.3:we said that the fact that swimmingwould
improvemymoodandthefactthatswimmingwouldhelpmesleepbetterwereeachpartsof
genuinely different explanations of why swimming was, in some respect, worth doing. In
contrast,we said that the fact that Joanne’s roof leaks and the fact that it rained last night
werenotpartsofgenuinelydifferentexplanations.
Whydidwe reach these conclusions? Itwas because even if it stoppedbeing the case that
swimming would help me sleep better, the fact that it would improve my mood would
continue to explain why swimming was, in some respect, worth doing; and vice versa. In
contrast, if Joanne’s roof didn’t leak, then, ceteris paribus, the fact that it rained last night
wouldnotexplainwhyhercarpet iswet. Itwasbecauseof thisdifferencebetween the two
examples that we said the former involved genuinely different explanations, whereas the
latterdidnot.
Thepoint,Isuggest,isthis:apropositionisseeminglypartofagenuinelydifferentexplanation
fromsomeotherpropositiononlyifitsexplanatorypowerdoesnotdependonthetruthofthe
latter.Sothefactthatitrainedlastnightisnotpartofagenuinelydifferentexplanationofwhy
thecarpet iswetfromthefactthattheroof leaksbecause,ceterisparibus, if theroofdidn’t
leak then the fact that it rained lastnightwould stopexplainingwhy the carpetwaswet (it
wouldn’tbewetanymore).29Thus:
DIFFERENCE For any propositions p, q and r, p is part of a genuinely differentexplanationofthefactthatrfromqonlyif,ceterisparibus,hadpbeenthecaseandqnotbeenthecase,pwouldstillpartiallyexplainthefactthatr.
29Cf. ‘If the rationalizing explanation is dependenton thephysiological explanation in an appropriatesense(e.g.,bybeingreducibletoit),thenintruththereisonlyoneexplanationhere.’(Kim1989,80)
125
The difference principle connects the property of being part of a genuinely different
explanationfromsomeotherpropositionwiththepropertyofbeinglogically independentof
thatotherthatproposition.30
3.3 Concludingtheargumentfromimpotentfacts
Here,then, iswhat Ihavecalled ‘theargumentfromimpotentfacts’: thefactthatmyfriend
hadwonanawardwouldnothaveexplainedwhyIcongratulatedherifIhadn’tbelievedthat
she had. That being so, since some proposition is part of a genuinely different explanation
fromsomeotheronlyifitsexplanatorypowerdoesnotdependonthetruthofthelatter,the
fact thatmy friendwon an award is not part of a genuinely different explanation of why I
congratulatedherfromthefactIbelievedthatshehadwonanaward.
Formally:
TheArgumentfromImpotentFacts
Premise1b If,ceterisparibus,Ihadnotbelievedthatmyfriendhadwonanaward(thoughshehad)thenthefactthatshehadwonanawardwouldnothavepartiallyexplainedwhyIcongratulatedher.
DIFFERENCE For any propositions p, q and r, p is part of a genuinely differentexplanationofthefactthatrfromqonlyif,ceterisparibus,hadpbeenthecaseandqnotbeenthecase,pwouldstillpartiallyexplainthefactthatr.
Conclusion1b The fact that my friend won an award is not part of a genuinelydifferent explanationofwhy I congratulatedher from the fact that Ibelievedthatshehadwonanaward.
4 TheargumentforPremise1
I said that Premise1 can be inferred from the conclusion of the argument from false belief
togetherwiththeconclusionoftheargumentfromimpotentfacts,here’show:Theargument
fromfalsebelieftellsusthatthereisafullexplanation,callit‘Δ*’,ofwhyIcongratulatedmy
friendthat includesthefactthat Ibelievedthatshehadwonanawardbutnotthefactthat
shehadwonanaward.Theargumentfromimpotentfactstellsusthatthefactthatmyfriend
wonanawardisnotpartofagenuinelydifferentexplanationofwhyIcongratulatedherfrom
thefactthatIbelievedthatshehadwonanaward.31
Now,sinceΔ*isafullexplanationofwhyIcongratulatedmyfriendthatincludesthefactthatI
believedthatshehadwonanaward, theargumentfromimpotent factsmeansthatthefact
30IntheAppendixto§(X)InotethatthosewhomIcall‘inclusivedisjunctivists’rejectthisprinciple.31Thismeans that, forany fullexplanation,Δ,ofwhy I congratulatedmy friend that includes the factthat I believed thatmy friend hadwon an award, there is no full explanation that is both genuinelydifferentfromΔandincludesthefactthatmyfriendwonanaward.
126
thatmy friend won an award cannot be part of an explanation of why I congratulatedmy
friendthat isgenuinelydifferent fromΔ*.Therefore, there isa fullexplanation,Δ*,ofwhy I
congratulatedmyfriendsuchthatthefactthatmyfriendwonanawardisneitherapartofΔ*
norisitpartofafullexplanationthatisgenuinelydifferentfromΔ*.
Formally:
Conclusion1a There is a full explanation of why I congratulated my friend thatincludes the fact that I believed that shehadwonan awardbut notthefactthatshehadwonanaward.
Conclusion1b The fact that my friend won an award is not part of a genuinelydifferent explanationofwhy I congratulatedher from the fact that Ibelievedthatshehadwonanaward.
Premise1 ThereisafullexplanationofwhyIcongratulatedmyfriendsuchthatthe fact that she had won an award is neither a part of that fullexplanationnorisitpartofagenuinelydifferentexplanation.
TheformalargumentforPremise1issetoutinfullintheAppendixtothischapter.
5 TheExclusionPrinciple
Premise1 isthefirstpremiseofTheExplanatoryExclusionProblem.Thesecondpremise isa
finalprincipleofexplanation:theexclusionprinciple.Theexclusionprinciplesaysthatifsome
proposition is not part of a full explanation of some explanandum, and is not part of a
genuinely different explanation of that explanandum then it does not explain that
explanandum. In this section I set out the argument for the exclusion principle and, as an
aside,discussitsrelationtoKim’swell-knownprincipleofcausalexclusion.
5.1 Theargumentfortheexclusionprinciple
The reasoningbehind theexclusionprinciple is straightforward: if you say that youcan fully
explain some explanandum without mentioning p, then p can’t just be added to that full
explanation (because it would be superfluous – and MINIMALITY precludes superfluous
explanans),sopandthatfullexplanationcan’ttogetherbepartofthesamefullexplanation.
Moreover,ifp isalsonotpartofagenuinelydifferentfullexplanation,thenwearedrawnto
theconclusion that,by the lawofexcludedmiddle,p isnotapartofany fullexplanationof
thatexplanandum(sinceitisn’tpartofthesamefullexplanationandisn’tpartofagenuinely
different explanation). But if it isn’t part of any full explanation, then it isn’t a partial
explanation–whichmeans,asIsetoutinthepreviouschapter,itdoesnotexplain.
127
Thus, if there isa fullexplanationof someexplanandum thatdoes includesome fact then if
that fact is not part of a genuinely different explanation of that explanandum, it does not
explainit.Or,inotherwords:
EXCLUSION For any propositions,pandq, if there is a full explanation ofwhyqsuchthatp isneitherapartofthatfullexplanationnor is itpartofagenuinelydifferentexplanation,thenpdoesnotpartiallyexplainq.
ThisisthesecondpremiseofTheExplanatoryExclusionProblem.
5.2 Theexclusionprincipleandtheprincipleofcausalexclusion
As an aside, before we conclude, it is worth noting that the exclusion principle is a close
relation of a principle that is central to the exclusion problem inmental causation, namely,
Kim’sprincipleofcausalexclusion:
Ifaneventehasasufficientcausecatt,noeventattdistinctfromccanbeacauseofe(unlessthisisagenuinecaseofcausaloverdetermination).(Kim2008,17)
ThereareseveralsimilaritiesbetweenKim’sprincipleandmine.32However,despitemyuseof
the‘exclusionprinciple’label,therearealsosomesignificantdifferences.
OnedifferencethatisparticularlyworthstressingisthatwhileKim’sprincipleisrestrictedonly
totheconsideration(orexclusion)ofsimultaneousevents,thereisnoanalogousrestrictionin
myexclusionprinciple.Thisdifference isparticularlyworthstressingbecause it is thereason
why my argument against the exclusion principle (see §(XI)) does not also apply to Kim’s
principleofcausalexclusion.
32 In particular, much of the conceptual apparatus of Kim’s principle of causal exclusion has anexplanatory analogue in the conceptual apparatus I have used. For instance, Kim’s (1993, 280)distinction between partial and sufficient causes, is, I suggest, the causal analogue ofmy distinctionbetweenpartialandfullexplanation.ItisthuspossibletotransposeKim’sprincipleintoanexplanatoryanalogueofit,usingthestructuralprinciplesofexplanationIhaveassumed(itisperhapsworthnotingherethatKim(1988,233)originallyformulatedhisprincipleastheprincipleofexplanatoryexclusion).Doingsorevealsboththerespectsinwhichmyexclusionprincipleissimilartohis,andthoseinwhichitisnot.
128
6 TheExplanatoryExclusionProblemfor(R1)
Here, then, is the argument of The Explanatory Exclusion Problem against the claim that I
congratulatedmyfriendbecauseshewonanaward(i.e.against(R1)):
TheExplanatoryExclusionProblemfor(R1)
Premise1 ThereisafullexplanationofwhyIcongratulatedmyfriendsuchthatthe fact that she had won an award is neither a part of that fullexplanationnorisitpartofagenuinelydifferentexplanation.
EXCLUSION For any propositions,pandq, if there is a full explanation ofwhyqsuchthatp isneitherapartofthatfullexplanationnor is itpartofagenuinelydifferentexplanation,thenpdoesnotpartiallyexplainq.
Conclusion1 The fact that my friend had won an award does not explain why Icongratulatedher.
7 Conclusion
Ihavedemonstratedhowtwoseemingly trivialclaimsaboutwhatexplainsanagent’saction
whentheyactinerrorandinignorance,togetherwithfiveplausibleprinciplesofexplanation
can lead to the somewhat counterintuitive conclusion that I did not congratulatemy friend
becauseshehadwonanaward,butonlybecauseIthoughtshedid.
Indeed, as is presumably clear, this result should generalise beyond this example – The
ExplanatoryExclusionProblemisaproblemforanyonewhothinksthatfactsaboutthingsthat
areexternaltoourmindscanexplainwhywedothethingsthatwedo.So,forinstance,ifthe
Problemistobebelievedthenonenevertakesone’sumbrellabecauseit israining,butonly
becauseonebelievesthatit israining.Similarly,oneneverwaitstocrosstheroadbecausea
cariscoming,butonlybecauseonethinksacariscoming.
ThestandardresponsetoTheExplanatoryExclusionProblemistoaccepttheconclusionandto
insistthatwhenIsaythatIcongratulatedmyfriendbecauseshewonanaward,thepurported
explanansofthatexpression(thefactthatmyfriendhadwonanaward)ismerelyellipticalfor
therealexplanans,whichitconversationallyimplies(thatIkneworbelievedthathadshewon
anaward).Iwilldealwiththisresponse,andothercommonresponsesin§(X).Beforethen,in
thenextchapter,IwanttoshowhowTheExplanatoryExclusionProblemcanbeusedtocreate
furtherproblemsforexplanatoryrationalism.
129
Appendix
A.1 TheargumentforPremise1
Forreference,hereistheargumentforPremise1,infull:
Premise1a If,ceterisparibus,myfriendhadnotwonanaward(butIstillbelievedthat she had), then (i) the fact that I believed that she hadwon anawardwouldhavepartiallyexplainedwhyIcongratulatedher;and(ii)the fact that she had not won an award would not have partiallyexplainedwhyIcongratulatedher.
FACTIVITY Foranypropositionspandq,ifppartiallyexplainsthefactthatqthenpisthecase.
ENDURANCE For any propositionsp, qand r, the following holds: Suppose thatqpartiallyexplainsthefactthatrwhenitisnotthecasethatp.Supposefurtherthatneitherpnornotpispartofthesameexplanationofrasq. Then, if, ceteris paribus, it were the case that p, q would stillpartiallyexplainthefactthatr.
SUFFICIENCY Foranypropositionq,andanyset,Δ, ifΔ isa fullexplanationof thefact thatq in somecircumstance, then, inanycircumstance inwhichall theelementsofΔpartiallyexplain the fact thatq,Δ fullyexplainsthefactthatq.
Premise1b If,ceterisparibus,Ihadnotbelievedthatmyfriendhadwonanaward(thoughshehad)thenthefactthatshehadwonanawardwouldnothavepartiallyexplainedwhyIcongratulatedher.
DIFFERENCE For any propositions p, q and r, p is part of a genuinely differentexplanationofthefactthatrfromqonlyif,ceterisparibus,hadpbeenthecaseandqnotbeenthecase,pwouldstillpartiallyexplainthefactthatr.
Premise1 ThereisafullexplanationofwhyIcongratulatedmyfriendsuchthatthe fact that she had won an award is neither a part of that fullexplanationnorisitpartofagenuinelydifferentexplanation.
130
(IX)
OtherUsesforTheExplanatoryExclusionProblem
InwhichIshowhowTheExplanatoryExclusionProblemcanbeusedtoarriveatsomeotherconclusionsthatareinconvenientforexplanatoryrationalism.Isetout the general form of the Problem, followed by the general form of theargumentforthefirstpremiseoftheProblem.IshowtheProblemcanbeusedtoargue that the fact that I read thatmy friendhadwonanawarddoesnotexplainwhyIcongratulatedher,andthatneitherthatfact,northefactthatshehad won an award, can explain why it was pro tanto rational for me tocongratulateher.
Recall that in §(VI), I said that if explanatory rationalism is to be consistentwith theprima
faciereasonableclaimssetoutin§§(II)-(IV),thefollowingmustbetrue:
(R1) Icongratulatedmyfriendbecauseshehadwonanaward.
(R2) IcongratulatedmyfriendbecauseIreadthatshehadwonanaward
(R3) Itwaspro tanto rational forme to congratulatemy friend because she hadwonanaward.
(R4) ItwasprotantorationalformetocongratulatemyfriendbecauseIreadthatshehadwonanaward.
In the previous chapter I showed how The Explanatory Exclusion Problem provides an
argumentagainst(R1).ThepurposeofthischapteristoshowthatTheExplanatoryExclusion
Problem also provides an argument against (R2), (R3) and (R4), by using it to reach the
followingconclusions:
Conclusion2 ThefactthatIreadthatmyfriendhadwonanawarddoesnotexplainwhyIcongratulatedher.
Conclusion3 Thefactthatmyfriendwonanawarddoesnotexplainwhyitwasprotantorationalformetocongratulateher.
Conclusion4 ThefactthatIreadthatmyfriendhadwonanawarddoesnotexplainwhyitwasprotantorationalformetocongratulateher.
131
1 ThegeneralformofTheExplanatoryExclusionProblem
ThegeneralformofTheExplanatoryExclusionProblemfortheclaimthatsomeproposition,x,
explainssomeproposition,z,isasfollows:
ThegeneralformofTheExplanatoryExclusionProblem
Premise# Thereisafullexplanationofwhyzsuchthatxisneitherapartofthatfullexplanationnorisitpartofagenuinelydifferentexplanation.
EXCLUSION For any propositions,pandq, if there is a full explanation ofwhyqsuchthatp isneitherapartofthatfullexplanationnor is itpartofagenuinelydifferentexplanation,thenpdoesnotpartiallyexplainq.
Conclusion# xdoesnotexplainwhyz.
TheExplanatoryExclusionProblemcanthusprovidetheargumentforConclusions2,3,and4,
ifweprovidetheappropriatespecificationof‘Premise#’.Howdowedoso?
1.1 ThegeneralformoftheargumentforPremise#
IntheargumentforPremise1ofthepreviouschapter,theonlypremisesthatwerespecificto
theexampleconsideredwerethese:
Premise1a If,ceterisparibus,myfriendhadnotwonanaward(butIstillbelievedthat she had), then (i) the fact that I believed that she hadwon anawardwouldhavepartiallyexplainedwhyIcongratulatedher;and(ii)the fact that she had not won an award would not have partiallyexplainedwhyIcongratulatedher.
Premise1b If,ceterisparibus,Ihadnotbelievedthatmyfriendhadwonanaward(thoughshehad)thenthefactthatshehadwonanawardwouldnothavepartiallyexplainedwhyIcongratulatedher.
Thegeneralformofthesepremisesis,fortheparticularpropositionsx,yandz,asfollows:
Premise#a If,ceterisparibus,xhadnotbeenthecase(butystillhad)then(i)thefactthatywouldhavepartiallyexplainedwhyz;and(ii)thefactthatnotxwouldnothavepartiallyexplainedwhyz.
Premise#b If, ceteris paribus,yhadnot been the case (but x still had) then thefactthatxwouldnothavepartiallyexplainedwhyz.
AlloftheotherpremisesintheargumentforPremise1were,youwillrecall,generalprinciples
ofexplanation.Asaresult,ifPremise#aandPremise#baretrueofx,yandz,thenPremise#
istrueofthemtoo(givenFACTIVITY,SUFFICIENCY,ENDURANCEandDIFFERENCE);andifPremise# is
trueofthemthenTheExplanatoryExclusionProblemimpliesthatxdoesnotexplainz.So,to
arrive at Conclusions 2, 3, and 4 we need only show that the appropriate specifications of
132
Premise #a and Premise #b are true of those cases. I consider the argument for each
conclusioninturn.
2 TheExplanatoryExclusionProblemfor(R2)
Doesthefactthat I readthatmyfriendhadwonanawardexplainwhy Icongratulatedher?
The Explanatory Exclusion Problem for (R2) concludes that it does not. To reach that
conclusionweneedtoestablishthefollowing:
Premise2 ThereisafullexplanationofwhyIcongratulatedmyfriendsuchthatthefactthatIreadthatmyfriendhadwonanawardisneitherapartof that full explanation nor is it part of a genuinely differentexplanation.
Now, as discussed, in order to arrive at Premise2 we need only demonstrate that the
appropriatespecificationsofPremise#aandPremise#baretrue.That is thepurposeof the
followingsections.
2.1 TheargumentforPremise2a
Firstly, suppose that, ceteris paribus, I hadn’t read that she had won an award, but I still
believed that she had –maybe I saw her win it, or heard about it from another friend, or
maybe (incredibly) I acquired the belief as the result of a brain aneurism. In such a
circumstance,wouldIstillhavecongratulatedher?OfcourseIwould!Ithoughtthatshe’dan
award!And,Isubmit,IwouldhavecongratulatedherbecauseIbelievedthatshehadwonan
award.
Moreover, continuing to suppose that, ceteris paribus, I hadn’t read that she had won an
award(butnonethelessbelievedthatshehad),wouldthefactthatIhadn’treadthatshehad
won an award explainwhy I congratulated her? Surely not!Whywould it? This is a prosaic
case,notaweirdone.
Thus,combiningthesetwoinsights,wearriveatthefollowingspecificationofPremise#a:
Premise2a If,ceterisparibus,Ihadnotreadthatmyfriendhadwonanaward(butI still believed that shehad) then (i) the fact that Ibelieved that shehadwonanawardwouldhavepartiallyexplainedwhyIcongratulatedher; and (ii) the fact that I hadnot read that shehadwonanawardwouldnothavepartiallyexplainedwhyIcongratulatedher.
2.2 TheargumentforPremise2b
Nowconsider:if,ceterisparibus,IhadnotbelievedthatshehadwonanawardeventhoughI’d
read that she had in the newspaper (perhaps I’m sceptical of the mainstream media, or
133
jealousymakesmewithhold),wouldIhavecongratulatedher?OfcourseIwouldn’t:aswe’ve
alreadyestablished.Ididn’tthinkthatshe’dwonanaward,soitwouldhavebeenoddofme
tocongratulateher (again, Ididn’t seeanythingelseofworth incongratulatingher,and I’m
notasarcasticsort).
But if Iwouldn’t have congratulated her then the fact that I had read that she hadwon an
award wouldn’t have explained why I congratulated her (since I wouldn’t have, and
explanandamustbethecase).Thus:
Premise2b If,ceterisparibus,Ihadnotbelievedthatmyfriendhadwonanaward(althoughIhadreadthatshehadwonanaward)thenthefactthat Ihad read that she had won an award would not have partiallyexplainedwhyIcongratulatedher(sinceIwouldn’thave).
2.3 TheargumentforPremise2
To run through theargument, for clarity:weknow, fromcondition (i)ofPremise2a, that if,
ceterisparibus,Ihadnotreadthatmyfriendhadwonanaward(buthadstillbelievedthatshe
had won an award) then there would have been a full explanation, call it ∆*, of why I
congratulated her that would have included the fact that I believed that she had won an
award.And,fromFACTIVITY,weknowthathadInotreadthatshehadwonanaward,the(false)
propositionthatIreadthatshehadwonanawardcouldnothavebeenapartof∆*.
Wealsoknow,fromcondition(ii)ofPremise2a,thatthefactthat Ididn’t readthatshehad
wonanawardwouldnothavebeenapartof∆*.So,since(hadInotreadthatshehadwonan
award) neither the (false) proposition that I read that she hadwon an awardnor its (true)
negationwouldhavebeenelementsof∆*,weknow,fromENDURANCE,thatalltheelementsin
∆*mustalsohaveexplainedwhyIcongratulatedmyfriendinthecaseinwhichIdidreadthat
shehadwonanaward.So,fromSUFFICIENCY,weknowthat∆*islikewiseafullexplanationof
whyIcongratulatedherwhenIdidreadthatshehadwonanaward.
Now,fromPremise2b,weknowthat if,ceterisparibus, Ihadn’tbelievedthatmyfriendhad
won an award then the fact that I had read that she had won an award would not have
explainedwhyIcongratulatedher(sinceIwouldn’thave).So,fromDIFFERENCE,weknowthat
the fact that I read that she had won an award cannot be part of a genuinely different
explanation of why I congratulated her from the fact that I believed that she had won an
award.Andsince∆* includes the fact that Ibelieved that shewonanaward, the fact that I
readthatshehadwonanawardcannotbepartofanexplanationofwhyIcongratulatedher
thatisgenuinelydifferentfrom∆*.Therefore:
134
Premise2 ThereisafullexplanationofwhyIcongratulatedmyfriendsuchthatthefactthatIreadthatmyfriendhadwonanawardisneitherapartof that full explanation nor is it part of a genuinely differentexplanation.
2.4 TheExplanatoryExclusionProblemfor(R2)
So,theProblemfor(R2)isasfollows:
TheExplanatoryExclusionProblemfor(R2)
Premise2 ThereisafullexplanationofwhyIcongratulatedmyfriendsuchthatthefactthatIreadthatmyfriendhadwonanawardisneitherapartof that full explanation nor is it part of a genuinely differentexplanation.
EXCLUSION For any propositions,pandq, if there is a full explanation ofwhyqsuchthatp isneitherapartofthatfullexplanationnor is itpartofagenuinelydifferentexplanation,thenpdoesnotpartiallyexplainq.
Conclusion2 ThefactthatIreadthatmyfriendhadwonanawarddoesnotexplainwhyIcongratulatedher.
3 TheExplanatoryExclusionProblemfor(R3)
Does the fact thatmy friendwonan awardexplainwhy itwaspro tanto rational forme to
congratulateher?TheExplanatoryExclusionProblem for (R3) concludes that itdoesnot. To
reachthatconclusionweneedtoestablishthefollowing:
Premise3 There isa fullexplanationofwhy itwaspro tanto rational formetocongratulatemy friend such that the fact that shewon an award isneither a part of that full explanation nor is it part of a genuinelydifferentexplanation.
Inthefollowingsections IwilldemonstratethattheappropriatespecificationsofPremise#a
andPremise#baretrue.
3.1 TheargumentforPremise3a
If,ceterisparibus,myfriendhadnotwonanaward,butIstillbelievedthatshehad,wouldit
stillhavebeenpro tanto rational forme tocongratulateher? I suggest that itwould: inany
normalcircumstancesifyouthinkthatyourfriendhaswonanawardtherationalthingtodois
tocongratulateher–ifyoudidn’tthenyouwouldbeactingirrationally(unless,say,youwere
veryjealous,orknewthatshedoesn’tliketobecongratulated–butthat’snotmyexample).
Whydoesn’tthefalsityofmybeliefseemtomatter?Itisbecause,asWedgwoodnotes:
Whenweassessachoiceordecisionasrationalorirrational,weareassessingitonthebasisofitsrelationtotheagent’sbeliefs,desires,andothersuchmentalstates–notonthebasisofitsrelationtofactsabouttheexternalworldthatcouldvarywhilethosementalstatesremainedunchanged.(Wedgwood2002,350)
135
Forinstance,wealreadyacknowledged(see§(II)1.1)thatitwasrationalforSallytorungiven
thatshebelievedthatabearwaschasingher1;furthermore,itwasrationalinspiteofthefact
thatnobearwaschasingher.Whatmatterstotheprotantorationalityofanactionis,asI’ve
suggested,thattheagenttakestheretobesomethingofworthindoingit–notthatitactually
is,insomerespect,worthdoing.
So,if,ceterisparibus,myfriendhadnotwonanaward,butIstillbelievedthatshehad,thenit
wouldhavebeenprotantorationalformetocongratulateher,andthatwouldhavebeen,in
part,becauseIbelievedthatshehadwonanaward.
Moreover, in this counterfactual case, the fact thatmy friendhadnotwonanawardwould
clearlynothaveexplainedwhyitwasprotantorationalformetocongratulateher.Again,the
caseis,exhypothesi,prosaicandnotweird.Thus:
Premise3a If,ceterisparibus,myfriendhadnotwonanaward(butIstillbelievedthat she had), then (i) the fact that I believed that she hadwon anawardwouldhavepartiallyexplainedwhyitwasprotantorationalformetocongratulateher;and(ii)thefactthatshedidnotwinanawardwould not have partially explainedwhy itwaspro tanto rational formetocongratulateher.
3.2 TheargumentforPremise3b
Nowconsider: if,ceterisparibus, Ihadnotbelievedthatshehadwonanawardeventhough
shehad,would itstillhavebeenprotantorational formetocongratulateher?Perhapsyou
thinkthattheanswerdependsonwhetherornotitwasrationalformenottobelievethatshe
hadwonanaward?Iwillreturntothequestionofwhetherornotthatmatters in§4.1,but
suppose,fornow,thatitwas.
IfIdidn’tbelievethatshehadwonanawardanddidn’ttakecongratulatinghertobe,inany
other respect,worth doing, then itwould not have been evenpro tanto rational forme to
congratulateher.Norational(non-sarcastic)person,withsuchbeliefsanddesires(etc.)would
congratulatetheirfriend.Thus:
Premise3b If,ceterisparibus,Ihadnotbelievedthatmyfriendhadwonanaward(thoughshehad)thenthefactthatshehadwonanawardwouldnothave partially explained why it was pro tanto rational for me tocongratulateher(sinceitwouldn’thavebeenprotantorationalformetocongratulateher).
1Ifyouareconcernedaboutwhetherornotherbeliefisrationalinthefirstplace,pleaseforestallthoseconcernsuntil§4.1.
136
3.3 TheExplanatoryExclusionProblemfor(R3)
Trusting that the reasoning isnow familiar, Iwill spare the readerademonstrationofhow I
thinkwecanarriveatPremise3fromPremise3aandPremise3b.So,theProblemfor(R3)is
asfollows:
TheExplanatoryExclusionProblemfor(R3)
Premise3 There isa fullexplanationofwhy itwaspro tanto rational formetocongratulatemy friend such that the fact that shewon an award isneither a part of that full explanation nor is it part of a genuinelydifferentexplanation.
EXCLUSION For any propositions,pandq, if there is a full explanation ofwhyqsuchthatp isneitherapartofthatfullexplanationnor is itpartofagenuinelydifferentexplanation,thenpdoesnotpartiallyexplainq.
Conclusion3 Thefactthatmyfriendwonanawarddoesnotexplainwhyitwasprotantorationalformetocongratulateher.
4 TheExplanatoryExclusionProblemfor(R4)
Finally:doesthe fact that I readthatmyfriendhadwonanaward in thenewspaperexplain
whyitwasprotantorationalformetocongratulateher?TheExplanatoryExclusionProblem
for(R4)saysitdoesnot.Thisiswhatweneedtoshowtogetthere:
Premise4 There isa fullexplanationofwhy itwaspro tanto rational formetocongratulatemyfriendsuchthatthefactthatIreadthatshehadwonan award is neither a part of that full explanation nor is it part of agenuinelydifferentexplanation.
4.1 TheargumentforPremise4a
Suppose, as we did in §2.1, that ceteris paribus, I hadn’t read that my friend had won an
award,butIstillbelievedthatshe’dwonanaward.IfIhadn’treadthatshe’dwonanaward,
but,say,I’dseenherwinit,woulditstillhavebeenprotantorationalformetocongratulate
her?Ofcourseitwould.Likewise,ifI’dheardaboutherawardfroma(reliable)frienditwould
havebeenprotantorationalformetocongratulateher.ButwhatifIacquiredthebeliefasthe
result of brain aneurism?Would it still havebeenpro tanto rational forme to congratulate
her?Somethinkitwouldnot,forinstance:
Ifanagenthasirrationalbeliefs,thosebeliefsarenotabletomakerationalanyactionsdoneintheirlight.(Dancy2000,60)
ItwouldnotberationalforHollytoputonwinterclothesifherbeliefthatit issnowingweredue to crazed conviction, say, or wishful thinking. Irrationality cannot beget rationality! Asubject’s beliefs contribute tomaking it rational for her to act in certain ways only if thosebeliefsarethemselvesrational.(Whiting2014,4)
137
If this is so, then it seems that the way that one acquires one’s beliefs is relevant to the
rationality of one’s actions; that is, we cannot so easily omit mention of them in the
explanationofwhyitisrationalforanagenttodosomeaction.
Insomerespect,Idisagree–Ithinkthatanirrationalbeliefcannonethelessexplainwhyitwas
pro tanto rational (but probably not all things considered rational) for someone to do
something,2asdoothers.3Nonetheless,evenifirrationalbeliefscan’texplainwhyactionsare
pro tanto rational, that does not mean that the experiences or appearances on which an
agent’sbeliefsarebasedarenecessarytoafullexplanationofwhyitisprotantorationalfor
thatagenttodosomeaction;itmeansonlythatthefactthattherelevantbeliefsoftheagent
arerational4isnecessarytoafullexplanation.
So, ifceterisparibus, Ihadn’treadthatmyfriendhadwonanaward,but Istillbelievedthat
she’dwonanawardandthatbeliefwas(still)rational,wouldithavebeenprotantorational
forme to congratulatemy friend?Of course!And itwouldhavebeenpro tanto rational, in
part,becauseIbelievedthatshehadwonanaward.
Moreover, the fact that I didnot read thatmy friend hadwon an award clearlywould not
explainwhyitwouldhavebeenprotantorationalformetocongratulateher.Thus:
2Anargument to thateffect: suppose thatBernardOrtcutt, spyextraordinaire, comes tobelieve thattheFBIhasdiscoveredthatheisaspyandhassentagentstoapprehendhim.Hisbeliefiswell-founded–FBIcounter-intelligenceexiststocapturespieslikehim;anormallyregularassethasgonemissing;andhe’sbeenrepeatedlytrailedbyablacksedanthisweek.Itis,Isubmit,protantorationalforBernardtogointohiding(sinceheisapatriot(soisn’tmindedtoturn),andhasnointerestinjailtime,itislikelyalsoall things considered rational). Now considerOrnard Bertcutt: Ornard has actually lived a ratherpedestrianlifebut,throughsomefreakco-incidence(apeculiarmentaldisorder,say),hismentalstatesareall identicaltoBernard’s–heisBernard’smentalduplicate.So,Ornard,likeBernard,believesthattheFBIisouttogethim.However,Ornard’sbeliefisnotonlyfalse,itisplainlynotrational.Nonetheless,it is, I submit, at least pro tanto rational for him to go into hiding. Some resistance to this view isunderstandable:tobelievethattheFBIischasingyouiscertainlyoutlandish,andOrnardonlybelievesitbecauseheiscrazy–sosurelygoingintohidingcan’tbeevenaprotantorationalthingforhimtodo?IfOrnard goes into hiding it’s because he’s not rational, it’s not the rational thing for him to do! Theproblem,however,withsayingthatitisn’tevenprotantorationalforOrnardtogointohidingisthatweareseemingly forcedtoquestionwhetherornot it isprotantorational forBernardtogo intohiding.Why?Becauseof thewidelyheldviewthatwhat it is rational foranagent todosuperveneson theirmentalstates(e.g.Broome2013,151),whichistosaythattherecanbenochangeinwhatitisrationalfor an agent to do without a change in their mind. Now, since Bernard and Ornard are mentalduplicates,giventhatwhatitisrationalforanagenttodosupervenesontheirmentalstates,therecanbenodifferencebetweenBernardandOrnard inwhat it is rational for them todo, since there isnodifferenceintheirbrainstates.3‘Givenmyirrationalbeliefthatsmokingwillprotectmyhealth,itwouldberationalformetosmoke.Given this hermit’s irrational belief that his life of self-inflicted pain would please God, he couldrationallylivesuchalife.’(Parfit2011,114)4Whateveryourpreferredstandardofrationalityforbeliefsis.
138
Premise4a If,ceterisparibus,Ihadnotreadthatmyfriendhadwonanaward(butI still believed that she had won an award) then (i) the fact that Ibelieved that she hadwon an awardwould have partially explainedwhy itwasprotanto rational formetocongratulateher;and (ii) thefact that I hadnot read that shehadwonanawardwouldnothavepartiallyexplainedwhyitwasprotantorationalformetocongratulateher.
4.2 TheargumentforPremise4b
Finally:if,ceterisparibus,IhadnotbelievedthatshehadwonanawardeventhoughI’dread
thatshehad,wouldithavebeenprotantorationalformetocongratulateher?Again,perhaps
youthinkitdependsonwhetherornotitwasrationalformenottobelievethatshehadwon
anaward.Andperhaps, further, you insist thatgiven theceterisparibus clause it can’thave
beenrational.Well,Idisagree.
Supposing that pathological jealousymakesmewithhold. The all things considered rational
thingformetodoistoceasewithholding(andperhapsseektreatment).Then,onceI’vedone
that, itwouldbepro tantorational formetocongratulatemy friend.But,giventhat Idon’t
believethatshehaswonanaward(that is,before Iceasewithholding), Isuggest, itcan’tbe
protantorationalformetocongratulateher.Iseenothingofanyworthindoingso–andnot
becauseIdon’tlikeher,ordon’tcareaboutherfeelings–butbecauseIdon’tbelievethatshe
has won an award. To insist that it is even pro tanto rational for me to congratulate her,
despitethefactthatIdon’tbelievethatshehaswonanaward,istoinsistthatitisprotanto
rationalformetodosomethingthatItaketobe,innorespect,worthdoing.Idon’tseehow
thatcouldberational.
So,giventhatitwouldn’thavebeenprotantorationalformetocongratulateher,thefactthat
Ireadthatshehadwonanawardcouldnotexplainwhyitwouldhavebeenprotantorational
formetocongratulateher.Thus:
Premise4b If,ceterisparibus,Ihadnotbelievedthatmyfriendhadwonanaward(althoughIhadreadthatshehadwonanaward)thenthefactthat Ihad read that she had won an award would not have partiallyexplainedwhy I itwaspro tanto rational forme to congratulateher(sinceitwouldn’thavebeenprotantorationalformetocongratulateher).
4.3 TheExplanatoryExclusionProblemfor(R4)
Again, I will not spell out the reasoning from Premise 4a and Premise 4b to Premise4. So,
concluding,theProblemfor(R4)is:
139
TheExplanatoryExclusionProblemfor(R4)
Premise4 There isa fullexplanationofwhy itwaspro tanto rational formetocongratulatemyfriendsuchthatthefactthatIreadthatshehadwonan award is neither a part of that full explanation nor is it part of agenuinelydifferentexplanation.
EXCLUSION For any propositions,pandq, if there is a full explanation ofwhyqsuchthatp isneitherapartofthatfullexplanationnor is itpartofagenuinelydifferentexplanation,thenpdoesnotpartiallyexplainq.
Conclusion4 ThefactthatIreadthatmyfriendhadwonanawarddoesnotexplainwhyitwasprotantorationalformetocongratulateher.
5 TheArgumentfromIllusion
Many have observed 5 a similarity between the problem that I have formalised as The
ExplanatoryExclusionProblemandtheargumentfromillusionintheliteratureonperception.
Totheextent thatweconceiveof theargument from illusionasaproblemfor the ideathat
the external world could explainwhywe believewhatwe believe then the argument from
illusion can be represented as an instance of The Explanatory Exclusion Problem, briefly, as
follows.
IseeatomatosoIbelievethatthereisatomatobeforeme.Ifnotomatohadbeenbeforeme,
butitstillappearedtomeasthoughonehad,thenthefactthatitappearedtomeasthougha
tomatowere therewouldpartiallyexplainwhy Ibelieved thata tomatowas there.And the
fact that there wasn’t a tomato wouldn’t. This provides the relevant specification of
Premise#a.
Andtheniftherehadbeenatomatobutithadn’tappearedtomeasthoughtherehad(blind
spotsinmyvision,say),thenIwouldn’thavebelievedthattherewasatomato,sothefactthat
therewasatomatowouldn’thaveexplainedwhy Ibelievedthattherewas (since Iwouldn’t
havebelievedthattherewas).ThisprovidesuswiththerelevantspecificationofPremise#b.
FromthesespecificationsofPremise#aandPremise#b,togetherwiththerelevantprinciples
of explanation,we can, in amanner that should now be familiar, arrive at The Explanatory
ExclusionProblem,andtheconsequentconclusionthatthefactthatthereisatomatobefore
doesnotexplainwhyIbelievethatthereis.
However,thereareotherversionsoftheargumentfromillusionaccordingtowhichitisnota
claim about what explains an agent’s beliefs. For instance, one interpretation of it is as a
problem for the idea that veridical perceptual experience provides a basis for knowledge.5(E.g.Stout1996;Dancy2000;Hornsby2008;Hyman2011;McDowell2013)
140
When the argument from illusion is conceived in that manner, it cannot be so simply
characterised as The Explanatory Exclusion Problem. Although, I submit, the problems are
nonethelessrelated.
6 Conclusion
I have now shown how The Explanatory Exclusion Problem can provide arguments against
(R1)-(R4).
Sinceexplanatory rationalismrequires the truthof (R1)-(R4) if it is tobeconsistentwith the
primafaciereasonableclaimssetoutinearlierchapters,Iwillneedtofindsomewaytoreject
the conclusions of The Explanatory Exclusion Problem. My solution will be to reject the
exclusionprinciple,which,Iwillargue,bymeansofseveralcounterexamples,isclearlyfalse.I
willargue, interalia,thatthefactthatmyfriendwonanawardexplainswhyIcongratulated
herbecauseitexplainswhyIbelievedthatshewonanaward,andtheexclusionprinciplefails
toprovideforthetransitivityofthatsortofexplanation.
Before then, however, I wish to consider the other, more commonplace responses to The
Explanatory Exclusion Problem, andwhat iswrongwith them. That is the focus of the next
chapter.
141
(X)
Hownormativereasonsdon’texplain
InwhichIrejecttwoaccountsofhownormativereasonsexplain.Ire-introducetalkofnormativereasons,definingthemasthingsthatmakeactions, insomerespect, worth doing. I ask how it is that wemanage to explain our actionswhenwesaythatweactedbecauseofanormativereasontherewastoact;forinstance:howisitthatIexplainwhyItookmyumbrellawhenIsaythatItookitbecauseitwasraining?IsuggestthatthefactthatitwasrainingexplainswhyItookmyumbrellaeither ‘elliptically’, ‘directly’or ‘indirectly’. Inote thatwhichansweroneacceptswilldependonone’sresponsetoTheExplanatoryExclusionProblem: elliptical theorists accept the conclusion of the Problem, directtheorists reject the firstpremise,and indirect theorists reject thesecond. I setouttheproblemswithellipticalanddirecttheories.
Weoftenexplainwhywedosomethingbycitingsomefactthatcountedinfavourofdoingit:I
saythatItookmyumbrellabecauseitwasraining;Marshallsaysthatheisgoingtothestation
because his daughter is on the 7 o’clock train. How are we to make sense of these
commonplaceexplanations?
IfTheExplanatoryExclusionProblemistobebelieved,weshouldnotinterpretthemliterally.
That is, thepurported explanantia (i.e. thatwhich follows the ‘because’) in such statements
arenottheactualexplanantia;thefactthatitisrainingdoesnotreallyexplainwhyItookmy
umbrella.Instead,whateverexplanatorypowerthesestatementshaveisduetotherebeinga
short-hand;when I say that I tookmy umbrella because itwas raining, the fact that itwas
rainingismerelyellipticalforwhatreallydoestheexplaining,whichisthefactthatIbelieved
thatitwasraining,orthatIknewthatitwas.
Alternatively,onecouldrejecttheconclusionofTheExplanatoryExclusionProblem,andinsist
that these remarks are literally accurate; the fact that it is raining really does explainwhy I
tookmyumbrella.Therearetwowaysofdoingthis:eitherbyrejectingthefirstpremiseofthe
Problem,orthesecond.
Theoriesthatrejectthefirstpremisesaythatthecontributionofthefactthat it israiningto
theexplanationofwhyItookmyumbrellais,insomesense,independentfrom,orinaddition
totheexplanatorycontributionofthefactthatIbelievedthatitwasraining.Theyarguethat
the fact that it is raining explainsmy actiondirectly (that is, unmediated by features ofmy
psychology).Accordingtothemostpopulartheories,thedirectexplanatoryrelationbetween
142
theworldandtheactionistheresultofthespecialconnectionbetweentheworldandactions
thatknowledgeengenders.
Incontrast, theoriesthatrejectthesecondpremiseacceptthattheexplanatorycontribution
ofthefactthatitwasrainingisnotinadditiontotheexplanationthatisalreadyprovidedby,
inter alia, the fact that I believe that it was raining. Thus they deny that the fact that it is
raining directly explains why I tookmy umbrella. However, they insist, the fact that it was
rainingdoesindirectlyexplainwhyItookmyumbrella,byexplainingwhyIbelievedthatitwas
raining.
Inshort:wewanttounderstandhowitisthatImanagetoexplainmyactionwhenIsay,‘Itook
myumbrellabecause itwas raining.’ There are threepossible accounts: the fact that itwas
rainingeitherexplainsmyactionelliptically,oritexplainsitdirectly,oritexplainsitindirectly.
Inwhat follows Iwill set out the problemswithelliptical anddirect theories,which are the
typicalresponsestothisproblem.InsubsequentchaptersIwilldefendmyownindirecttheory.
Before then it will helpme better characterise what is at stake in this discussion if we re-
introducetalkof‘normativereasons’;thatisthefocusofthefirstsection.
1 Normativereasonexplanations
1.1 Normativereasons
Iwanttore-habilitatetheterm‘normativereason’,whichIabandonedin§(I).Letusdefineit
asfollows:
Definition Foranyp,pisanormativereasonforAtoφifandonlyifpmakesA’sφing,insomerespect,worthdoing.1
Afewpoints,alreadydiscussedin§(I)4,areworthstressinghere:Firstly,whilethisdefinition
iscloselyalignedtotheconventionaldefinitionofnormativereasons, itdepartsfromit inso
farasIamnotsayinganythingabouthownormativereasons,sodefined,relatetoexpressions
like‘thereasonstherearetoact’or‘thereasonsforwhichanagentacted’–formypurposes
theterm‘normativereason’isstrictlyatermofartmeaninganythingthatmakesanaction,in
somerespect,worthdoing.
Secondly,Iwillassume,toavoidambiguity,that‘countinginfavourof’and‘making, insome
respect, worth doing’ are equivalent relations. This assumption has no bearing on the
argumentofthischapter.
1Note:thisdefinitionmakesnoexplicitassumptionsabouttheontologyof‘p’.
143
Finally,itisworthrecallingthatthatwhichcountsinfavourofanaction(i.e.thatwhichmakes
it,insomerespect,worthdoing),istypicallynotafeatureoftheagent’spsychology,whichis
tosaythatnormativereasonsaretypicallynotfeaturesofanagent’spsychology.
1.2 Normativereasonexplanations
In§(VII)1,Inotedthattheword‘explains’and‘explanation’havedifferentmeanings.Ifwesay
thatthefact that it rained lastnight isanexplanationofwhyJoanne’scarpet iswet,weare
givingonesenseof‘explanation’,inwhichanexplanationisanexplanans.Thisisthesenseof
‘explanation’and‘explains’thatIhavefocussedoninpreviouschapters–itisthesensethatis
involvedintheconceptsoffullandpartialexplanation.
However,Isuggestthat‘Itookmyumbrellabecauseitwasraining’isanexplanationofwhyI
tookmyumbrellaindifferentsenseoftheword–thesentenceisanelucidationofwhyItook
my umbrella, it is not the explanans of why I took it. It is in that sense of the word
‘explanation’thatIsuggestthatwecallsentencesinwhichanormativereasonforanagentto
act appears in the position of an explanans of why they acted, 2 ‘normative reason
explanations’.
It is a fact that we often give normative reason explanations of our actions. I gave a few
examplesinmyopeningremarks,herearesomemore3:Sandraisgoingtotheshopsbecause
she is out ofmilk; I’m flying to Bodrumbecause that’swheremy father lives; TheresaMay
madeadealwiththeDUPin2017becausethatwastheonlywayforhertoformamajority
government.4In all of these examples, something that made the action, in some respect,
worthdoing(forthatagent),whichistosay,anormativereasonforthemtodoit,appearsto
explaintheiraction.And,indeed,whensuchexplanationsaregiven,youunderstandwhythe
actionwasdone–soanexplanationofsomesorthascertainlybeenprovided.
2 Theoriesofnormativereasonexplanation
IntheintroductiontothischapterIaskedhowitwasthatImanagedtoexplainmywhyItook
myumbrellawhen Isaidthat I tookmyumbrellabecause itwasraining.Theremarksof the
previous section should have made clear that this is a specific instance of a more general
2Thatis,sentenceslike‘Aφ’dbecausep’or‘thefactthatpexplainswhyAφ’d’wherethefactthatpisanormativereasonforAtoφ.3 For the following examples, assume that the relevant supporting conditions (desires, evaluativejudgements,evaluativefacts…)areinplacesuchthat:thefactthatSandraisoutofmilkisanormativereasonforhertogototheshops;thefactthatmyfatherlivesinBodrumisanormativereasonformeto go there; the fact thatmaking a dealwith theDUPwas the onlyway forMay to form amajoritygovernmentwasanormativereasonforhertodoso.4Ofcourse,thepurportedexplanantiaoftheseremarksare,atbest,partialexplanations.
144
question: how do we manage to explain our actions when we give a normative reason
explanation of them? Answering this question is the job of what I will call a ‘theory of
normativereasonexplanation’.
Which theoryofnormative reasonexplanationoneholdsdependsonone’s response toThe
Explanatory Exclusion Problem: thosewho accept the conclusion of the Problem insist that
when we give a normative reason explanation it is not really the normative reason that
explainsouraction.Instead,theyargue,thenormativereasonisellipticalforthatwhichreally
doestheexplaining–whichistheagent’sawarenessof,orbeliefinthenormativereason.This
istheellipticaltheoryofnormativereasonexplanation.
ThosewhorejectthefirstpremiseoftheProbleminsistthat,whenwegiveanormativereason
explanation of an agent’s action, the normative reason explains the agent’s action directly;
that is, the explanatory relations involved are unmediated by features of an agent’s
psychology.Thesearedirecttheoriesofnormativereasonexplanation.
Finally,thosewhorejectthesecondpremiseoftheProblemaccepttheprimacyoffeaturesof
theagent’spsychologyinexplainingtheiraction,butnonethelessinsistthatnormativereasons
canexplainanagent’sactionindirectly,byexplainingthosefeaturesoftheagent’spsychology
thatexplaintheiraction.Theseareindirecttheoriesofnormativereasonexplanation.
The focusof thenext twosections isoncritiquingelliptical anddirect theoriesofnormative
reasonexplanation,respectively.Subsequentchaptersaredevotedtothedefenceofmyown
indirecttheoryofnormativereasonexplanation.
3 Ellipticaltheories
Elliptical theories5accept the conclusionof TheExplanatory ExclusionProblem; they suggest
thatwhenIsaythatItookmyumbrellabecauseitwasraining,thepurportedexplanans(the
fact that itwas raining) isnot theactual explanans.Nonetheless,when I say that I tookmy
umbrellabecauseitwasraining,youunderstandwhyItookit–thatis,inspiteoftheapparent
inaccuracyofwhatIsaid,youstillunderstoodwhyItookmyumbrella.Thequestionis:how?
Howdowemanage toexplainouractions if thepurportedexplanans inanormative reason
explanationisnottheactualexplanans?
The elliptical theorist’s response, is to say that, in a normative reason explanation, the
normativereasonisellipticalfortheactualexplanans,so,forinstance,whenIsaythatItook
myumbrellabecauseitwasraining,thefactthatitwasrainingisellipticalforsomefeatureof5The name for these theorieswas inspired byMaria Alvarez’s (2010, 180) related (but not identical)discussionofwhatshecalls‘Humeanexplanations.’
145
mypsychology,which iswhat reallydoes theexplaining. That is,whenwegiveanormative
reason explanation, ‘we suppose that, properly understood, it should be seen as
enthymematic, i.e. asanacceptable shorthandversionof the full explanation.’ (Dancy2000,
121)
FigureX-1:Ellipticaltheoriesofnormativereasonexplanation
Althoughtheyarerarely6explicitlyadvocated,ellipticaltheoriesprovidewhatisprobablythe
defactoaccountofhowfactsabouttheworldexplainouractions.
3.1 Ellipticalforwhat?
What feature of our psychology is it that normative reasons aremeant to be elliptical for,
whenwegiveanormativereasonexplanation?Opinionsdiverge.OneviewisthatifIsaythatI
took my umbrella because it was raining the conversational implicature7is the fact that I
believedthatitwasraining,8anotherviewisthatitisthefactthatIactedforthereasonthatit
was raining,9andanother still is that it is the fact that I knew that itwas raining.10Of these
threeviewsonlythelatterisrobusttoaparticularsortofchallengeposedbyGettiercases,11
soIwillassumethattheconversationalimplicatureofsayingthatItookmyumbrellabecause
itwasrainingisthatIknewthatitisraining.
6Sandis(2013)and, Ithink,Dancy(2014)arerareexceptions.Sandis(2012,178fn.24)alsoattributesthisviewtoMichaelSmith.7Which,recall,iswhatreallyexplainswhyItookmyumbrella.8Thisisprobablythedefactoview,and,Ithink,isexplicitlythepositionofSandis(2013).9SeeDancy(2014).10Gibbons (2010,359) comes close toadvocating this view–although I thinkhis eventualposition iscloser to the indirect theory that I advocate in §(XII), since, on the same page, hemakes clear thatnormativereasonscanexplain.11Thechallenge is this: recall, from§(IV)1.4 thatwhenEdmundhadaGettierbelief (i.e. justifiedandtruebutnotknowledgeable)thattheicewasthinwecouldnotsaythathestayedattheedgebecausetheicewasthin.However,iftheconversationalimplicatureofsaying‘Hestayedattheedgebecausetheicewasthin’ismerelythathebelievedthattheicewasthin,thenthereisnoreasonwhyweshouldnotsayit–thisisthenaproblemforellipticaltheoriesthattakethepurportedexplananstobeellipticalforthefactthattheagentbelievedit.Incontrast,iftheconversationalimplicatureisthatheknewtheicewasthin,thenweshouldnotsaythathestayedattheedgebecausetheicewasthin,sincehedidnotknowit.Dancy’s(2014)account,accordingtowhichtheconversationalimplicatureisthathestayedattheedgeforthereasonthattheicewasthincouldaccountforthefactthatwedon’tsaythathestayedattheedgebecausetheicewasthinifhewerewillingtosaythatanagentonlyactsforthereasonthatpiftheyknowit;butheisn’t,soitcan’t.
Psychological
factimplies explainsNormativereason
Action
146
3.2 Theproblemswithellipticaltheories
Elliptical theories claim that normative reason explanations provide some explanation of an
agent’sactiononlybecausetheyimplythattheagentknewthatnormativereason.Critically,
ellipticaltheoristsinsist,thenormativereasondoesnoexplanatoryworkofitsown.
I raise two related concerns with this view: first, it makes normative reasons explanatorily
inert,contrarytotheprevailingviewthattheyoughttohaveexplanatorypower;andsecond,
it renders ordinary language explanations of our actions thoroughly unsuited to the task to
whichwehabituallyputthem.
Onthefirst:the‘explanatoryconstraint’,sonamedbyJonathanDancy(2000),12andtowhich
many13subscribe, says that any theory of reasons must account for someone doing some
actionbecauseofanormativereasonthatthereisforthemtodoit.Thisconstraintseemslike
a modest one: assuming that normative reasons indeed have some normative import, the
explanatory constraint requires only that normative reasons havemore than just normative
significance.AsUlrikeHeuersuccinctlyputsit,thisrequirement,‘expressesnothingmorethan
theeverydayassumption thatwe sometimes…do somethingbecause it is rightor justified.’
(Heuer 2004, 47) Indeed, onemustwonderwhat the point of normativity is if we can’t do
thingsbecausetherearesuchnormativereasonsforustodothem(whyrecommendanaction
ifthatrecommendationcan’taffectwhetherornotyoudoit?).
Theproblemforellipticaltheoriesisthattheyclearlyfailtheexplanatoryconstraint.Theyhold
thatit isneverthenormativereasonpersethatexplainstheagent’saction,butonlywhatis
impliedbyit–therebyrenderingnormativereasonsexplanatorilyinert.14
The second problem: denying that normative reasons explain our actions contradicts our
habitualpatternsofspeech.Weroutinelycitenormativereasonsbywayofexplanationofour
actions and it does quite severe disservice to our ordinary language expressions to suppose
thatwhenIsay,‘Laurathrewawaythemilkbecauseithadgoneoff’therealexplanansisnot
what I say it is15but only what is implied by it. Ordinary language may be occasionally
imprecise or misleading, but to accept such ubiquitous misrepresentation as a part of our
everydayaccountsofactionsseemstobeahighpricetopay.Ithinkwecandobetter.12Dancy’sworkis,tomyknowledge,alsothefirstappearanceofthisargument.13(e.g.Dancy2000,101;Smith2004,175;Hornsby2007,301;Raz2009,194;Hieronymi2011,415)14ThisargumentissimilartoTheRightReasonsProblem(see§(III)3.2).However,whilethatargumentrequired that an agent should be able to do something for reasons that make it worth doing, thisargumentrequiresthatanagentshouldbeabletodosomethingbecauseofwhatmakesitworthdoing,indeed, they should be able to do it because it is worth doing. The problem for The ExplanatoryExclusionProblemisthatitisincompatiblewiththeideathatsuchfactscouldexplainanagent’saction.15Thatis,whatfollowsthe‘because’.
147
4 Directtheories
The only way to satisfy the explanatory constraint and to accept the literal form of our
everydayexpressionsistoconcedethat(non-psychological)normativereasonscanexplainour
actions. Thismeans rejecting the conclusionof The Explanatory ExclusionProblem. The first
waytodothat,whichweconsidernow,istorejectthefirstpremise.
RecallthegeneralformofthefirstpremiseofTheExplanatoryExclusionProblem:
Premise# Thereisafullexplanationofwhyzsuchthatxisneitherapartofthatfullexplanationnorisitpartofagenuinelydifferentexplanation.
Now,asweestablished in§(VIII), it seemsas thoughwheneverwegiveanormative reason
explanation we can make a claim that fits the Premise # form about the explanatory
contributionofthenormativereasontotheexplanationoftheaction16;thatis,foranyaction
andanynormative reason todo thataction, there isalwaysa fullexplanationof thataction
such that that normative reason is neither a part of that full explanation nor is it part of a
genuinelydifferentfullexplanation.Directtheoriesdenythisclaim.
4.1 Hownormativereasonsdirectlyexplainactions
Directtheoriessaythatwhenanormativereasonexplainsanactionitaddssomethingtothe
explanationof that action that is independentof, and in addition to,what the fact that the
agentknewitprovides.
FigureX-2:Directtheoriesofnormativereasonexplanation
Buthoware factsabout theworld supposed todirectlyexplainouraction?Here isa typical
response:followingGilbertRyle(1949),onecanconceiveofknowledgeasacapacityorability.
Inparticular, one can conceiveof knowledgeas the capacity to respond to a fact about the
world,thatis,thecapacitytorespondtoanormativereason.
Thesuggestionisthat,whenanagentactsfromknowledge:
- ‘Thefactthatthingsactuallyarethewaytheybelievethemtobeweighswiththem’(Hornsby2008,254);or
- Thefactexertsarationalinfluenceontheagent’swill(McDowell2013);or
16There are, perhaps, exceptions in ‘weird’ cases, where the fact that p and the fact that the agentbelievesthatparebothnormativereasonsforthemtoφ(seeDancy2000,124)–butthesearepeculiarenoughthattheycanbeignored.
Normativereason explains
Action
148
- Theagentisguidedbythefact(Hyman2015);or
- Theagentexhibitsarationalresponsetothefact.(Smith2004)
Thus, knowing some fact engenders a special, direct connectionbetween theagent and the
fact. When one knows something the fact itself guides one or impresses itself upon one’s
actionandtherebyaccountsforwhatonedoes:thatishownormativereasonsdirectlyexplain
anagent’saction.
Notalldirecttheoriesrelyonknowledgetoaccountforthedirectconnection,however.Some,
suchasDancy’s(2000)non-factivetheoryofnormativereasonexplanationdonotthinkthata
normative reasonneedevenbe true inorder toexplainanagent’s action, so longas itwas
believed.Moregenerally,evenamongstknowledge-baseddirecttheories, theprecisenature
of how the normative reason explains depends on the way in which one rejects the first
premise of The Explanatory Exclusion Problem. I consider the main ways of being a direct
theoristintheAppendixtothischapter.
4.2 Theproblemswithdirecttheories
4.2.1 What’sweirdaboutdirectnormativereasonexplanation
Howexactlydoes‘responding’toanormativereason,being‘guided’byoneor‘actinginlight
of it’ make that normative reason directly explanatory? The nature of the explanation that
knowledge is supposed toengender is thoroughlymysterious,andaccountsof itare replete
withmetaphorsbutthinondetail.Iftheconceptof‘respondingtothefact’isnotcausal(and
none seem to think it is), what exactly is the nature of the direct connection between the
agentandthefact,whentheyrespondtothefact,thatmakesthatfactexplaintheiraction?17
Now, if itwere self-evident that there is suchadirect connection then theuseofmetaphor
might well be unproblematic. For instance, we can say that maglev trains are ‘guided’ by
magnetswithoutneedingtobeliteralabouttherelationbetweenthemagnetsandthetrain
because it is seemingly clear that something themagnets are doing is directly affecting the
train. However, it is very much not clear that normative reasons are directly affecting my
action–soweneedanaccountofwhatisoccurringthatisnotcouchedinmetaphorsinorder
toconvinceusthatthenormativereasonreallyisdirectlyrelatedtotheaction.Iamconcerned
that no such account is available because there is no such direct relation. This is the first
problemfordirecttheories.
17Similar remarks can bemade for direct theories that aren’t knowledge-based, such as non-factivetheories.SeeremarksintheAppendixtothischapterforfurtherdetail.
149
4.2.2 YoucannotcrediblyrejectPremise1
Nowforthesecondproblemfordirecttheories.In§§(VIII)2-4,Isetouttheargumentforthe
followingclaim:
Premise1 ThereisafullexplanationofwhyIcongratulatedmyfriendsuchthatthe fact that she had won an award is neither a part of that fullexplanationnorisitpartofagenuinelydifferentexplanation.
Directtheoriesrejectthisclaim,inthisinstance,andtheyrejectthatasimilarsortofclaimcan
bemadeforanynormativereasonthatpurportstoexplainanagent’saction.Theproblem,I
suggest,isthatthereisnocrediblewaytorejectPremise1.
First, recall that Premise1 followed from four principles of explanation together with the
followingtwoclaims:
Premise1a If,ceterisparibus,myfriendhadnotwonanaward(butIstillbelievedthat she had), then (i) the fact that I believed that she hadwon anawardwouldhavepartiallyexplainedwhyIcongratulatedher;and(ii)the fact that she had not won an award would not have partiallyexplainedwhyIcongratulatedher.
Premise1b If,ceterisparibus,Ihadnotbelievedthatmyfriendhadwonanaward(thoughshehad)thenthefactthatshehadwonanawardwouldnothavepartiallyexplainedwhyIcongratulatedher.
Assumingthattheseclaimsareuncontentious,18theonlyoptionsforonewhowantstoreject
Premise1aretorejectoneofFACTIVITY,ENDURANCE,SUFFICIENCY,orDIFFERENCE.
Ithinkthatthereisnounproblematicwayforadirecttheoristtorejectoneoftheseprinciples
of explanation. I provide a full account of my reasoning in the Appendix to this chapter,
however,bywayofoverviewhere:denyingFACTIVITYcomesatthecostofdenyingsomething
thatisseeminglyobviouslytrue(i.e.thatexplanationisfactive).DenyingDIFFERENCEmakesthe
concept of being a genuinely different explanation obscure, and relying on the denial of
difference to account for normative reason explanation results in an implausibly ubiquitous
levelofoverexplanation.Meanwhile,theoriesthatdenyENDURANCEseeminglystretchcredulity
by insisting that, in some cases, the mind plays no explanatory role in action. And lastly,
theories that deny SUFFICIENCY must insist that normative reasons are indispensable to the
explanationofanactioniftheyareknown–andthereisnogoodaccountofwhythatshould
bethecase.
18AndIknowofnoonewhowoulddenyeither.
150
The upshot, I suggest, is that there is no credible way to reject the first premise of The
Explanatory Exclusion Problem for normative reason explanation, and therefore no credible
wayofbeingadirecttheorist.
5 Conclusion
IhavediscussedtwopossibleresponsestoTheExplanatoryExclusionProblemfornormative
reason explanation, elliptical theories and direct theories. I suggested that both of these
theoriesaredeeplyproblematic.
Ifnormativereasonsdon’texplainelliptically,anddon’texplaindirectly,then, Isuggest,they
must explain indirectly, that is, by explaining those features of the agent’s psychology that
explain their actions. In the next chapter I will provide the basis for my indirect theory by
showingthattheexclusionprincipleisfalse.Subsequentchapterswillthensetoutanddefend
myindirecttheory.
Appendix
A.1 Fourdirecttheoriesofnormativereasonexplanation
Direct theoriesofnormative reasonexplanation reject the idea that the firstpremiseofThe
Explanatory ExclusionProblem is always trueof normative reasons.Asnoted in §4.2.2, this
meansthatdirecttheoriesmustrejectoneofthefollowingprinciplesofexplanation:
FACTIVITY Foranypropositionspandq,ifppartiallyexplainsthefactthatqthenpisthecase.
DIFFERENCE For any propositions p, q and r, p is part of a genuinely differentexplanationofthefactthatrfromqonlyif,ceterisparibus,hadpbeenthecaseandqnotbeenthecase,pwouldstillpartiallyexplainthefactthatr.
ENDURANCE For any propositionsp, qand r, the following holds: Suppose thatqpartiallyexplainsthefactthatrwhenitisnotthecasethatp.Supposefurtherthatneitherpnornotpispartofthesameexplanationofrasq. Then, if, ceteris paribus, it were the case that p, q would stillpartiallyexplainthefactthatr.
SUFFICIENCY Foranypropositionq,andanyset,Δ, ifΔ isa fullexplanationof thefact thatq in somecircumstance, then, inanycircumstance inwhichall theelementsofΔpartiallyexplain the fact thatq,Δfullyexplainsthefactthatq.
151
Since any direct theorymust reject one of these principles,we can categorise the different
directtheoriesaccordingtowhichoftheseprinciplestheyreject,asfollows19:
- Non-factivist: Theories that reject FACTIVITY and insist that when normative reasonsexplainanagent’sactiontheydosoquathecontentoftheagent’sbelief.
- Inclusive disjunctivist: Theories that reject DIFFERENCE and insist thatwhen normativereasons explain what an agent does they do so as part of an explanation that isgenuinelydifferentfromtheexplanationintermsoftheagent’spsychology.
- Exclusivedisjunctivist:Theories that rejectENDURANCEand insist thatwhennormativereasonsexplainwhatanagentdoestheydosoaspartofthefullexplanationinsteadofthefactsaboutwhatanagentbelieves.
- Supplementarist: Theories that reject SUFFICIENCY and insist that when a normativereasonexplainsanagent’s action theydo soaspartof the full explanation togetherwiththefactsaboutwhatanagentbelieves.
Inwhat follows Iwill setout theaccountofhownormativereasonsexplain ineachof these
strategies,followedbytheproblemstheyface.
A caveat regarding all these theories: they are all to a greater or lesser extent, ofmy own
construction. Most of the literature from which these theories are drawn is actually a
discussion of whether or not (and how) normative reasons can be the reason for which an
agentacts. I,however,amansweringasimplerquestion:cannormative reasonsexplainour
actions? The theories below have been inspired by responses to the former question, and
whileIsuggestthatcertainauthorsholdsomeformsofthetheoriesbelow,nothinginwhatI
say depends on these attributions being accurate. That is, so long as they are correct
characterisationsofpossibletheories,itisnotvitalthatthosetowhomIattributethemwould
agreewiththeattribution.
A.2 Non-factivisttheories
Non-factivist theories reject the factivity principle. According to non-factivist theories (e.g.
Dancy 2000; Comesaña andMcGrath 2014), saying that I tookmy umbrella because it was
raininghas theconversational implicaturethat itwasraining,butdoesnotentail that itwas
raining;indeed,theystress,thatimplicatureiscancellable.
19There isa fourthstrategythat Ihaven’tconsideredhere.Onecoulddifferentiatetheexplanantiaofnormative reasons and psychological facts (I believe that Stout (1996) adopts this strategy, but onecouldalsodevelopsuchastrategybasedonHornsby(2008)andMcDowell’s(2013)disjunctivismaboutacting)–andconsequentlyarguethatnormativereasonscanexplainsomethinganagentdoese.g.thefactthatthereisbeerinthefridgeexplainsthefactthatIgotbeer(Icouldn’thavedonesohadtherenot been beer there). I don’t consider this strategy here because I think it suffers all the failings ofellipticaltheorywithoutitssimplicity–wewanttobeabletogiveanormativereasonexplanationofthefactthatIwenttothefridge,notmerelythefactthatIgotbeer.
152
These theories suggest thatwhenanormative reasonexplainsanagent’saction factsabout
whattheagentbelievesarereallysecondfiddleinexplanatorytermstothenormativereason:
theyactasmereenablingconditions forthenormativereason(quawhattheagentbelieves)
toexplainwhattheagentdoes.
The idea seems to be this: a normative reason cannot explain an agent’s action unless the
agentbelievesit20–sothefactthattheagentbelievesitmustatleastenabletheexplanation.
However, ‘there is a difference between a consideration that is a proper part of an
explanation,andaconsiderationthatisrequiredfortheexplanationtogothrough,butwhich
isnotitselfapartofthatexplanation.’(Dancy2000,127)
Accordingtothisstrategy,thefactsaboutwhattheagentbelievesdon’thavetheexplanatory
forcenecessarytoexplaintheagent’saction,onlythenormativereasondoes.Thatis,thefact
thatIbelievethatit israiningenablesthecontentsofmybelief,namely,thatit israining,to
explainwhyItookmyumbrella(andsotheyareindependentpartsofthesameexplanation–
althoughit’sthenormativereasonthatis, insomesense,themajorparty).Theysaythatwe
should understand expressions like ‘Sally ran because she thought a bear was chasing her’
appositionally,wherethereferencetoSally’sbelievingisaqualificationonthetruthvalueof
theexplanans,butnotapartoftheexplanansitself.21
Non-factivisttheoriescanthusacceptthatthesamefullexplanationisavailablewhetherthe
agent’sbeliefistrueorfalsebecausethetruthorfalsityofthecontentoftheagent’sbeliefhas
nobearingonwhetherornotitisinthatfullexplanation.
A.2.1 Theproblemfornon-factivisttheories
The first problem with non-factivist theories is that they are absurd. If we admit that
falsehoods can explain that seems to make the concept of explanation itself thoroughly
mysterious.Ifsomeonesaysthattheytooktheirumbrellabecauseitwasrainingeventhough
itwasn’t,IwouldnotthinktheywerecancellingaGriceanimplicature,Iwouldthinkthatthey
misspoke,becausewhattheysaidisjustplainlycontradictory.
Secondly,non-factivisttheoriesresttheirclaimtoplausibilityontheideathatbeliefsaremere
enablingconditions–theydon’thavetherequisiteexplanatoryforce inandofthemselves. I
amscepticalthattheperceivedexplanatoryweaknessof‘enablingconditions’isgenuineand
isn’tmerelyaconsequenceoftheirsalience inagivencontext.Were itmerelyaquestionof
saliencethentheexplanatorypoweroffactsabouttheagent’spsychologywouldberestored
20Recallthediscussionof§(VIII)3.1.21Dancyproposesthissortofreading:‘Sallyranbecauseabearwaschasingher,asshebelieved.’
153
andwewouldagainhavetoquestionwhatnormativereasonsaddbywayofexplanation.But
evenputtingsuchscepticismaside,itis,asTurri(2009)notes,farfromclearthatfactsabout
what an agent believes are aptly categorised as enabling conditions.22And if they aren’t
enablingconditions,concernsaboutwhatnormativereasonsreallydobywayofexplanation
return.
Thirdly,non-factivist theories struggle toexplainwhy it is that,whenanagenthasaGettier
belief,wedon’tgiveanormativereasonexplanationof theiraction.23For instance,wedon’t
saythatEdmundstayedattheedgebecausetheicewasthinwhenhis justifiedbeliefthat it
wasthinisonlyaccidentallytrue–althoughthenon-factivistwouldinsistthatweshould.24
Fourthly, it isworth noting, from a rhetorical perspective, that even the progenitor of non-
factivisttheories,JonathanDancy,hassinceabandonedthem(seeDancy2014).
A.3 Inclusivedisjunctivisttheories
Inclusivedisjunctivist theories reject thedifferenceprinciple. Theyaccept that there is a full
explanationofwhy Icongratulatedmyfriendthatdoesnot includethefactthatshewonan
award. 25 Nonetheless, they say, there is a genuinely different explanation of why I
congratulatedmyfriendthatdoes includethefactthatshewonanaward.Theysuggestthat
even though the explanation in terms of normative reasons would not exist without the
explanationintermsofbeliefs,itisnonethelessgenuinelydifferentfromit.26
Whatwouldmakeyoubelieve this?Well,asalreadynoted,27therearestructural similarities
betweentheargument from illusion in the literatureonperceptionand theproblemfor the
22Turri(2009,505–6)arguesthatitisnormallyoddtoaskwhyanenablingconditionforanexplanationobtains,but that it isnotnormallyodd toaskwhyanagentbelievedwhat theybelieved– thereforefactsaboutwhatanagentbelievesaren’tenablingconditions.23Seefn.11.24See,forinstance,Dancy(2014,89)forthiscriticismofnon-factivisttheories.25This is what makes them inclusive disjunctivists as opposed to the exclusive disjunctivists, to beconsideredinthenextsection.Ihaveadoptedtheinclusivevs.exclusivedistinctionfromRuben(2008)andStout(2009);theysaythatinclusivedisjunctivistsacceptsomeroleforthehighestcommonfactor,while exclusive disjunctivists do not. Note Pautz (2010, 298–99) draws the same distinction betweendifferenttypesofdisjunctivism,buthecallsinclusivedisjunctivismthe‘overdeterminationversion’;andexclusivedisjunctivismand‘therestrictiveversion’.26This is, I think, the view that JohnHyman sets out: ‘If Jamesmerely believed that going to churchwouldpleasehismotherbutdidnotknowthatitwould,wecansaythathewenttochurchbecausehebelievedthat itwouldpleasehismother,butwecannotsaythathewenttochurchbecause itwouldpleasehismother.But ifheknewthat itwouldpleasehismother,wecansayeither thathewent tochurchbecauseheknewthat itwouldpleasehismotheror thathewent tochurchbecause itwouldpleasehismother.’(Hyman2011,366–67)27See§(IX)5.
154
explanationofactionthatIhaveconstruedasTheExplanatoryExclusionProblem.28Onecould,
inspired by disjunctivist responses to the former, adopt a disjunctivist view of action
explanationasaresponsetothelatter.Theideaisthis:althoughitisalwaysexplanatory,the
highestcommonfactorofallactionexplanations(explanationintermsofthefactsaboutwhat
the agent believes) is not the limit of the resources available for action explanation29–
sometimeswe canalso explain an agent’s action directly, and differently, with a normative
reason.Sowhile theaction isalready fullyexplainedby factsaboutwhat theagentbelieves
(and relevant supplementary facts (e.g. facts about what they want, judge good etc.)) the
normativereasonaddsanother,additionalexplanationofit.
A.3.1 Theproblemsforinclusivedisjunctivisttheories
The first problem with inclusive disjunctivist theories is that, in rejecting the difference
principlewe are rejecting a seemingly plausible account of (at least part of) what it is that
makesexplanationsgenuinelydifferent.Asaresult,Ifind,webegintoloseourhandleonwhat
itisthatmakestheexplanationgenuinelydifferent.
Secondly,andrelatedly,it’snotatallclearwhatwouldbeinthefullexplanationoftheagent’s
actionthatincludesthenormativereason.GivenMINIMALITY,itcannotincludealltheelements
ofthefullexplanationthatincludesthebelief–soitmustbesomeothersetoffacts.Butwhat
other set of facts is also sufficient forme to congratulatemy friend andwhich explainsmy
doingsodirectly?Icannotthinkofany.
And lastly, and perhapsmost obviously, the inclusive disjunctivist accepts thatmy action is
overexplained,becausetherearetwogenuinelydifferentexplanationsofit.Thisisproblematic
because, asDancyputs it, ‘[it]wouldmean that there are somehow toomanyexplanations
around’(2000,171).30Andthisisnotabenigncaseofoverexplanation–it’snotatallclearthat
thefactthatIcongratulatemyfriendisexplainedintwodifferentways.
28Consider:Forperception,whenit,forinstance,appearsasthoughthereisatomatooneisalwaysineitheroneoftwostates:eitheroneisseeingatomatooritmerelyappearstooneasthoughthereisatomato.Thefactthatitappearsasthoughthereisatomatoisthehighestcommonfactorofthesetwostates.CompareHorsnby’s(2008)construalofaction:whenoneactsonthebeliefthatitisrainingoneisalways inoneof twostates–actingontheknowledgethat it is rainingormerelyactingonthebeliefthatitisraining.Actingonthebeliefthatitisrainingisseeminglythehighestcommonfactorofthesetwostates.29As McDowell puts it: ‘The point of the disjunctive approach is to reject a highest common factorconception,notinthesenseofdenyingthattherearecommonfeaturesbetweenthedisjuncts,butinthesenseofrefusingtorestrictourresourcesforrationalexplanationtothosethatareavailableforthe“worse”disjunct.’(McDowell2013,27)30Davis(2005)alsomakesthisargument.
155
Theproblemisn’tjustconfinedtothiscaseeither.Whattheinclusivedisjunctivistrequiresis
thatwheneverwecangiveanormativereasonexplanationofanaction,theagent’sactionis
overexplained. But if this sort of overexplanation is meant to occur whenever we give a
normative reason explanation, then, given the ubiquity of such explanations, inclusive
disjunctivismseemstorequireareallyimplausibleincidenceofoverexplanation.
Maybe you want to insist that overexplanation should not trouble us as much as
overdetermination, because there aren’t independent determining factors, there are only
differentexplanatoryfactors.However,Ithinkthisresponseismoretroublethanit’sworth–
it’snotcleartomethatitmakesoverexplanationunproblematicand,asI’venoted,westartto
lose our grasp on what makes genuinely different explanations genuinely different if they
aren’t,insomesense,independentexplanations.
A.4 Exclusivedisjunctivisttheories
Exclusivedisjunctivisttheoriesrejecttheenduranceprinciple.Theyrejecttheideathat,whenI
congratulatemy friend, the fact that I believed that she hadwon an award explains why I
congratulatedher.Accordingtothisstrongersortofdisjunctivism(henceexclusiveratherthan
inclusive)eitheranagentactsbecausetheybelievetheworldtobeacertainwayortheyact
because the world is that way, but never both. On this account, when a normative reason
explainsanagent’sactionitdoessoinsteadofthefactsaboutwhatanagentbelieves.31
Why would you believe this? That is, why think that sometimes beliefs don’t explain?
Stoutlandmakesthefollowingargumentforthisview:
Ifsomeonegoestoaroombecauseameetingisbeingheldthere,thatisadequatejustificationforhereffort.Ifitturnsoutthatthemeetingisn'tthere,wehavetoreviseourjustification,andhenceourexplanation,andsayshewent to theroombecauseshebelieved themeetingwasthere.Herbeliefbecomesanexplanatory factor, that is tosay, just incaseshewasmistakenabout the situation originally appealed to as justification. This is the general case: beliefsbecomeexplanatoryfactorswhenagentsaremistakenaboutthesituationsoriginallytakentojustifythem.
Thesituation isanalogoustosomeone’s flippingaswitchto turnona light,without the lightgoingon.Inthiscasehetriedtoturnonthelight,justbecausehefailed,whichwouldnotbe
31 Both Collins (1997) and Stoutland (1998) adopt this response to the Problem. I may havemis-interpretedHyman’s (2011) view in fn.26, inwhich case I think this ishisposition. Sandis (2012,119)alsoattributesthisviewtoAlvarez(2010),thoughI’mnotconvinceditisherposition.Inparticular,Alvarez agrees that beliefs still explain in veridical cases (cf. ‘It is always possible (and sometimesnecessary, namely when the agent acted on a false belief) to give explanations in the psychologicalform.’ (Alvarez 2013, 149)). More generally, it’s not clear to me that Alvarez is responding to TheExplanatory Exclusion Problem as I have characterised it – her discussion focuses on the pragmaticconsiderationsthatdeterminewhetherornotonegivesanormativereasonexplanationoroneintermsof psychological facts. However, to the extent that she does respond to the redundancy objection, Ithinksheisbettercharacterisedasadoptingthesupplementaristtheory.
156
true ifhesucceeded in turning itonwithoutdifficulty.Fromthe fact thatwhatsomeonedidwhenhe failed to turnona light is thathe tried, itdoesn't followthatwhathedidwhenhesucceeded also included trying. Analogously, given that what explains my going to a roomwhereIdonotfindmyfriendisthatIbelievedmyfriendwasthere,itdoesnotfollowthatwhatexplainsmygoing to the roomwhenmy friend is there isalso that Ibelieved shewas there.(Stoutland1998,61)
IthinkthatthemostcoherentreadingofStoutland’saccountisasarejectionoftheideathat
wheneveranagentacts, theyactonthewaytheytakethingstobe.That is:eitheranagent
actsonthewaytheytakethingstobeor,touseMcDowell’s(2013)phraseology,they‘actin
lightofthefacts’,butnotboth.32Iftheyactonthewaytheytakethingstobe,thenthefact
thattheytakethemtobethatwayexplainstheiraction,butiftheyactinlightofthefacts(i.e.
inlightofanormativereason),thenitisthefactsthatexplaintheiraction.Thatis,eitherfacts
aboutwhattheybelieveexplaintheiractionsornormativereasonsexplaintheiractions,but
notboth.
A.4.1 Theproblemsforexclusivedisjunctivisttheories
Firstly,byrejectingtheenduranceprincipletheexclusivedisjunctivistiscommittedtotheview
thatwhensomeirrelevantfalsepropositionbecomestrue,thatcandestroypre-existingpartial
explanationrelations.Forthereasonssetoutin(VIII)2.3.1,thatseemsimplausible.
However,even if it turnsout that theenduranceprinciple is false, theexclusivedisjunctivist
theoryisadifficultpositiontomaintainbecausethereiswidespreadsupportfortheviewthat
beliefsalwaysplaysomeroleinexplainingwhatanagentdoes.
Exclusivedisjunctivism insists that there is nohighest common factor that is relevant to the
explanation of action, whether one is right or mistaken. That is, it insists that there is no
significantcommonfactorbetweensomeonewhoactsonsomethingtheyknowtobethecase
andsomeonewhoactsonsomethingtheymerelybelieve.Ifindthisincredible.
32This reading has Stoutland endorsing (exclusive) disjunctivism about acting. This is different fromdisjunctivismaboutwhatexplainsanagent’saction,orwhattheirreasonforactingis.It’sworthnotingthatHornsby(2008)andMcDowell(2013)giveaconsiderablymorethoroughaccountofdisjunctivismaboutacting,buttheirsisofan inclusivekind–theyconcedethatwheneveranagentactstheyactontheway they take things tobe.Sowhile I takeStoutlandtobeadisjunctivistabout thesamething, Idon’t invoke them here since his disjunctivism is far stauncher. There is also an alternative reading(perhapstruertohisprecisestatements)onwhichStoutlandembracesanexclusivedisjunctivismaboutactionexplanation(i.e.anexclusiveversionofthedisjunctivismconsideredintheprevioussection).Allthat such a disjunctivism insists is that the facts aboutwhat an agent believes donot always explaintheiraction, it is silenton thequestionofwhetherornotanagentalwaysactson theway they takethingstobe.Thislattersortofdisjunctivismisweakerthantheformerinsofarasitisentailedbyit.Ifindthelattersortlesscoherentwithouttheformersinceitisnotcleartomehowonecouldactonthewayonetakes things tobewithout the fact thatone takes themtobe thatwayplayingsomerole inexplainingwhatonedoes;forthatreasonIdon’tconsideritfurther.
157
Consider:when I believe that it’s raining I ameither in a situation inwhich I know that it’s
raining or I merely believe (i.e. without knowing) that it is raining. Given that I take my
umbrellawhicheverofthetwosituationsIamin,doesn’tthepossibilitythatthetwosituations
may be indistinguishable tome justmean that I act on theway I take things to be in both
situations?33AndgiventhatIactonthewayItakethingstobe,thefactthatItakethemtobe
that way must explain my action. I find positions that deny this reasoning impossible to
believe.
Moreover,Stoutland’ssupposedargumentbyanalogyisnotsomuchanargumentforhisview
as it ismerelytheapplicationof it toanotherarea. Itseemscleartome(andtoothers(e.g.
O’Shaughnessy 1973), that someone who goes to turn on a light tries to turn on the light
whetherornottheyendupdoingso.34Stoutland’srejectionofthatviewisjustthesamesort
of (exclusive)disjunctivismashis rejectionof theview thatonealwaysactson thewayone
takes things tobe.35Andonewho isnotpersuadedofhisview in the latter isunlikely tobe
persuadedbytheapplicationofitsomewhereelse.
Ofcourse,myincredulousstaremaydonothingtoaltertheopinionofsomeonewhobelieves
sucha theory;however, Idon’t think I amalone inmy incredulity. Surely thereare theories
thatareeasiertobelieve?
A.5 Supplementaristtheories
Finally,supplementariststrategiesrejectthesufficiencyprinciple.Theyarguethatjustbecause
factsaboutwhatanagentbelievesplaya role inexplaining theiraction thatdoesnotmean
that normative reasons are not needed for a full explanation of the agent’s action; indeed,
they argue, normative reasons are an independent (and necessary) part of the same
explanationasthefactsaboutwhatanagentbelieves.36
Supplementarist strategies differ from non-factive strategies in that they insist that the
normative reason can explain the agent’s action only if it is true; for instance, the
supplementaristmightinsist,thenormativereasononlyexplainsanagent’sactioniftheagent
knows it.What the supplementarist rejects is the idea that the same full explanation of an
33Thisisnotadenialofdisjunctivismtoutcourt–forinstance,aninclusivedisjunctivist(e.g.McDowell2013) can acknowledge that one always acts on theway one takes things to be; they just deny thatactingonthewayonetakesthingstoisallthatapersoneverdoes.34Comparethesecommonplaceexpressions:‘hetriedandfailed’;‘hetriedandsucceeded’–Stoutland’saccountmakestheformertrueofanyonewhotries,andmakesthelatteracontradiction.35Indeed,Dancy(2008b)discussesadisjunctivistaccountoftryingtoact.36As noted in fn. 31 to the extent that Alvarez (2010) gives a response to The Explanatory ExclusionProblem,Ithinkthismaybeherview.
158
agent’s action is availablewhether or not the agent knowswhat theybelieve. Instead, they
say,thenormativereasonisanindispensablepartoftheexplanationwhen(andonlywhen)it
is known. Importantly, the supplementarist doesnot violateMINIMALITY because theydonot
thinkthatthesetthatomitsthenormativereasonisstillsufficienttoexplaintheagent’saction
whentheagentactsfromknowledgeofthenormativereason.37
A.5.1 Theproblemsforsupplementaristtheories
Thefirstproblemforsupplementaristtheoriesisthatinrejectingthesufficiencyprinciplethey
rejectaseeminglyplausibleprincipleofexplanation.Thatis,ifsomesetofpartialexplanations
is sufficient to explain some explanandum in one case, given that whenever those partial
explanationsallexplainanexplanandum theyaresufficient toentail it,whywould theyalso
notbesufficienttoexplainit?
However,even if therewereagoodargumentforrejectingthesufficiencyprinciple, it isnot
clearwhyitshouldbefalseinthiscase.Thatis:it’snotclearwhyknowinganormativereason
makes that normative reason an indispensable part of the full explanation of the agent’s
action. Given that what I believe doesn’t explain my action when I am mistaken, so an
explanation in termsofpsychological facts suffices,why is that explanation insufficient (and
thenormativereasonindispensable)whenmybeliefhappenstoknowledgeable?
I suppose the supplementarist theoristwill answer thisquestionby saying that it’s because,
whentheagentknowsthenormativereason,thenormativereasonis,itself,apartofthestory
ofwhytheyactedbecauseitguidesthemortheyrespondtoit.Whenanagentdoesn’tknow
thenormativereason,itdoesn’tguidethem,soitisn’tpartofthereasonwhytheyacted.Then
I fallback tomy first concern (see§4.2.1): I justdon’t knowwhat itmeans foranormative
reasontoguidesomeoneanduntilIdo,Ican’tseewhyIshouldacceptthatnormativereasons
areindispensabletoanexplanationonlyiftheyareknown.
37AlthoughtheydoviolateC-MINIMALITY(see§(VIII)fn.23).
159
(XI)
TheExclusionPrincipleisFalse
In which I show that the exclusion principle is false. I provide twocounterexamples to the exclusion principle, one involving causal explanationand another involving non-causal explanation. I suggest that they arecounterexamples because in each case the purportedly excluded fact explainsthe explanandum by explaining something that, in turn, explains theexplanandum. Isuggestthattheproblemwiththeexclusionprinciple isthat itdiscriminates against all but the most proximal explanations of any givenexplanandum, and that this is problematic at least partly because we aretypically interested inmoredistal explanations. I explainwhereour reasoningwentwrongandwhichfullexplanationanapparentlyexcludedfactispartof.
InthepreviouschapterIarguedthatweshouldnotaccepttheconclusionofTheExplanatory
Exclusion Problem (at least as far as normative reason explanations are concerned). I also
arguedthatweshouldnotrejectthefirstpremiseoftheProblem.Ifweshouldnotacceptthe
conclusion and we should not reject the first premise, the only remaining response to the
Problemistorejectthesecondpremise,theexclusionprinciple.Thepurposeofthischapteris
to show that the exclusion principle is false and to explainwhy it is false. The next chapter
builds on these insights to provide the makings of an indirect theory of normative reason
explanation.
1 Twocounterexamplestotheexclusionprinciple
Recalltheexclusionprinciple:
EXCLUSION For any propositions,pandq, if there is a full explanation ofwhyqsuchthatp isneitherapartofthatfullexplanationnor is itpartofagenuinelydifferentexplanation,thenpdoesnotpartiallyexplainq.
In this section I want to discuss two counterexamples to EXCLUSION, one involving causal
explanationandtheothernon-causalexplanation.
1.1 Acausalcounterexample
JeancontractsHIVafterhavingbeentransfusedwith infectedblood. Itgoesundiagnosedfor
solongthat,tragically,hedevelopsAIDS.Thismuchseemsclear:JeandevelopedAIDS,inpart,
becausehewastransfusedwithHIV-infectedblood.Thatis:
(a) ThefactthatJeanwastransfusedwithHIV-infectedbloodpartiallyexplainswhyhedevelopedAIDS.
160
However,wecanalsogive the followingexplanationofwhyhedevelopedAIDS:hehadHIV
and it went untreated. The fact that he had HIV and the fact that it went untreated are, I
suggest, parts of a single full explanation of why he developed AIDS. Is the fact that he
contractedHIV froman infectedblood transfusionapartof thatexplanation? I suggestnot:
thefactthathehadHIVisalreadyenough1forJeantodevelopAIDS,2sothefactthathewas
transfusedwithHIV-infectedbloodwouldbesuperfluous inanyfullexplanationthatalready
includedthefactthathehadHIV.3Thus,thereisafullexplanationofwhyhedevelopedAIDS
thatincludesthefactthathehadHIVbutdoesnotincludethefactthathewastransfusedwith
HIV-infectedblood.4
Moreover, the fact that Jean was transfused with HIV-infected blood only explains why he
developedAIDSgiven thathehadHIV.5So, fromDIFFERENCE, the fact thathewas transfused
with HIV-infected blood cannot be part of a genuinely different explanation of why he
developedAIDSfromthefactthathehadHIV.
So,itfollowsthat:
(b) ThereisafullexplanationofwhyJeandevelopedAIDSsuchthatthefactthathewas transfusedwithHIV-infectedblood isneitherapartof that fullexplanationnorisitpartofagenuinelydifferentexplanation.
Now notice that (b) is the antecedent condition of EXCLUSION while (a) is the denial of the
consequent condition. That is, the fact that Jean was transfused with HIV-infected blood
explainswhyhedevelopedAIDSinspiteofthefactthatthereisafullexplanationofwhyhe
developedAIDSthatitisneitherapartofandnorispartofagenuinelydifferentexplanation.
So,giventhat(a)and(b)aretrue,EXCLUSIONmustbefalse.
1Togetherwith the fact that itwentuntreatedand the fact thata sustained,untreatedHIV infectionleadstoAIDS,etc.2Wecanseethisbynotingthatafullexplanationofthissort(hehadHIV,itwentuntreated,untreatedinfectionsleadtoAIDS…)wouldhavebeenavailable,ceterisparibus,howeverhecontractedHIV.3AndMINIMALITYprecludesthepossibilityoffullexplanationshavingsuperfluouselements.4Wecanarriveatthisconclusionusingtheestablishedreasoningfrompreviouschapters,asfollows:if,ceterisparibus,JeanhadnotbeentransfusedwithHIV-infectedbloodbutstillhadHIV(nevermindhowhe contracted it), thenhewould still havedevelopedAIDS, and the fact thathehadHIVwouldhave(partially) explainedwhy he developed AIDS.Moreover, in this counterfactual circumstance, the factthathehadnotbeentransfusedwithHIV-infectedbloodwouldnothavebeenpartoftheexplanationofwhy he developed AIDS. This supplies us with the appropriate specification of Premise #a, which,together with FACTIVITY, ENDURANCE and SUFFICIENCY, leads us to the conclusion that there is a fullexplanationofwhyhedevelopedAIDSthatincludesthefactthathehadHIVbutdoesnotincludethefactthathewastransfusedwithHIVinfectedblood.5If,ceterisparibus,JeanhadnotcontractedHIVdespitehavingbeentransfusedwithHIV-infectedbloodthen he wouldn’t have developed AIDS, so the fact that he was transfused with HIV-infected bloodwouldnothaveexplainedwhyhedevelopedAIDS.ThissuppliesuswiththeappropriatespecificationofPremise#b.
161
1.2 Anon-causalcounterexample
PerhapsyouagreethatEXCLUSIONisfalseforcausalexplanation,butyouinsistthatitistrueof
non-causalexplanation.Ifthatweretrue,andonemaintainedthatreasonexplanationisnon-
causal, then The Explanatory Exclusion Problem for normative reason explanation would
return.Thepurposeof this section is todemonstrate that EXCLUSION isequally false fornon-
causalexplanation.
Considerthefollowingchessopening:
FigureXI-1:TheHyperAcceleratedDragon
Thisgamehasjustbegun.White’skingandkingsiderookareintheirstartingpositions(e1and
h1,respectively).Itistypicalintheearlystagesofachessgameforaplayertoseektocastle.
Foraplayerplayingwhite,castlingonherking’ssidealwaysinvolvesmovingthekingfrome1
tog1,and(aspartofthesameturn)therookfromh1tof1.However,Whitecannotcastleon
herking’s sidebecauseof thebishop (on f1)between thekingand thekingside rook,which
obstructsthemove.Thatis:
(c) The fact that there isabishopbetween thekingand thekingside rookpartiallyexplainswhyWhitecannotcastleonherking’sside.
Itakeitthattheexplanatoryrelationshiphereisclearlynon-casual(oratleastnotcausalinthe
familiarsenseinwhichJean’sstoryisacausalstory).
Now,theWorldChessFederation’shandbookstatesthat:
Castling is prevented temporarily… if there is any piece between the king and the rookwithwhichcastlingistobeeffected.(FIDE,n.d.,3.8.2.2)
Thisfact(concerningtherulesofchess),togetherwiththefactthatthereisapiecebetween
White’skingandherkingsiderookfullyexplainswhyWhitecannotcastleonherking’sside.
162
Butthatfullexplanationdoesnotincludethefactthatthereisabishopbetweenthekingand
thekingsiderook.6So,itseems,thereisafullexplanationofwhyWhitecannotcastleonher
king’ssidethatincludesthefactthatthereisapiecebetweenthekingandtherookbutdoes
notincludethefactthatthereisabishopbetweenherkingandking’ssiderook.7
Moreover, the fact that there is a bishopbetween the king and the rook only explainswhy
Whitecannotcastleifthereisapiecebetweenthekingandtherook.8Thatbeingsoweknow
(fromDIFFERENCE)thatthefactthatthereisabishopbetweenthekingandtherookisnotpart
ofagenuinelydifferentexplanationofwhyWhitecannotcastle fromthefactthatthere isa
piecebetweenthekingandtherook.
So,itfollowsthat:
(d) There is a full explanation of whyWhite cannot castle such that the fact thatthere is a piece between the king and the rook is neither a part of that fullexplanationnorisitpartofagenuinelydifferentexplanation.
Nownotice,again, that (d) is theantecedentconditionofEXCLUSIONwhile (c) is thedenialof
theconsequentcondition.So,giventhat (c)and (d)are true,andgiventhat theexplanatory
relationsinvolvedarenon-causal,EXCLUSIONmustbefalseofevennon-causalexplanation.
2 Whythesearecounterexamplestotheexclusionprinciple
Why are these cases counterexamples to the exclusion principle? Answering that question
meansanswering thequestionofhow the seeminglyexcluded factsnonethelessexplain the
explanandatheyareexcludedfrom.Myanswerisstraightforward:theyexplainbyexplaining
thosefactsthatinturnexplaintheexplananda.
6Again,thelatterisjustsuperfluoustothatfullexplanation,andMINIMALITYprecludessuchsuperfluity.7Again,wecouldarguetothisconclusionusingtheconventionalreasoning:If,ceterisparibus,therehadbeennobishopbetweentherookandthekingbuttherehadstillbeenapiecebetweenthekingandtherook(supposethattherewasaknightthereinstead),thenthefactthattherewasapiecebetweenthekingandtherookwouldstillhaveexplainedwhyWhitecouldnotcastle.Moreover,hadtherebeennobishopbetweenthekingandtherook,thefactthattherewasnobishopbetweenthekingandtherookwouldnothaveexplainedwhyWhitecouldnotcastle.ThisprovidesuswiththerelevantspecificationofPremise#a,so,fromFACTIVITY,ENDURANCEandSUFFICIENCYwecanconcludethatthereisafullexplanationofwhyWhitecannotcastlethat includesthefactthatthere isapiecebetweenthekingandtherookbutdoesnotincludethefactthatthereisabishopbetweenthekingandtherook.8Ifthebishophadn’tbeenapiece(perhapsoneisplayingachessvariantthatexcludesbishops,butforwhichtherulesareotherwisethesame), thenWhitewouldhavebeenabletocastle,sothefact thatthere isabishopbetween thekingand the rookwouldnothaveexplainedwhyWhitewouldn’thavebeenable to castle (sinceWhitewouldhavebeenable to castle). This is the relevant specificationofPremise#b.Consider:iftherehadbeenapennyontheboardbetweentherookandthekingwewouldnotsaythatWhitecannotcastle;however,inthechessvariantinwhichbishopsaren’tpieces,theyare,fromthegame’sperspective,nodifferentfrompennies–thatis,theyarejustirrelevant.
163
2.1 Revisitingthecausalcounterexample
InthefirstcounterexampleIinsistedthatthefactthatJeanwastransfusedwithHIV-infected
bloodexplainswhyhedevelopedAIDS.Howdoesitdoso?Itdoessobyexplainingwhyhehas
HIV.
Of course, the fact that JeanhasHIV is themore immediateexplanationof the fact thathe
developedAIDS.And,ofcourse,thewayonecontractsHIVmakesnodifferencetowhetheror
notonedevelopsAIDSgiventhatonehasHIV. Ifyourquestionwerewhy,giventhathehad
HIV, did Jean develop AIDS, then the relevant explanation includes the fact that it went
untreated,andthefactsabouthowasustained,untreatedHIV infection leadstoAIDS–and
that explanation needmake nomention of howhe contractedHIV. But itwould be odd to
alwaystakethefactthathehadHIVasgiven!Andonceyoudon’ttakeitasagiven,apartof
theexplanationofhowhecametodevelopAIDSiswhyhehasHIVinthefirstplace.Andhe
hasHIVbecausehewastransfusedwithHIV-infectedblood.
Summarising:thefactthatJeanwastransfusedwithHIV-infectedbloodexplainsboththefact
thathehasHIVandthefactthathedevelopedAIDS,andthefactthathehasHIValsoexplains
thefactthathedevelopedAIDS.Symbolically:ppartiallyexplainsq,qpartiallyexplainsrandp
partially explains r. The question is:why doesp partially explain r? I think the simplest and
mostnaturalansweristhattheexplanatoryrelationsinvolvedaretransitive.
Giventhetransitivityofthe(partial)explanationrelationsinvolved,thefactthatpexplainsq
togetherwith the fact thatq explains r ensures thatp explains r. So, the fact that Jeanwas
transfusedwithHIV-infectedbloodexplainswhyhedevelopedAIDSbecause itexplains that
which explains why he developed AIDS and because of the transitivity of the explanatory
relationsinvolved.
2.2 Revisitingthenon-causalcounterexample
Whatof the chessexample?What is the relationshipbetween the fact that there is apiece
betweenthekingandthekingsiderookandthefactthatthereisabishopbetweentheking
andthekingsiderook?
Well,thefactthatthereisapiecebetweenthekingandthekingsiderookisanexistentialfact.
Andexistentialfactsareexplainedbytheirinstances:
‘Why is it that something is F? Because A is F. An existential quantification is explained byprovidinganinstance.’(Lewis1987,223)
‘Existentialquantificationsaretruebecauseoftheirtrueinstances.’(Schnieder2011,460)
164
So,why is thereapiecebetween thekingand thekingside rook?Because there is abishop
between thekingand thekingside rook (andabishop isapiece…)– the fact that there isa
bishopbetweenthekingandthekingsiderookexplainswhythereisapiecebetweentheking
andthekingsiderook.
Summarising:thefactthatthereisabishopbetweenthekingandthekingsiderookexplains
bothwhythereisapiecebetweenthekingandthekingsiderookandwhyWhitecannotcastle
onherking’sside,andthe fact that there isapiecebetweenthekingandthekingsiderook
also explainswhyWhite cannot castle. Symbolically, again:p partially explainsq,q partially
explainsrandppartiallyexplainsr.And,again, Ithinkthemostnaturalexplanationofthese
facts is that theexplanatoryrelations involvedare transitive. That is:pexplainsrbecausep
explainsqandqexplainsr.
3 What’swrongwiththeexclusionprinciple
Having understood how the apparently excluded facts of these two counterexamples
nonetheless explain, we are now in position to see what was wrong with the exclusion
principle.
First, somebasic terminology: ifpexplainsq andq explains rand the explanatory relations
involvedare transitive thenwe can say thatp is adistal explanationof r,whileq is amore
proximal explanation of r. So, the suggestion of the previous sectionwas this: the fact that
Jean was transfused with HIV-infected blood and the fact that there is a bishop between
White’s king and her kingside rook are distal (partial) explanations of their respective
explananda.Whereas,bycomparison,thefactthatJeanhadHIVandthefactthatthere isa
piecebetweenthekingandthekingsiderookaremoreproximal(partial)explanationsoftheir
respectiveexplananda.Isuggestthattheproblemwiththeexclusionprinciple,andthereason
whyitiswrong,isthatitdeniestheexplanatorypowerofdistalexplanations.
Itwillalwaysbetrueofanydistal (partial)explanationofsomeexplanandum that there isa
full explanation of that explanandumofwhich it is not part.Moreover, a distal explanation
doesnotexplaintheexplananduminawaythatisgenuinelydifferentfromthemoreproximal
explanation. So itwill seemingly alwaysbe trueof anydistal explanation that there is a full
explanationof someexplanandum such that thatdistal explanation is not apartof that full
explanationnorisitpartofagenuinelydifferentfullexplanation.Thatis,adistalexplanation
ofsomeexplanandumwillalwayssatisfytheantecedentconditionofEXCLUSION.
165
Thatbeing so, theonlyway topreserveEXCLUSION is todeny theexplanatory statusofdistal
explanations–but that it is a veryheavyprice topay. Itmeans insisting thatonly themost
proximateexplanationofsomeexplanandumexplainsit.Andthisisabsurd!
SupposewesaythatFranzdevelopedlungcancerbecausehesmoked.Thisexplanationis, in
some sense, an extremely distal explanation of why he developed lung cancer. A more
proximal explanation would be that he regularly inhaled carcinogens. A more proximal
explanationstillwouldbethefactthatcellsinhislungsmutated.9
Whateachofthesemoreproximalexplanationshaveincommonisthatafullexplanationof
the fact that Franz developed lung cancer can be given in terms of them which makes no
mentionof,norevenentails, themoredistalexplanation.For instance,giventhatFranzwas
inhalingthatmixofcarcinogens,itdoesn’tmatter(ceterisparibus)whetherhegotthemfrom
smokingorfrompassivesmoking.Likewise,giventhatthecellsinhislungsmutateditdoesn’t
matter(ceterisparibus)tohisdevelopinglungcancerwhetherthecellmutationwastheresult
of carcinogen inhalation or exposure to radiation. So, if the exclusion principle is to be
believed, we can’t even say that Franz developed cancer because he regularly inhaled
carcinogens. According to the exclusion principle, only themostproximal explanation of his
lungcancerexplainsit.Thisissurelyabsurd.10
The absurdity of the exclusion principle should be clear whenwe see that even an agent’s
beliefs can be excluded from the explanation of their actions. Suppose that, when I
congratulatemyfriend,Ibelievethatcongratulatingherwillmakeherhappyand,say,Idesire
tomakeherhappy(orwhathaveyou).GiventhatIbelievethatcongratulatingherwillmake
her happy, the explanation of why I believe it is excluded from the explanation of why I
9‘TheformationofcovalentbondsbetweenthecarcinogensandDNAproducingDNAadducts,andtheresulting permanentmutations in critical genes of somatic cells is themajor established pathway ofcancercausationbycigarettesmoke.’(Hecht2006,609)10Yablo(2008)makesananalogouscriticismofRussel’sremarksaboutcausation,whichfollow:
If the cause is a process involving change within itself, we shall require… causal relationsbetween its earlier and laterparts;moreover itwould seem thatonly the laterparts canberelevanttotheeffect…Thusweshallbeledtodiminishthedurationofthecausewithoutlimit,andhowevermuchwemaydiminish it, therewill still remainanearlierpart,whichmightbealteredwithoutalteringtheeffect,sothatthetruecause…willnothavebeenreached.(Russell1917,135)
Russellsuggeststhatonlythe‘laterpart’(i.e.themoreproximalpart)ofthecausecanbetakentobethe cause itself, since the ‘earlier part’ (i.e. the more distal part) of the cause can obtain withoutguaranteeingtheeffect,ifthelaterpartdoesnotobtain.Thisisthecausalanalogueof (whatItaketobe)theimplicationoftheexclusionprinciple.InresponsetoRussell’sthesisaboutwhatittakestobeacause,Yablonotesthat, ‘ifthis…weretrulydisqualifying…essentiallyeverythingwouldberobbedof itsintuitive causal powers.’ (2008, 298) My point is that the exclusion principle has exactly the sameimplicationforexplanation: if itweretruethenessentiallyeverythingwouldberobbedof its intuitiveexplanatorypower.
166
congratulateher.That is, Idon’tneed tobelieve that shehaswonanaward tobelieve that
congratulatingherwillmakeherhappy,andif Idon’tbelievethatitwouldmakeherhappyI
wouldn’tcongratulateher,evenifIbelievedthatshehadwonanaward.ThefactthatIbelieve
that congratulating her would make her happy is a more proximal explanation of why I
congratulateher.ButitisabsurdtoinferfromthisthatthefactthatIbelievethatshehadwon
anawarddoesnotexplainwhyIcongratulatedher!
Theexclusionprincipleisadamagingexplanatoryprejudice–discriminatingagainstallbutthe
mostproximalexplanationsofanygivenexplanandumwillactuallyimpedeourabilitytooffer
the explanations that we ordinarily give, becausewe are typically interested inmore distal
explanationsofanygivenexplanandumthanthemostimmediatelyproximalexplanation.For
instance, if you’re looking toprevent lung cancer then itmatters that Franzdeveloped lung
cancerbecausehesmoked–knowingthatcellmutationexplainshiscancerdoesn’thelpyou
much.What’smostwrongwiththeexclusionprinciple,then,isthatitforcesustosaythatthe
explanationsweareinterestedinaren’treallyexplanations.11
4 Wheredidwegowrong?
Supposing you accept my reasoning, some questions still remain: if the fact that Jean was
transfusedwithHIV-infectedbloodexplainswhyhedevelopedAIDS then itmust bepart of
somefullexplanationofwhyhedevelopedAIDS.However,asestablishedin(b),thereisafull
explanation ofwhy he developed AIDS such that the fact that hewas transfusedwith HIV-
infected blood is neither a part of that explanation nor is it part of a genuinely different
explanation.Sohowcould itbepartofanyexplanation? Itwasexactlythis lineofreasoning
that,in§(VIII)5,ledustotheexclusionprinciple.Sincewehaveestablishedthattheexclusion
principleisfalse,weshouldseewhatwaswrongwiththisreasoning.
The mistake, I suggest, was the implicit assumption that if two full explanations of some
explanandum are not identical then they must be genuinely different explanations of that
explanandum.Thepointtorecogniseisthatfortwofullexplanationstobegenuinelydifferent
requiresmore than just non-identity, it requires,per the difference principle, some formof
independence.Thatis,theanswertothequestionofhowitcouldbepartofanyexplanationis
thattwofullexplanationscanbenon-identicalwithoutbeinggenuinelydifferent.Forinstance,
therearetwofullexplanationsofthefactthatJeandevelopedAIDSsuchthat,althoughthey
11Recallthatin§(VIII)5.2,InotedthatmyargumentagainsttheexclusionprincipleisnotanargumentagainstKim’sprincipleofcausalexclusion(inspiteoftheapparentsimilarityoftheirnamesandforms).This isbecause,asmayquitebeclear,Kim’sprincipledoesnotexcludedistalcausationsinceit isonlyrestrictedtotheexclusionofsimultaneouscauses.
167
arenotgenuinelydifferentexplanationsofwhyhedevelopedAIDS,theyarenonethelessnot
identical; and, in particular, the fact that he was transfused with HIV-infected blood is an
elementofoneandnottheother.
5 Whichexplanationaredistalexplanationspartof?
Anotherquestion:whichfullexplanationofwhyhedevelopedAIDSincludesthefactthatJean
was transfusedwith infectedblood?Weknow it cannotbe the same fullexplanationas the
onethatincludesthefactthathehadHIV,becausehowhecontractedHIVisredundantinthat
explanation.Sowhichfullexplanationisit?
Let‘[HIV]’standforthefactthatJeanhasHIVandlet‘Δ1’standforafullexplanationofwhy
JeandevelopedAIDSofwhich[HIV]isapart.Nowlet‘[Transfusion]’standforthefactthathe
hadaninfectedbloodtransfusionandlet‘Γ’standforafullexplanationof[HIV]thatincludes
[Transfusion].Finally, let ‘Δ2’standforthesetobtainedbysubstitutingΓfor[HIV] inΔ1.12My
suggestionisthis:Δ2isafullexplanationofwhyJeanhadAIDS.
That is, ifwesubstitute the fullexplanationofwhy JeanhadHIV,Γ, for the fact thathehad
HIV,[HIV], intothefullexplanationofwhyhedevelopedAIDS,Δ1,thatproducesanotherfull
explanation ofwhy he hadAIDS, Δ2.Now, Δ2 is clearly not identical to Δ1, but it is also not
genuinely different from it. Moreover, since [Transfusion] is included in Δ2, and Δ2 is a full
explanation of why he developed AIDS, [Transfusion] is a partial explanation of why Jean
developedAIDS.
To answer the question then, in ‘plain’ English: the fact that Jean had an infected blood
transfusionisapartofthefullexplanationofwhyhedevelopedAIDSthatisgotbysubstituting
thefullexplanationofwhyJeanhadHIVforthefactthathehadHIVintothemoreproximal
fullexplanationofwhyhedevelopedAIDS.
6 Conclusion
Ihavearguedthattheexclusionprincipleisfalseandthatseeminglyexcludedfactscanexplain
anexplanandumbyexplainingsomethingthat,inturn,explainsthatexplanandum.Inthenext
chapterIwillsetouttheprinciplesofmyindirecttheoryofnormativereasonexplanation.
12Thatis:Δ! = Γ ∪ Δ! ∖ 𝐻𝐼𝑉 .
168
(XII)
Explainingwhyweact
In which I say how normative reasons (and the appearance of them) explainwhy we act. I suggest that normative reasons explain an agent’s action byexplaining theirbelief that, in turn,explains theagent’saction. I suggest thatthey explain an agent’s belief by explaining the appearance of them that, inturn, explains the agent’s belief. I set out the implications of this view forexplanatory rationalism and for anti-psychological theories of reasons moregenerally.
In lightof theremarksof thepreviouschapter,myproposedanswertothequestionofhow
normativereasonsexplainouractionsisperhapsclear:
Factsaboutanagent’s[normative]1reasonsexplainanagent’sactionswhenevertheyexplainwhy she has the (true) beliefs she has about her [normative] reasons, beliefs that in turnexplainheractions.(Smith1998,38)
This is the indirect theoryofnormativereasonexplanation.Accordingtothis theorythefact
that my friend won an award explains why I congratulated her because it explains why I
believedthatshehadwonanaward,which,inturn,explainswhyIcongratulatedher.
Theindirecttheoryofnormativereasonexplanationhasbeenvariouslyconsidered,endorsed
or rejected, by others, though mainly in passing.2 However, I do not think it has been
consideredasthoroughlyasitoughttohavebeen,because,Isuggest,itsimplicationsforwhat
anagent’sreasonforactingcouldbeareprofound(regardlessofwhetherornotoneaccepts
explanatoryrationalism).
Moreover, as I will show, the indirect theory applies equally well as an account of how
experiencesexplainactions–thatis,anexperienceexplainsanagent’sactionifitexplainsthe
beliefthatexplainstheagent’saction.
In thenextchapter Iwill considerwhat the indirect theoryshouldsayabouthownormative
reasonsandtheappearanceofthemexplainwhyitisrationaltoact.
1Theseare‘justifying’reasonsinSmith’soriginal–thesubstitutionImakeispurelyterminological.2See:Collins(1997,111),Smith(1998,38),Dancy(2000,109–101),Davis(2005,74–75),Saporiti(2007,306),Raz(2009,197)andGibbons(2010,359).DustinLocke(2015)givesamorethoroughtreatmentofatheoryofthiskind,however,Locke’streatmentdifferssignificantlyfrommine,particularlyinsofarasitisfocusedonwhatitisforsomethingtobeanagent’sreasonforacting.
169
1 Whennormativereasonsexplain
Theprinciplebehindthe indirecttheory isthatwhenanagentactsforareason‘there isthe
“proximal”explanationoftheaction,givenbyspecifyingthepsychologicalstateoftheagent.
Thenthereisthe“distal”explanationoftheaction,givenbyspecifyingwhatisresponsiblefor
theagentgetting into that state.’ (Dancy2000,109)For this tobea fruitfulaccountofhow
normativereasonsexplainactions, itmustbethecasethatanormativereasoncanbe‘what
gets’anagentintothestateofbelievingthatnormativereason.
FigureXII-1:Theindirecttheoryofnormativereasonexplanation
Buthowdoesafactabouttheworldexplainone’sbelief in it?Thatexplanatoryrelation is, I
suggest,alsoindirect.
1.1 Hownormativereasonsexplainbeliefsinthem
Why do I believe that the earth is spherical? Because it is spherical! That which I believe
explainswhyIbelieveit.Likewise:ifI’mbeingrainedon,andI’mjumpingtoavoidthepuddles
andIseeothersrushingforshelter,ifyouweretothenaskmewhyIbelieveit’srainingIwill
reply,incredulously,‘Becauseitisraining!’
HowcouldthefactthatitisrainingexplainwhyIbelievethatitisraining?Perhapsyouthink
alongthese lines: it’snotthefact that itwasrainingthatexplainswhy Ibelievedthat itwas
raining,itisonlythefactthatitappearedtomeasthoughitwasraining.Ifitweren’training
butitstillappearedtomeasthoughitwas,Iwouldstillhavebelievedthatitwas.Andsoon.
Thisview,asIarguedin§(IX)5,isjustanotherinstanceofTheExplanatoryExclusionProblem,
which the previous chapter discredited. In short: even though how things appear to be
intermediate between the world and our beliefs about it, that does not prevent the world
frombeingabletoexplainthosebeliefs.
Towit:thefactthatthingsappeartobeacertainwaymaybethemoreproximalexplanation
ofanagent’sbeliefthattheyarethatway.However,ifthingsactuallyarethatway,andthey
appear to be that way because they are that way, then, given the transitivity of the
explanatory relations involved, the fact that they are that way can explain why an agent
believesthemtobethatway.Thisis,Isuggest,anindirecttheoryofbeliefexplanation.
Psychologicalfactexplains explains
Normativereason
Action
170
FigureXII-2:Theindirecttheoryofbeliefexplanation
SupposethatIbelievedthatitwasrainingbecauseI lookedoutofthewindowandsawrain.
Whydiditappeartomeasthoughitwasraining?Well,inpart,becauseitwasraining.Thefact
thatitwasrainingexplainswhyIsawrainandthuswhyitappearedasthoughitwasraining,
which explains why I believed that it was raining. So, because the explanatory relations
involvedaretransitive,thefactthatitwasrainingexplainswhyIbelievedthatitwasraining.
Ifyoudoubtthattheexplanatoryrelationsinvolvedaretransitive,pleasesuspendyourdoubts
forthemoment:thisisasubjectIwilldiscussatlengthinsubsequentchapters.
Likewise,Ibelievedthatmyfriendhadwonanaward,inpart,becauseIreadthatshehadwon
anaward.AndIreadthatshehadwonanaward,inpart,becauseshehadwonanaward.3The
factthatshehadwonanawardexplainswhyIreadthatshehadwonanaward,whichexplains
whyIbelievedthatshehadwonanaward.Again,becausetheexplanatoryrelationsinvolved
aretransitive,wecansaythatthefactthatmyfriendwonanawardexplainswhy Ibelieved
thatshehadwonanaward.
So,thewaythatnormativereasonsexplainourbeliefsinthemisbyexplainingwhyitseemed
tousasthoughthosenormativereasonswerethecase.
1.2 Hownormativereasonsexplainactions
Whattheindirecttheoryofnormativereasonexplanationsuggestsisthis:Icongratulatedmy
friend,inpart,becauseIbelievedthatshehadwonanaward,andIbelievedthatshehadwon
anaward,inpart,becauseIreadthatshehadwonanaward,andIreadthatshehadwonan
award,inpart,becauseshewonanaward.So,becausetheexplanatoryrelationsinvolvedare
transitive,wecansaythatIcongratulatedmyfriend,inpart,becauseshewonanaward.
FigureXII-3:TheexplanationofwhyIcongratulatedmyfriend
3We could spell out the chain further: I read that she had won an award, in part, because thenewspaperprintedanarticleabout it,andtheyprintedanarticleabout it, inpart,becauseareporterwroteanarticleabout it,andtheywroteabout it, inpart,becausetheywitnessedherwintheawardandtheywitnessedherwinit,inpart,becauseshewonit.Thepointisthatgiventhetransitivityoftheexplanatoryrelationsinvolvedthereisnorequirementtospellouttheseintermediatesteps.
Howthingsappeartobeexplains explains
Theworld Belief
Ireadthatshehadwonan
award
Myfriendwonanaward
Ibelievedthathadshewonan
awardexplains
Icongratulated
herexplains explains
171
1.3 Whennormativereasonsdon’texplain
Recallthatwedon’ttendtogivenormativereasonexplanationsininstancesoffalsebeliefsor
in Gettier cases – we say that, in those cases, the normative reason does not explain the
agent’s action. The indirect theoryofnormative reasonexplanation tellsuswhy:because in
both falsebelief casesandGettier cases, thenormative reasondoesnotexplain theagent’s
beliefinit.
First, a false belief case: when Sally ran because she (mistakenly) believed that a bear was
chasingher,wedon’tsaythatsheranbecauseabearwaschasingher.Whynot?Becausethe
proposition thatabearwas chasingher doesnotexplainwhy shebelieved that abearwas
chasingher.Clearenough.
FigureXII-4:TheexplanationofwhySallyran
Second,aGettiercase:recallthatEdmund’snormallyreliablefriendtoldhim,onawhim,that
theiceinthemiddleofthelakewasthin,althoughshehadnoideaabouttheactualstatusof
theice.AsaresultEdmundskatedbytheedgeofthelake;thatis,heskatedbytheedgeofthe
lake,inpart,becausehebelievedthattheiceinthemiddlewasthin.
Now, as it turns out, it actually was thin. However, as we have already acknowledged (see
§(IV)1.4),thefactthattheiceinthemiddleofthelakewasthindoesnotexplainwhyEdmund
skatedattheedge.Whynot?Wellaccordingtothe indirecttheorytit isbecause itdoesnot
explainwhyhebelieved that itwasthin;and itdoesnotexplainwhyhebelievedthat itwas
thinbecause itdoesnotexplainwhyhis friendtoldhimthatthe icewasthin (which iswhat
explainswhyhebelievedthatitwasthin).
FigureXII-5:TheexplanationofwhyEdmundskatedattheedgeofthelake
2 Implicationsforexplanatoryrationalism
Recall that explanatory rationalism requires that an agent’s reason for acting must both
explaintheiractionandexplainwhyitwasprotantorationalforthemtodoit.AsIsetoutin
Sheheardabear-likesound
AbearwaschasingSally
Shebelievedthatabearwaschasingherexplains
Sheranexplains
Edmund’sfriendtoldhimthattheicewasthin
Theiceinthemiddleofthelakewasthin
Hebelievedthattheicewasthinexplains
Heskatedattheedge
explains
172
§(VI),ifexplanatoryrationalismistobeconsistentwiththeprimafaciereasonableclaimsset
outin§§(II)-(IV),thefollowingmustbetrue:
(R1) Icongratulatedmyfriendbecauseshehadwonanaward.
(R2) IcongratulatedmyfriendbecauseIreadthatshehadwonanaward
(R3) Itwaspro tanto rational forme to congratulatemy friend because she had
wonanaward.
(R4) ItwasprotantorationalformetocongratulatemyfriendbecauseIreadthat
shehadwonanaward.
Thediscussionoftheprevioussectionhasdemonstratedhowitisthat(R1)and(R2)aretrue:
theexplanans ineachstatementexplainswhy I congratulatedmy friendbyexplainingwhy I
believedthatshehadwonanaward.
FigureXII-6:Explainingwhyweact
In the next chapter I will discuss (R3) and (R4), but before then I want to discuss the
implications of the indirect theory of normative reason explanation for anti-psychological
theoriesofreasonsmoregenerally.
3 Implicationsforanti-psychologicaltheoriesofreasons
RecallthatTheExplanatoryExclusionProblemwasthemotivatingargumentforpsychologism
aboutthereasonsforwhichweact.Thereasoningwentlikethis:anagent’sreasonforacting
must always explain their action (recall §(IV)1.2), however, only features of an agent’s
psychologycanexplaintheiractions(asperTheExplanatoryExclusionProblem),therefore,an
agent’sreasonforactingmustalwaysbeafeatureoftheirpsychology.
Thisargumentis,aswehavenoted,themajorargumentagainstanytheorythatsuggeststhat
reasonsaresometimesnotfeaturesofourpsychology.4Therejectionoftheexclusionprinciple
underminedthatargument,strengtheningthecaseforanti-psychologicaltheories.
Theindirecttheoryofnormativereasonexplanationnowprovidesuswithanaccountofwhyit
iswrong:normativereasonscanexplainactions,andtheydosobyexplainingthebeliefsthat
4Note: anti-psychological theoriesdon’t insist that reasonsarenever featuresofourpsychology, justthattheyatleastsometimesaren’t.
Howthingsappeartobeexplains explains
Normativereason Belief
explainsAction
173
explain our actions. This theory does not deny the primacy of the psychological, on which
psychologisminsists,butitnonethelessprovidesarolefortheworldinexplainingwhatagents
do. The indirect theory of normative reason explanation thus provides a response to
psychologismforallanti-psychologicaltheories,andnotmerelyexplanatoryrationalism.
4 Conclusion
One of the challenges I set myself was showing how it was that normative reasons could
explain an agent’s action. I take that challenge to have now beenmet. In particular, I have
arguedthat, regardlessofwhetherornotyouacceptexplanatoryrationalism(andtheother
arguments for it that follow), the indirect theory of normative reason explanation can still
provideyouwithanaccountofhowitisthatnormativereasonscanexplainouractions.
Thenextchapterdiscusseshowit isthatnormativereasons(andperceptualexperiencesand
thelike)canexplainwhyitisrationalforanagenttodosomething.Beforethatdiscussion,in
the Appendix to this chapter, I consider two potential objections to the indirect theory of
normativereasonexplanation.
Appendix
A.1 Objections
Objection1 Youhavesuggestedthatanormativereasonexplainsanagent’sactiononlyif
it explains a feature of their psychology that explains their action. However, here is a
counterexampletothatclaim:
Thatitisabouttorainmayexplainwhyeveryoneiscomingin.Istheirbeliefthatitisabouttoraintobeexplainedbyitsbeingabouttorain?Orisitrathertheblacknessofthecloudsandthesuddendrop in temperature? These are not themselves to be explained by its being about to rain. Thecloudsarenotblackbecauseitwillshortlyrain.(Dancy2000,112)
For expositional simplicity, let’s restrict Dancy’s example to claims about someone in
particular;callhim‘Jim’.ForJim,theargumentgoeslikethis:
(a) ThefactthatitisabouttorainisanormativereasonforJimtocomein;
(b) ThefactthatitisabouttorainexplainswhyJimiscomingin;and
(c) ThefactthatitisabouttoraindoesnotexplainwhyJimbelievesthatitisabouttorain.
174
If (a), (b) and (c) are all true then it is not true that, as I have argued, a normative reason
explainsanagent’sactiononlyif5itexplainsafeatureoftheirpsychologythatinturnexplains
theiraction.6
Response Iamhappytoagreethat(a)istrue,soImustrejecteither(b)or(c).Thereare
some7whoreject(b),onthegroundthatfactsaboutthefuturecan’texplainpresentactions.I
have some sympathy for that response; however, itmay not fare sowell with facts known
from inference, and may rely on a causal analysis of the explanatory relations involved.
Regardless, to mymind there is a more compelling response: I do not know how one can
consistentlymaintainboth(b)and(c).
TheonlyreasonthatDancygivesindefenceof(c)isthattheblacknessofthecloudsetc.isnot
explainedby the fact that it is raining.That is trueenough.But tomakethatargument is to
embracean indirect theoryofbeliefexplanation–namelythatsomefactpexplainsthefact
thatanagentbelievesthatponlyifitexplainssomefactthat,inturn,explainsthefactthatp.
Thatis,Dancypresupposestheindirecttheoryofbeliefexplanationinsuggestingthatthefact
thatitisabouttoraincan’texplainwhyJimbelievesthatitisabouttorainbecauseitdoesnot
explain the blackness of the clouds etc.My concern is that I don’t knowwhywe should be
indirecttheoristsaboutbeliefexplanationbutnotaboutactionexplanation.
Ofcourse,ifoneweretodeny(c),andinsistthatthefactthatitisabouttoraindoesexplain
why Jim believes that it is about to rain, then one might wonder what the explanatory
connection between those two facts is (given that it is unmediated by any perceptual
experience).However,Idon’tseehowonecanwonderthiswithoutlikewisewonderingwhat
theexplanatoryconnectionbetweenthefactthatitisabouttorainandthefactthatJimcame
inis.Mypointisthis:Idon’tseewhatbasisonecouldhaveforthinkingthatboth(b)and(c)
aretrue–eitheroneaccepts(b)and,forthesamereason,rejects(c);oroneaccepts(c)and,
forthesamereason,rejects(b).
So,ifthereisnoconsistentbasisforthetruthofboth(b)and(c),thenthisisnoobjectionto
theindirecttheoryofnormativereasonexplanation.
5Atleastinnon-weirdcases.6Anargumentofthiskindcanlikewisebemadeforfactsthatarebelievedonthebasisofinference(i.e.whereaperceptualexperiencethatexplainsone’sbeliefthatpisalsoseeminglyisnotexplainedbythefact that p). Alvin Goldman (1967) considered these sorts of examples in his causal theory ofknowledge,which is similar to the indirect theory in several respects.However,his response to themdiffersfrommine.7Forinstance:Davis(2003,456),Gibbons(2010,359)andLocke(2015,194).
175
Objection2 Evenifweacceptthisindirecttheoryofnormativereasonexplanation,itdoes
notguaranteethatyoucongratulatedyourfriendbecauseshewonanaward(i.e. itdoesnot
guaranteethetruthof(R1)).Whynot?Becauseonemaybescepticalastowhetherornotthe
factthatyourfriendwonanawardreallyexplainsthefactthatyoubelievedthatshehad.
Forinstance,supposethatyourfriend’sawardisparticularlyobscureandyoujusthappento
stumbleuponasmallarticleaboutitinanewspaperthatyouwouldn’tnormallyread.Under
these circumstances then (at least on, for instance, a difference-making account of
explanation)onemightsaythatthefactthatyourfriendwonanawarddoesnotreallyexplain
whyyoubelievedthatshehad,8anditconsequentlydoesnotexplainwhyyoucongratulated
her.Thus, in spiteofyour indirect theoryofnormative reasonexplanation, (R1)maystillbe
false.
Response Ihave three responses to thisobjection.First, I could justaccept that (R1) is
not truewhenmy discovering aboutmy friend’s award is so chancy. Theremight be other
examples (e.g. my friend wins a Nobel Prize), for which the explanatory connection is
sufficientlyrobusttocounterfactualsforthefactthatmyfriendwonanawardtocountasan
explanationofmyaction.Onestrategycouldthusbetojustrestrict(R1)toaclaimaboutsuch
cases.Therewouldbelittlelostformytheoryinmakingsucharestriction.
Mysecondresponse,however,istonotethatevenifmyfriend’sawardhadbeenobscure,and
I had only foundout about it because I stumbled upon an article about it in a newspaper I
wouldn’tnormallyread,Istillthinkthatthefactthatmyfriendwontheaward(partly)explains
whyIbelievedthatshedidand,therefore,whyIcongratulatedher.So,totheextentthatyour
account of explanation implies that it doesn’t, it is not really the sort of account that I am
anywayinclinedtoaccept.9
8Since there are very nearby possible worlds in which you don’t see the article, but in which shenonethelesswinstheaward.9Whatmaybeatworkhere isanexplanatoryanalogueofNedHall’s (2004)claimthat thereare twoconcepts of causation: productive and counterfactual dependence. I suggest that difference-makingaccountsofexplanationcanbeunderstoodasparticularkindofexplanatoryanalogueofcounterfactualdependenceconceptsofcausation(evenwithoutrestrictingthediscussiononlytocausalexplanation).Now,whilstIdon’twanttorejectdifference-makingaccountsofexplanation,IwouldsuggestthatthereisabonafidemannerofexplanationthatistheexplanatoryanalogueofHall’sproductivecausation.TherelevanceofthisobservationtothisdiscussionisthatHalldemonstratedthatonecanhaveproductivecausationwithoutcounterfactualdependence,andviceversa.Thus,mysuggestion(modestlymade)isthattheremaybetwodifferentconceptsofexplanationthatcanlikewisecomeapart.Thus,itmaybethateventhoughthefactthatmyfriendwonanawarddoesnotexplainwhyIbelievedthatshehadifwehaveinmindaparticularsortofcounterfactualdependenceconceptofexplanation(i.e.difference-making),itmaynonethelessstillexplainitifwehaveinmindtheproductiveconcept.
176
Mythirdresponse:inthechaptersthatfollowIarguethatifanagentknowsthatpthenthat
entailsthatthereisaparticularexplanatoryrelationbetweenthefactthatpandthefactthat
theybelievethatp. Ifyouacceptthis,thenyoumusteitherrejecttheclaimthatIknewthat
myfriendhadwonanawardoracceptthatthefactthatmyfriendhadwonanawardexplains
why Ibelievedthatshehad.10Since Ido thinkthat (even inthechancycase) Iknewthatmy
friendhadwonanaward,Iamalsoinclinedtothinkthatthefactthatshehadwonanaward
explains why I believed that she had, and, thereby, explains why I congratulated her.
However,if,inthechancycase,youdon’tthinkthatIknewthatmyfriendhadwonanaward,
then you could (aswedid for the first response) just restrict (R1) to a claim about cases in
whichIknewthatmyfriendwonanaward,withoutmuchlossformytheory.
10If you reject the claim about knowledge that I go on to make you are, of course, under no suchcompulsion.However,ifyourejectthisclaimthenyouwillperhapshavemoresignificantobjectionstomytheorythanjustthis.
177
(XIII)
Explainingwhyitisrationaltoact
Inwhich I saywhen something explainswhy it’s rational to act, andwhen itdoesn’t. I suggest that normative reasons or appearances explain why it isrational to act only if they explain those beliefs that in turn explainwhy it isrationaltoact.Inotethatitistemptingtoinferthatifanagent’sbeliefexplainswhyitisrationalforthemtodosomeactionthenwhateverexplainsthatbeliefalso explains why it is rational for them to do that action. I show how thatinference leads toanapparentdilemmaforexplanatory rationalism. I counselagainst that inference, by noting that different kinds of explanatory relationsmaynotbetransitivewitheachother. I thensetout thetaskahead:showingthat the explanatory relations concernedare transitivewhen, andonlywhen,explanatoryrationalismneedsthemtobe.
Thereisaseeminglyclearwayinwhichtheaccountofthepreviouschaptercouldalsobeused
asanaccountofhowitisthatnormativereasons(ortheappearanceofthem)explainwhyitis
rational foranagent toact:wecansay thatanormative reason (or theappearanceofone)
explainswhy it is rational for an agent to act if it explains the belief that explainswhy it is
rationalfortheagenttoact.
So,forinstance,whenIlookoutofthewindowandseerain,wecansaythatthefactthatitis
rainingexplainswhyitisrationalformetotakemyumbrellabecauseitexplainswhyIbelieved
that itwas raining.Thisproposalvindicatesexplanatory rationalism’sclaimthat thingsother
than features of an agent’s psychology could explain why it is rational for them to do
something.
However,asIwillargue,weshouldnotassumethatjustbecausesomefactexplainsanagent’s
beliefandtheirbeliefexplainswhyitisrationalforthemtodosomeaction,thatthatfactalso
explainswhyitisrationalforthemtodothataction.Thatis,Isuggest,thereareoccasionson
whichtheexplanatoryrelationsinvolvedaren’ttransitive.
In the next chapter I will provide an account of what separates those cases in which the
explanatoryrelationsinvolvedaretransitivefromthoseinwhichtheyaren’t.
1 Anotherindirecttheory
RecallthatIneedtodemonstratethefollowing:
(R3) Itwaspro tanto rational forme to congratulatemy friend because she had
wonanaward.
178
(R4) ItwasprotantorationalformetocongratulatemyfriendbecauseIreadthat
shehadwonanaward.
The question of how these claims could be true is an instance of amore general question,
namely:howisitthateithernormativereasons(quafactsabouttheworld)ortheappearance
ofthemcouldexplainwhyitis(protanto)rationalforanagenttoact?Myansweristhatthey
explainindirectly.
FigureXIII-1:Explainingwhyitisrationaltoact
The fact that it is raining explains why it appeared to me as though it was raining, which
explainswhyIbelievedthat itwasraining,whichexplainswhyitwasrationalformetotake
myumbrella.So,becausetheexplanatoryrelationsinvolvedaretransitive,thefactthatitwas
rainingexplainswhyitwasrationalformetotakemyumbrella.
Likewise, as we established in the previous chapter, the fact thatmy friendwon an award
explainswhy I read that shehadwonanaward,whichexplainswhy Ibelieved that shehad
wonanaward,whichexplainswhy itwasprotantorational formetocongratulateher.And
this is why (R3) and (R4) are true: we can say that it was pro tanto rational for me to
congratulatemyfriendbothbecauseshewonanawardandbecauseIreadthatshehadwon
anaward,andwecansaythatbecausetheexplanatoryrelationsinvolvedaretransitive.
2 Isexplanationtransitive?Anapparentdilemma
Ihavesaidthatnormativereasonsandhowthingsappeartobecanexplainwhyitisprotanto
rational foranagenttodosomeactiononly if,andbecause, theyexplainoneoftheagent’s
beliefs,which,inturn,explainswhyitisprotantorationalforthemtodothataction.
AsIhavenoted,thisindirectmannerofexplanationreliesonthetransitivityoftheexplanatory
relations involved. With that in mind, we might then reason from this observation to the
followingtheoryaboutwhatexplainswhyitisrationaltoact:
- TheNaïveTheory:ifthefactthatAbelievesthatpexplainswhyitisprotantorationalforA toφ thenwhateverexplainswhyAbelievesthatpalsoexplainswhyitwasprotantorationalforAtoφ.
Howthingsappeartobeexplains explains
Normativereason Belief
explains
Rationalityofaction
179
Inwhat follows Iwillarguethat thenaïvetheory is false.Before Idoso,however, Iwant to
show how the naïve theory might seem to create a dilemma for explanatory rationalism,
irrespectiveofwhetheritistrueorfalse.
Recallwhatexplanatoryrationalismhastosayaboutthereasonsforwhichweact:
- Explanatoryrationalismaboutthereasonsforwhichweact:Foranyp,pisA’sreasonforφingifandonlyifpexplainswhyitisprotantorationalforAtoφandexplains(intherightway)whyAφ’d.
This,combinedwiththenaïvetheory,meansthatifAφsbecauseAbelievedthatpanditwas
protantorationalforAtoφbecauseAbelievedthatp,thenwhateverexplainswhyAbelieved
thatpisalsoA’sreasonforφing.
Butnoticethatsometimeswecanexplainwhysomeonebelievedsomethingbycitingthings
thatareplainlynotthereasonsforwhichtheyact,suchasbrainaneurismsorpsychosis.Were
thenaïvetheorytrue,then,theexplanatoryrationalistwouldbeforcedtoconcludethat,ifan
aneurismexplainswhyanagenthassomebelief,andtheyact rationallyon thatbelief, then
theirreasonforactingisthattheyhadabrainaneurism.Thisisanundesirableconclusion.1
Andyet, if thenaïve theory isn’t true, thenexplanation isnotagenerally transitive relation.
Butthe indirecttheoryofnormativereasonexplanation,and,moregenerally,myaccountof
why(R1)to(R4)aretrue,reliesonthetransitivityofexplanation.
Here, then, is the apparent dilemma: either (i) we accept the naïve theory, which means
rejectingexplanatory rationalismabout reasons forwhichweact;2or (ii)wereject thenaïve
theorybyrejectingthetransitivityofexplanation,whichmeansgivinguponthetruthof(R1)
to(R4).3Thenextsectionisdevotedtoarguingthatthisdilemmaisonlyapparentandthatthe
appearanceofadilemmais theresultofamistake–thesamemistakethat ledtothenaïve
theory.
1Ofcourse,thefactthatonehasabrainaneurismorpsychosiscouldbeareasonforwhichonedoessomething (e.g. goes to the doctor). However, if that brain aneurism just causes one to have anirrelevantbeliefthroughsomearationalprocess,wewouldtypicallynotwanttoconcludethatwhenanagentactsbecauseofthatbelief,theirreasonforactingisthattheyhadabrainaneurism.2Orsayingthatwhenabrainaneurismcausesmetobelievethatitisrainingthatmyreasonfortakingmy umbrella was that I had a brain aneurism. I take this to be worse than rejecting explanatoryrationalism.3 Which means, in turn, that explanatory rationalism reduces to psychologism. It reduces topsychologismbecause,ifexplanationcan’tbetransitive,thenonlyfeaturesofanagent’spsychologycanexplainwhytheyactorwhyitisrationalforthemtoact.
180
3 Theapparentdilemmaisnotadilemma
Onesimple,butmistaken,waytoarriveatthedilemmaistothinkthatthereisonlyonekind
ofexplanatoryrelation.Werethattruethenanyinstanceofafailureoftransitivity(i.e.ifthe
naïvetheorywerefalse)wouldmeanthatexplanatoryrelationsarenottransitive.4However,
as the discussions of previous chapters have suggested, there is not just one kind of
explanatoryrelation.Wecanatleastdistinguishcausalfromnon-causalexplanatoryrelations,
andIsuggestthatthereareprobablystillmorefine-graineddistinctionsbetweenexplanatory
relationsthatthatdistinctionignores.
Amorenuancedbut,as Iwillargue,equallymistakenapproach is tomaintain thatalthough
there are different kinds of explanatory relation, if any explanatory relations are transitive
thentheyarealltransitivewitheachother.Isuggestthatit isthisfallaciousassumptionthat
ledtothenaïvetheory.
Myargument against this assumptionproceeds in two stages: firstly Iwill demonstrate that
someexplanatoryrelationsaretransitive;secondlyIwillgiveanexampleinwhichexplanatory
relationsarenot transitive.Together theseamount toacounterexample to theclaim that if
anyexplanatoryrelationsaretransitivethentheyallare.
3.1 Someexplanatoryrelationsaretransitive
Firstly,as§(XI)demonstrated,wehavegoodreasonstothinkthatsomeexplanatoryrelations
are transitive: if they weren’t then distal explanations of some explanandumwould never
reallybeexplanationsofthatexplanandum.5Since,inbothordinaryandscientificlife,weare
mostly interested in somewhat distal explanations, and since the explanations we normally
givearesomewhatdistal,ourordinaryandscientificexplanatorypracticeassumesthatatleast
someexplanatory relations are transitive. So, Iwill take it as a given that someexplanatory
relationsaretransitive,becausethatisthebestaccountofwhydistalexplanationsthatinvolve
suchrelationsexplain.
4Insofarasachainofexplanatoryrelationsdoesnotguaranteeanexplanatoryrelationlinkingthetwoexplanationsofthechain.5Ofcourse it’spossiblethatsomedistalexplanationsexplainforreasonsotherthanthetransitivityoftheexplanatoryrelationsinvolved.I’mnotsurewhatthosereasonscouldbe,butonemightbeabletoconstructexamples.Mypointisratherthatthetransitivityoftheexplanatoryrelationsinvolvedisreallythebestand simplestaccountofhowdistalexplanationsexplain,and it reallydoesexplainhow theyexplaininatleastsomecases.
181
3.2 Notallexplanatoryrelationsaretransitivewitheachother
InthissectionIwillprovideanexampleinwhichthetransitivityofexplanationfails,andIwill
suggest that it is because the explanatory chain involves different kinds of explanatory
relationsthataren’ttransitivewitheachother.
3.2.1 Whenexplanationisn’ttransitive
RecallthatSallybelievedthatabearwaschasingherbecausesheheardabear-likenoise.Why
didshehearabear-likenoise?Becausethewindrustledthetreesinsuchandsuchaway.That
is,thenoisethatSallyheardhadnothingtodowithanybear,butitnonethelesssoundedvery
bear-like.
WehavealreadynotedthatitwasprotantorationalforSallytorunbecauseshebelievedthat
abearwaschasingher.Andwhatexplainsherbeliefthatabearwaschasingherwasthatshe
heard a bear-like noise (it appeared to her as though a bear was chasing her), and what
explains thatwas the fact that the trees rustled (in theway that they did). Thus, there is a
chainofexplanatoryrelationsfromthefactthatitwasprotantorationalforSallytoruntothe
fact that the trees rustled. But shouldwe say that itwaspro tanto rational for Sally to run
(evenpartly)becausethetreesrustled?
Idon’t thinkweshould. Itseemsasthougheven ifoneweretogivethissortofexplanation
one would then be forced to add that Sally believed that a bear was chasing her. And, I
suggest,thatisbecauseitisonlyreallythebeliefthatisdoingtheexplaining.6
However, if the fact that the trees rustled does not (even partially) explain why it was pro
tantorationalforSallytorun,7thentheexplanatoryrelationsinvolvedinthiscasemustnotbe
transitive.8
Perhapsyouobject:thisdoesnotviolatetransitivitybecausetheoddrustlingofthetreesdoes
not explainwhy she heard a bear-like sound.Of course, you say, it explainswhy she heard
something,butitdoesn’texplainwhythatsoundsoundedlikeabear.Whilethiswouldbeall
thebetterformeifIagreedwithit,unfortunatelyIdon’t:ofcourseotherfactorsmaydomore
toexplaintheparticularcharacterofthenoisemade,butthefactthatthetreesrustledinthat
wayis,Isuggest,apartoftheexplanationofwhysheheardabear-likenoise.
6IfyouworrythatthisistooclosetotheellipticalviewthatIrejected,notehowwearen’tforcedtoaddanythingaboutthebeliefwhenwesaythatItookmyumbrellabecauseitisraining.7Ofcourse,ifyourviewofexplanationisthatitispurelyarelationofcounterfactualdependence,thenitdoesexplainit.However,Ithinkthatisawrongaccountofexplanation(seemyresponsetoObjection2intheAppendixtothepreviouschapter).8Thisfollowsnecessarily.
182
3.2.2 Whyexplanationwasn’ttransitive
The rustling in the treesexplained thenoise thatSallyheard, thenoiseexplainedherbelief,
and her belief explained why it was rational for her to run. However, I’ve suggested, the
explanatoryrelationshere9aren’talltransitivewitheachother.Whynot?Because,Isuggest,
theyaredifferentsortsofexplanatoryrelation.
Whatsortofexplanation istheexplanationofwhy itwasprotantorational forSallytorun?
Well, it isn’tcausal.Even ifweallowthatourbeliefscauseouractions, it’s stillnot thecase
that they cause it to be rational for us to act – anymore than the fact that a person is in
troublecauses it toberight tosavethem.Causation justseemstobethewrongconcept to
invoke when describing this sort of explanatory relation. That is, the explanatory relation
betweenmybeliefandtherationalityofmyactionisanon-causalexplanatoryrelation.
Incontrast,theonlysenseinwhichtherustlingofthetreesexplainswhySallyheardabear-
likesoundisacausalone;thatistheexplanatoryrelationhereisacausalexplanatoryrelation.
So, theexplanatory relationsbetween (i) the rustling treesand thenoiseSallyheardand (ii)
herbeliefandtherationalityofrunningareclearlyoftwodifferentkinds.AndIsuggestthat
thedifferentcharacteroftheexplanatoryrelationsinvolvediswhytheyaren’ttransitivewith
eachother.NotethatI’mnotsayingthatcausalexplanatoryrelationsarenevertransitivewith
non-causalexplanatoryrelations10–all Iamsayingisthattheseparticularsortsofcausaland
non-causalexplanatoryrelationsaren’ttransitive.
3.3 Theapparentdilemmaisn’tadilemma
Isaidthatthenaïvetheoryandtheapparentdilemmaitproduceswere(atbest)theresultof
the assumption that if anyexplanatory relations are transitive thenall explanatory relations
are transitive with each other. The argument of the previous two sections disproves that
assumption–whilesomekindsofexplanatoryrelationsaretransitive, itdoesnotfollowthat
differentkindsofexplanatoryrelationsaretransitivewitheachother.
Thus, there is no dilemma for explanatory rationalism because the naïve theory and the
consequent apparent dilemmawerebasedon themistaken assumption that all explanatory
relationsmustbetransitivewitheachother.
9That is:between (i) the treesand thenoise; (ii) thenoiseand thebelief; and (iii) thebeliefand therationalityofrunning.10That is, theremaybeother kindsof causal andnon-causal explanatory relations that are transitivewitheachother.
183
However, explanatory rationalism is not out of difficulty yet, in the next section I will
discussion a new challenge posed by this response. Namely this: why should it be that the
explanatoryrelationsinvolvedaretransitivewheneverexplanatoryrationalismneedsthemto
be,andnotwhenitneedsthemnottobe?
4 Thechallenge
We are talking about three sets of explanatory relations: the relation from theworld to an
appearance, from an appearance to a belief, and then the relation of that belief to the
rationalityofanagent’saction.WhatIseemtobesayingabouttheseexplanatoryrelationsis
this:sometimestheyaretransitivewitheachother,andsometimestheyaren’t.Inparticular,I
amsaying that theexplanatory relations inFigureXIII-2are transitive,while theexplanatory
relationsinFigureXIII-3aren’t.
FigureXIII-2:Achainofexplanatoryrelationsthatarealltransitive
FigureXIII-3:Achainofexplanatoryrelationsthatarenotalltransitive
Isn’t this just ad-hoc? Why should we suppose that the relations are transitive when
explanatoryrationalismneedsthemtobe,butaren’twhen itneedsthemnottobe?That is
theremainingchallengeforexplanatoryrationalism.
I want to consider one response to this challenge that is natural, but won’t work, before
settingthestageformyownresponse.
5 Theunsuccessfulnaturalstrategy
A natural strategymight be to reason as follows: perhaps in the second case the failure of
transitivity is not in the explanatory relation per se, but in the putative explanans (i.e. the
rustlingofthetrees);it’sneitherthecontentofSally’sbeliefnorisitsomethingthatherbelief
was based on – it’s of no epistemic importance. And perhaps because it’s of no epistemic
importance, it’s just thewrong sort of thing to do the right sort of explaining. I could then
refinemy theory in away that excluded these sorts of (epistemically unimportant) putative
Ireadthatshehadwonan
awardnewspaper
Myfriendwonanaward
Ibelievedthatshewonan
awardexplains
Congratulatingherwasprotantorationalexplains explains
Sallyheardabear-likesound
Thetreesrustled
Shebelievedthatabearwaschasingherexplains
Runningwasprotantorational
explainsexplains
184
explanans for some principled reason, and I could then insist that whatever met those
conditionsandexplainedmybelieftherebyalsoexplainedwhatthebeliefexplained.
There are twoproblemswith this strategy: firstly, it is not at all clear tome that one could
make such a restriction without it still being, in some respect, ad-hoc. Secondly, andmore
seriously,explanatoryrelationsmayfailtobetransitiveevenwhentheputativeexplanansisa
normativereason(i.e.therightsortofthing),asthenextexampledemonstrates.
5.1.1 Anotherexampleoffailureofexplanatorytransitivity
SupposethatIaminasealedroomthat,unbeknownsttome,isslowlybeingfilledwithcarbon
monoxide. Suppose further that, after a short while, the carbon monoxide causes me to
hallucinate,and,bysheerchance, Ihaveahallucinationofareliable friendbursting intothe
roomandwarningmethatitisfillingupwithcarbonmonoxide.Itisrationalformetoleave,11
andIdulydoso.12
The fact that the room was filled with carbon monoxide (partly) explains why I had the
hallucination.Ofcourse,itisonlyapartofthatexplanation–otherfactorsofmypsycheand
the likewilldomore toexplain the content ofmyhallucination,but the fact that therewas
carbonmonoxide in the room is, I suggest, a part of the full explanation ofwhy I had that
hallucination.
Moreover,thefactthatIhadthathallucinationexplainswhyIbelievedthattherewascarbon
monoxide in the room.And the fact that I believed that therewas carbonmonoxide in the
roomexplainswhyitwasprotantorationalformetoleave.
Again,wehaveanexplanatorychainconnectingthefact that therewascarbonmonoxide in
the roomwith the fact that itwaspro tanto rational formeto leave.Butdoes the fact that
therewascarbonmonoxideintheroomexplainwhyitwasprotantorationalformetoleave?
Idon’tthinkso.
It’stemptingtoputitlikethis:Iwasn’treallyawareofthatfact.Iamnotreallyrespondingto
the fact that there is carbonmonoxide in the roombecausemybelief is just truebyhappy
accident.That is,thatfact justseemstostand inthewrongsortofrelationtothefactthat I
believedthatitwasraining,i.e.itstandsinapurelycausalexplanatoryrelationtothatfact.11Assume,ifyoulike,thatIknowthatIamnotpronetohallucinations,andalsodon’tbelievethatthereisanythingthatmightmakebehallucinatenow.12Thanks for thisexampleareowedto JohnRoberts,whoput it tome inaseminar.This iswhatonemightcalla‘deviant’Gettiercase,involvingbothajustified,truebeliefthatfallsshortofknowledgeanda deviant causal chain linking the fact itself with the belief. This sort of case is a well-establishedproblem for Goldman’s (1967) causal theory of knowledge (see e.g. McDonnell 2015). I will discussdeviantcausalchainsingreaterdetailinthenextchapter.
185
5.1.2 Thenaturalstrategyisunsuccessful
Nowwecanseewhythenaturalstrategyisunsuccessful:intheexamplejustconsidered,the
would-beexplanans is the right sortof thing– it isanormative reason, indeed, it isactually
whatIbelieve–andyettheexplanatoryrelationsinvolvedstillaren’ttransitive.So,asawayof
distinguishingbetweenthedifferentcases,thenaturalstrategyfails.That is, it’snotbecause
ofthenatureofthewould-beexplanansthattransitivityofexplanationfails.
The next section sketches out my strategy, which will be developed in detail in the next
chapter.
6 Themysteriousstrategy
Here is thechallenge: Ineedaprincipledaccountofwhy it is that theexplanatory relations
fromthebelievedfacttotherationalityoftheactionaretransitiveinthe‘award’casebutnot
ineitherthe‘Sally’caseorthe‘carbonmonoxide’case.Whatisthataccount?
I’vehintedatmyansweralready,andMcDowellexpressestheintuitionaboutwhatmarksout
the ‘award’ case from the others nicely thus: ‘we can say that the fact itself is exerting a
rational influenceontheagent’swill;wecansay that…theagent is respondingrationally to
thefactitself.’(2013,17)Now,in§(X)4.2.1,Ilamentedtheinscrutablymetaphoricalcharacter
ofthesesortsofremarks,so,whilsttheywilldoasanexpressionoftheintuition,Iwouldbe
fallingwellshortofmyownstandardifIleftitthere.
InthechaptersthatfollowIwanttogiveanon-metaphoricalcharacterisationofwhatitisthat
distinguishesthe‘award’casefromtheothers.Inparticular,Iwillargue,thereisamysterious,
transitive,non-causalexplanatoryrelationbetweenthefactthatmyfriendwonanawardand
thefactthatIbelievedthatshedidwhichislackingintheothercases.
Inthesubsequentchapter, Iwillargue,thismysteryrelationistransitivewiththenon-causal
explanatoryrelationthatobtainsbetweenthefactthatIbelievedthatmyfriendhadwonan
awardandthefactthatitwasprotantorationaltocongratulateher.
This argument provides a principled reason for saying why the explanatory relations are
transitive in the ‘award’ casebutnot in theothers:because, in the ‘award’ case there is an
explanatoryrelationbetweentheworldandmybeliefthatislackingintheothercases.
186
(XIV)
TheMysteryRelation
In which I introduce the mystery relation. I suggest that a mysterious, non-causalrelationobtainsbetweenabeliefandthejustificationthatitisbasedonwhen that belief is justified. I argue that the mystery relation must be non-causal,because,asdeviantcausalchainsdemonstrate,amerelycausalrelationbetweenabeliefandsomejustificationfor it isnotsufficientforthatbelieftobejustified.Isuggestthatthisexactsamemysteriousrelationrelates:thebeliefthatptothefactthatpwhenthebeliefthatpisknowledgeable;ajustificationfor the belief that p to the fact that p when that justification affords theopportunityforknowledge;andanactiontosomebeliefthatexplainswhyitisrationalwhen thataction isdone intentionally. Iargue, furthermore, that thismysteryrelationisatransitive,explanatoryrelation.
Hereare somequestions:whatdistinguishesa justifiedbelief fromamerely justifiableone?
Whatseparatesaknowledgeablebelieffromabeliefthatoneholds,whenoneisinaposition
toknowit,withoutknowingit?Whatisthedifferencebetweenajustificationthataffordsthe
opportunity for knowledge from one that doesn’t? And, lastly,what distinguishes an action
doneintentionallyfromamerebodilymovement?
Idonotclaimtoknowtheanswertothesequestions.However,inwhatfollowsIwanttosee
what can be said without offering a theory of the difference between these cases. For
instance, the difference between a justified belief and a merely justifiable one is already
characterised in terms of the epistemic basing relation – the question is how that relation
shouldbeunderstood.Butevenwithoutansweringthatquestion,wecanstillsaysomethings
aboutthebasingrelation.Inparticular,itiswidelyagreedthattheproblemofdeviantcausal
chainsfrustratesapurelycausalanalysisofthebasingrelation,so,Isuggest,wecansuppose
that the basing relation is not merely causal. Which is to say, I suggest, that there is a
mysteriousnon-causalrelationbetweenabeliefandthatwhichitisbasedon,whenthatbelief
isjustified.
Isuggestthatthisexactsamemysteriousrelationrelates: thebeliefthatp tothefactthatp
whenthebeliefthatp isknowledgeable;ajustificationforthebeliefthatptothefactthatp
when that justification affords theopportunity for knowledge; and anaction to somebelief
when that action is done intentionally. I argue, furthermore, that thismystery relation is a
transitive,explanatoryrelation.
187
Thischaracterisationofthemysteryrelationprovidesthebasisforthediscussionofthenext
chapter, inwhichIsaythatthemysteryrelationistransitivewiththenon-causalexplanatory
relation thatobtainsbetweenthe fact that Ibelievedthatmy friendhadwonanawardand
thefactthatitwasprotantorationaltocongratulateher.
Iofferthefollowinganalysismodestly.ThereisnosenseinwhichItakemyselftohavesolved
theproblemof deviant causal chains for the contexts considered. I alsodonot think I have
offeredmuchofananalysisofwhatdifferentiatesdeviantcasesfromnon-deviantcasesother
thansomethingdifferentiatesthem,andthatwhateveritmustbemustbeinsomesensenon-
causal,explanatoryandtransitive.Itaketheseobservationstoberelativelyanodyne,although
somemaydisagreewiththem.Myhopeisthateventheseblandobservationswillsufficefor
thepurposesofthenextchapter.
1 Themysteryrelationandjustifiedbelief
Thereisacommonlyrecogniseddistinctionbetweenajustifiedbeliefandamerelyjustifiable
belief.Hereisatypicalexample:
Ajustifiablebeliefisonethebelievercouldbecomejustifiedinbelievingifhejustputtogetherin the right way what he already believes. To illustrate, a woman might have adequateevidence for believing that her husband is unfaithful to her, but systematically ignore thatevidence.However,whenherfather,whomsheknowstobetotallyunreliableinsuchmattersandbiasedagainstherhusband,tellsherthatherhusbandisunfaithfultoher,shebelievesiton that basis. Then her belief that her husband is unfaithful is unjustified but justifiable.(PollockandCruz1999,79)
The woman in this example has some justification for believing that her spouse has been
unfaithful(wedon’tknowwhatit is),whichsheignores.Inspiteofignoringthejustifications
she has for believing it, she still ends up forming the belief that her husband has been
unfaithful,butbases iton the fact thather father toldher thatherhusbandwasunfaithful,
which is not a justification for believing it (because her father is known to be biased and
unreliable).
Herbeliefismerelyjustifiable,andnotjustified,becauseajustifiablebeliefisabeliefforwhich
one has some justification (which she does), but a justified belief is one that isbased on a
justificationonehasforit(whichhersisnot).
Asanaside:Iamstrayingintoepistemologyhere.Torestraintheboundsofmyassertions,let
mestatemyassumptionsplainly:all that ismeantbya ‘justification’here is something that
couldexplainwhyitisjustifiableforonetobelievethatpanduponwhichone’sbeliefthatp
couldbebased.WhatIamcalling‘justification’ismoretypicallycalleda‘reasonforbelief’in
the literature,but Iavoidthe‘reasons’terminologytoavoidconflatingthatwiththepresent
188
discussion.1Moreover, I will take it as a given that appearances could be justifications for
belief.2
1.1 Theepistemicbasingrelation
Whatdoesitmeanforabelieftobebasedonsomejustificationforit?Thedefactoanalysisof
the‘basing’relationisacausalone.However,itiswidelyacknowledged(e.g.Korcz2015)that
thepossibilityofdeviantcausal chainsbetweenabeliefanda justification for it frustratesa
purelycausalanalysisofbasingbydemonstratingthatthefactthatajustificationforabelief
standsinacausalrelationtoitisnotenoughtoensurethatthebeliefisjustified.Typicallyone
hastoqualifyacausalanalysisbysayingthatthejustificationmustcausethebelief‘intheright
way’–butapurelycausalanalysisofwhatthis‘rightway’is,islacking.
Re-purposingtheexampleabove:supposethatEvaseesherhusbandkissinganotherwoman.
Supposethatthisisadequateevidenceofherhusband’sinfidelity(supplantmorecompelling
evidence ifyouaren’tconvinced).She ignorestheevidenceandcarriesonbelievingthather
husband is faithful.Her father isoutof thepicture this time,but suppose, instead, that the
stressofignoringwhatshehasseen(itisadifficultthingtoignore)causesabrainaneurism(in
spite of the fact that she does manage to ignore what she has seen) that, by incredible
coincidence,causeshertobelievethatherhusbandhasbeenunfaithfultoher.3Believingthat
herhusbandhasbeenunfaithfultoher,shesuesfordivorce.4
In this case her belief that her husband has been unfaithful is justifiable, and it is caused
(albeit,inaroundaboutway)bythejustificationshehasforbelievingit.Butitseemswrongto
sayeither thatherbeliefwasbasedon the justification thatshehad,or, indeed, that itwas
reallyajustifiedbelief.5
So, even if, as is popularly thought, ‘the basing relation is at least partly a causal relation,’
(PollockandCruz1999,79)theneedtostipulatenon-deviancyofthecausalchainprovidesat
1AlthoughIthinkananalogoustreatmentofreasonsforbeliefispossible,itisbeyondthescopeofthisdiscussiontoprovideone.2This will mean excluding the more extreme forms of internalism about epistemic justification (e.g.Davidson 2001d) but, adherents to that view are in the minority, as Littlejohn notes: ‘most statistinternalistsdefendtheviewthatexperiencesconstituteourreasonsforbelief.’(Forthcomming,4)3We could stress the independence of these events from Eva’s point of view: she has ignored theevidence,sowhenaskedwhyshebelievesthatherhusbandwascheatingonhershewon’tevencitethatfactthatshe’sseenhimkissinganotherwoman–perhapssheactuallymanagedtoforgetit.Whoknowswhatshewouldsay,theimportantpoint isthatshewouldn’tsaythatshesawhimkissingthatwoman.4Supposewhateveryouneedtosuppose inorder toensure that itwas rational forher todoso (thearationalityofthewaysheacquiredherbeliefnotwithstanding).5Thisexampleisstrongerthan,for instance,Plantinga’s(1993,69fn.8)classicexamplebecausewhatdoesthecausingisalsoajustificationforthebelief,whereasinPlantinga’scaseitisnot.
189
leastprimafacieevidencethatthattherelationisnotmerelycasual.Moreover,intheabsence
ofacompelling,purelycausal,solutiontotheproblemofdeviantcausalchains,6Iwillassume
thatthereisnosuchsolution,andthatthebasingrelation,whateveritmaybe,isnotmerely
causal.7
Thus, ifabelief isbasedonsome justification for it (i.e. if it is justified) thenthere isanon-
causal relation (as well as, perhaps, a causal relation) between the belief and the
justification(s)itisbasedon–let’scallitthemysteryrelation.
At thispointweshoulddosomeontologicalhousekeeping.The relataof thebasing relation
areprobablynotfacts–theyare,perhaps,beliefs,experiencesoreventsorwhathaveyou.In
thesameveinIthinkthattherelataofthemysteryrelationproperarealsoprobablynotfacts.
However,itwillgreatlysimplifymyformalexposition,at,Ithink,nocosttomyargument,ifwe
treatthemasrelationsbetweenfacts.8
(M1) For any proposition,p, ifA has a justified belief thatp then the fact thatAbelievesthatpismysteriouslyrelatedtosomejustificationforit.
2 Themysteryrelationandknowledge
NowIwanttoconvinceyouthatthesamemysteriousrelationobtainsbetweenthefactthat
p9and the fact thatAbelieves thatpwhenanagent knows thatp.Why should youbelieve
this?
Firstconsiderthat,asinthecaseofthebasingrelation,thepossibilityofdeviantcausalchains
frustratesattemptstogiveapurelycausalanalysisoftherelationshipbetweenthefactthatp
andanagent’sbelief thatpwhen theyknowthatp. Inorder tomaintaina causal theoryof
knowledge,onemustinsistthatthefactcausesthebeliefintherightway.
6McCain (2012) offers a purely causal solution by, as Korcz (2015) puts it, ‘removing the chain’ anddenyingthatthebasingrelationistransitive.However, IshareKorcz’sconcernthatthistheoryfailstocapture pre-theoretical accounts ofwhat our beliefs are based on, so I don’t find it very compelling.Moregenerally,totheextentthatonetakescausalrelationstobetransitive,apurelycausalanalysisofdeviantcausalchainswillbeimpossible,assumingthatthesamecausalrelationobtainsbetweeneachlinkinthechain.7RecentsolutionstotheproblemofdeviantcausalchainsbyHyman(2015)andSosa(2015)supportmycontentionthatrelationsaffectedbythemarenotmerelycausal.Theyarguethatcausalrelationsarestill necessary for such relations, but that non-deviancy is only guaranteed by a further non-causalrelation:‘themanifestationofacompetence’.8Thereneedbenothingsignificantaboutthismove:iftherelataofgenuinebasingrelationsormysteryrelations are the truth-makers of propositions (e.g. Sally’s believing that a bear is chasing her to thepropositionthatshebelieves it), thenthebasingrelationsandmysteryrelations Imentionaremerelythecounterpartsofthegenuinerelationsinanontologyoffacts.9Again,someontologicalhousekeeping:properlyspeakingitisprobablythetruth-makerofthefactthatp,thatisrelatedtothebelief,butIamagaintransposingthistalktoanontologyoffactsforthesakeofsimplicity.
190
Let’s revisit Eva’s case: Eva is in a position to know that her husband has been unfaithful10
becausesheiswatchinghimkissanotherwoman.Andhisdoingthatiswhatcauseshertosee
himdoingitwhichiswhatcauseshertoputtheeffortintoignoringitwhichiswhatcausesher
tohaveabrainaneurismwhichiswhatcauseshertobelievethathe’sbeenunfaithfultoher.
In this convolutedway, the fact that her husbandhasbeenunfaithful causesher tobelieve
that he has been unfaithful at a time when she is in a position to know that he has: she
believesitwithoutknowingit,eventhoughsheisinapositiontoknowitandeventhoughthe
factthathehadbeenunfaithfuliswhatcausedhertobelievethathehad.
Thus,giventhatacausalrelationbetweenthebeliefthatpandthefactthatpisnotsufficient
forthatbelieftobeknowledge(indeed,eveniftheagentisinapositiontoknowthatp)and
giventhatthe‘therightway’ofsomefact’scausingabeliefinitcannotbeanalysedinpurely
causal terms11we can say that if an agent knows thatp then there is a non-causal relation
betweentheirbeliefthatpandthefactthatp.
So,even if a causal relationbetween the fact thatp and thebelief thatp is required foran
agenttoknowthatp(anotinconsiderable‘if’),anon-causalrelationbetweenthefactandthe
beliefisstillalsorequiredfortheagenttoknowthatp.
Butwhysupposethatthisnon-causalrelationbetweenthefactandthebeliefisthesameas
the mystery relation that basing relations entail? Because, I suggest, the cases are directly
analogous:thedistinctionbetweenajustifiedbeliefandamerely justifiableoneisanalogous
to the distinction between knowing that p andmerely believing that p12when one is in a
positiontoknowthatp.
Indeed, theanalogousnessbetween the cases is already recognised in thediagnosisofboth
thesecasesascasesofdeviantcausalchains.Thesimplestexplanationofthisanalogyisthat
thenon-causalrelationthatislackinginthemerelycausalcasesineachisofacommonkind,
whichIhavecalledthe‘mysteryrelation’.Thus:
(M2) Foranyproposition,p,ifAknowsthatpthenthefactthatAbelievesthatpismysteriouslyrelatedtothefactthatp.
3 Themysteryrelationandopportunitiestoknow
NowIwanttoconvinceyouthat thesamemysteryrelationdistinguishesthose justifications
thataffordtheopportunityforknowledgefromthosethatdonot.
10Surely!11Whichweassumegiventhatsuchananalysisseemsisbothunavailableandanywayimplausible.12I.e.Believingwithoutknowingthatp
191
We said that Evawas in aposition to know thatherhusbandwas cheatingonher. Sally, in
contrast, isnot inapositiontoknowthatabear ischasingher(not leastbecauseone isn’t).
Likewise(recallingtheexamplefromthepreviouschapter),whenIaminaroomthatisslowly
fillingwith carbonmonoxide, I amnot in aposition to know that the room is full of carbon
monoxide (therearenoalarms,novisiblewarningsetc.).WhatdifferentiatesEva fromSally
andIwithrespecttoourepistemicpositions?
Hereisawaytocharacterisethedifference:whileweallhaveajustificationforbelievingthat
p,onlyEvahasa justification thataffords theopportunity forknowledge.Characterisingour
question in these terms, and generalising it beyond these cases, we can ask: what
differentiates a justification for believing thatp that affords the opportunity for knowledge
from one that doesn’t? I think it is the mysterious relation that a knowledge-affording
justification stands in to the fact that p that separates it from a justification that does not
affordtheopportunityforknowledge.
Tostart,considertheseremarksbyMcDowell,inhisdiscussionofhowperceptualknowledge
ispossible(giventhepossibilityofillusion):
Supposesomeone ispresentedwithanappearance that it is raining. It seemsunproblematicthatifhisexperienceisinasuitablewaytheupshotofthefactthatitisraining,thenthefactitself can make it the case that he knows that it is raining. (McDowell 1982, 474 emphasisadded)
What is this suitable way? What is the relation between the ‘fact itself’ and the man’s
experiencethatmakesitpossibleforthefactto‘makeitthecasethattheheknowsthatitis
raining’?Well,itisn’tmerelycausal.
Suppose that another version ofme, call him ‘Twinny’, on some other very similarworld is
likewise in a room that is slowly filling with carbon monoxide. Twinny doesn’t have the
hallucination.However,hisfriendseesthecarbonmonoxidelevelsoftheroomonamonitor
(whichTwinnyhadnoaccessto),and,accordingly,burstsintotheroomtowarnhim.Twinny
leavestheroom,justasIdid.
Now,what Twinny and I experience is subjectively indistinguishable (ex hypothesi), and our
perceptual experiences justify each of our beliefs (we both know that we aren’t prone to
hallucinationsetc.).However,onlyTwinnyhadanexperiencethatcanaffordtheopportunity
forknowledge.13AndthatistheupshotofthefactthatthejustificationthatIhadforbelieving
thattheroomwasfullofcarbonmonoxidewascausedbythefactthattheroomwasfullof13OfcourseIcoulddothetests,gathertheevidenceandsoforth(assumingIlivedthatlong)andthenIwould be in a position to know that the roomwas full of carbonmonoxide – but there and then, incircumstancesastheywereoriginallydescribed,Iwouldnotbeinapositiontoknow.
192
carbonmonoxidebya ‘deviant’causalchain–whereasthechain inTwinny’scasewas ‘non-
deviant’;thatis,itisonlyinTwinny’scasethathisexperienceisinasuitablewaytheupshotof
thefact.
Generalising:wecansaythatthefactthatp’shavingcausedajustificationforthebeliefthatp
doesnotensurethatthatjustificationisonethataffordsanopportunityforknowledge.So,I
suggest, theremustbeanon-causalrelation(aswellas,perhaps,acausalrelation)between
the fact that p and a justification for believing that p if that justification is to afford an
opportunityforknowingthatp.
What is that non-causal relation? My answer is presumably clear: it is the samemystery
relationasrelatesjustifiedbeliefstothefactstheyarebasedon,andknowledgetothatwhich
isknown.Inotherwords:
(M3) Foranypropositions,jandp,ifjaffordstheopportunityforknowledgethatpthenjismysteriouslyrelatedtothefactthatp.
4 Themysteryrelationandactingforareason
Finally,Iwanttoconvinceyouthatthesamemysteryrelationdistinguishesactionsdonefora
reasonfrommerebodilymovements.
Considerthefollowingexample:
A climbermightwant to rid himself of theweight and danger of holding anotherman on arope, andhemight know that by looseninghis holdon the ropehe could rid himself of theweightanddanger.Thisbeliefandwantmightsounnervehimas tocausehimto loosenhishold, and yet itmight be the case that he never chose to loosen his hold, nor did he do itintentionally.(Davidson2001b,79)
Inthisexamplethefactthattheclimberbelievesthat looseninghisgripwouldridhimselfof
danger (partially)explainswhy it ispro tanto rational for theclimber to loosenhisgrip,and
that belief also causes him to loosen his grip. However, he does not loosen his grip
intentionally. This example thus (famously) createsaproblem for apurely causal analysisof
what it is to act intentionally: although there is a causal chain between the belief and the
action,theagentdoesnotactintentionallybecausethecausalchainis‘deviant’.
Suppose that, had the climber acted on his beliefs and desires in the ‘non-deviant’way, he
wouldhaveactedintentionally.WecanthensaythatDavidson’sclimberwas inapositionto
dowhathedidintentionally,eventhoughhedidn’t.14Sowhatdifferentiatessomeonewhoφs
14Perhapsbeinginapositiontodosomethingintentionallyisjusthavinganintentiontodoit.Ileaveitforthereadertodecide.
193
intentionally from someone (like the climber) who is in a position toφ intentionally, and,
indeed,φsbutdoesnotdosointentionally?
Thetraditional‘answer’tothisquestionistosaythatifanagentactsintentionallythen,inter
alia, a belief that explainswhy their action is rationalmust cause them todo it in the right
way.15So, again, assuming that this elusive ‘right way’ cannot be analysed in purely causal
terms,wecansaythatifanagentactsintentionallythenthereisanon-causalrelation(aswell
as, perhaps, a causal relation) between features of their psychology that explain why their
actionisrationalandtheiraction.16
Andagain,owingtotheanalogousnessofthiscasetotheothersalreadyconsidered,Isuggest
that thisnon-causal relation is the very samemystery relation thatwas requiredof justified
belief,ofknowledgeandofopportunityforknowledge-affordingjustifications.Thus:
(M4) IfAφsintentionallythen,forsomeproposition,p,thefactthatAbelievedthatpexplainswhyitwasprotantorationalforAtoφandismysteriouslyrelatedtothefactthatAφ’d.17
5 Asummaryoftheexamples
Ihavesuggested thatwhatdistinguishesdeviantcausal chains fromnon-deviantones in the
contexts considered is that, in the non-deviant cases, the relata are not merely causally
related.Inparticular,Ihaveargued,thesamenon-causal‘mystery’relationispresentineach
case,sothat:
(M1) For any proposition,p, ifA has a justified belief thatp then the fact thatAbelievesthatpismysteriouslyrelatedtosomejustificationforit.
(M2) Foranyproposition,p,ifAknowsthatpthenthefactthatAbelievesthatpismysteriouslyrelatedtothefactthatp.
(M3) Foranypropositions,jandp,ifjaffordstheopportunityforknowledgethatpthenjismysteriouslyrelatedtothefactthatp.
(M4) IfAφsintentionallythen,forsomeproposition,p,thefactthatAbelievedthatpexplainswhyitwasprotantorationalforAtoφandismysteriouslyrelatedtothefactthatAφ’d.
Asanaside:weshouldnotethatwhilemysteryrelationsarenon-causalrelations,theydonot
necessarily exclude causal relations. That is, for instance, it is quite possible that a justified
15Cf. ‘Anaction isperformedwitha certain intention if it is caused in the rightwaybyattitudesandbeliefsthatrationalizeit.’(Davidson2001c,87)16Notethatthisisanecessarybutnotsufficientcondition.17Thisformulationexcludes,forconvenience,instanceswhereanagentdoessomethingforitsownsake(e.g.IsangbecauseIfeltlikesinging).InadoptingthisformulationIamnotclaimingthatsuchactsarenotdoneintentionally,itisjustmoreconvenienttousethisformulation.
194
beliefmaybebothcausallyandmysteriouslyrelatedtothejustificationitisbasedon.Indeed,
dependingonyourviews intheseareas itmightbethatacausalrelation isanecessary (but
notsufficient)conditionforthepresenceofamysteryrelation.
In the following sections I will argue further that the mystery relation is a transitive,
explanatoryrelation.
6 Mysteryrelationsareexplanatoryrelations
Now I want to convince you thatmystery relations are explanatory relations. Theymay be
other thingsalso,but Iaimtoconvinceyou that theyaredefinitelyexplanatory. Iwill argue
thatineachofthecasesofthemysteryrelationabovethereisanon-causalexplanationofthe
explanandumthatislackingwhenthemysteryrelationisabsent.
6.1 Explainingjustifiedbeliefs
Itisgenerallyacknowledgedthatthejustificationonwhichanagent’sbeliefisbasedexplains
whytheybelievedit(e.g.Harman1970).Whatsortofexplanationdoesitprovideuswith?
If we accept that the basing relation is partly a causal relation then, even though a causal
analysis is insufficient forestablishing it, it’s still possible that theexplanatory importof the
basing relation is merely causal. So, is the way that a justified belief is explained by the
justificationonwhichitisbasedmerelycausal?
The fact that Sally heard a bear-like sound headed her way (in a forest that she knew to
contain bears) is a justification for her to believe that a bear is chasing her.Moreover, her
beliefthatabearischasingherisbasedonthatjustificationforit.Sowecansay,aswehave
noted,thatSallybelievesthatabearischasingherpartlybecausesheheardabear-likesound.
Thatis:ajustificationforSallytobelievesomethingpartially18explainswhyshebelievedit.
Now notice that for Eva a justification for her to believe something also explains why she
believed it: shebelieves thatherhusbandhasbeenunfaithfulbecause itappearedtoheras
thoughhewaskissinganotherwoman(which,indeed,hewas).Thatis,forbothEvaandSallya
justification for their belief explains their belief. However, the sense of the explanation
providedby the justification inSally’s case seems tobe importantlydifferent to the sense it
providesinEva’scase–andthatdifferencecannotbecharacterisedincausalterms(becausea
causalrelationobtainsinbothcases).
18Otherpartsofthefullexplanationofherbeliefinclude,forinstance,thefactthatsheknewthewoodtocontainbears.
195
So, there is a seemingly non-causal explanationof Sally’s belief that is lacking in Eva’s case,
which is to say that theexplanatory importof thebasing relation isnotmerelycausal (note
thatEva’sbeliefisnotbasedonanything,sothereisnosimilarnon-causalexplanationofit).
6.2 Explainingknowledgeablebeliefs
Anotherexample:supposethatEva’shusbandiskissingSean’swife.Seanalsoseesithappen,
butstraightforwardlyconcludesthathiswifehasbeenunfaithfultohim.Seanknowsthathis
wifehasbeenunfaithfulandhebelievesthathiswifehasbeenunfaithfulpartlybecauseshe
hasbeenunfaithful.
Now,it isalsotrueofEvathatshebelievesthatherhusbandhasbeenunfaithfulbecausehe
hasbeenunfaithful,eventhoughshedoesn’tknowthathehas–but,again likethe justified
belief case, the sense of the ‘because’ seems different. Even given that there is a causal
explanationinSean’scase,thefactthathiswifeischeatingonhimpartiallyexplainsthefact
thathebelievessheisinawaythatisnotmerelycausal,becauseitexplainsitinawaythatis
qualitativelydifferenttothemerelycausalexplanation(i.e.theEvacase).
6.3 Mysteriousrelationsarepartialexplanatoryrelations
Iwillsparethereaderarehearsalofthisreasoningfortheothercasesconsideredandcutto
the chase: wherever we compare a deviant causal case with a non-deviant case (in the
contextsconsidered),itseemsasthoughthereisanon-causalexplanatoryrelationinthenon-
deviantcasethat is lacking inthedeviantone. Indeed, itseemstomethat it ispreciselythe
differentcharacteroftheexplanatoryrelationsinvolvedinthenon-deviantcasethatallowsus,
inthesecontexts,todistinguishthenon-deviantexamplesfromthedeviantones.
So,sincethereappearstobeanon-causalexplanatoryrelationwhereverwehaveamystery
relation, and sincemystery relations, aswe have established, are non-causal, I suggest that
mysteryrelationsarenon-causal(partial)explanatoryrelations.
Thus,whatseparatesEvafromSeanisthefactthatthejustificationSeanhasforhisbeliefnon-
causally (aswellas,perhaps,causally)explainshisbelief,whereas, forEva itmerelycausally
explainsit.
Likewise,whatseparatesmeinmycarbonmonoxidefilledroomfromTwinnyisthatthefact
that the roomwas filled with carbonmonoxide non-causally (as well as, perhaps, causally)
explainswhy it appeared toTwinnyas thoughhis friendwaswarninghimabout the carbon
monoxide,whereasitonlycausallyexplainswhyitappearedtomethatway.Andsoon.
196
7 Mysteryrelationsaretransitive
Finally, I want to convince you that mystery relations are transitive. I will show that the
conditionsfortransitivityaresatisfiedinalloftheaboveexamples,andIwillarguefurtherthat
the transitivity of the explanatory relations involved is the best explanation of why
knowledgeablebeliefsaremysteriouslyrelatedtothebelievedfacts.
7.1 Itistransitiveintheexamples
WhenIcongratulatedmyfriendIknewthatshehadwonanaward,sothefactthatIbelieved
thatshehadwonanawardwasmysteriouslyrelatedtothefactthatshehadwonanaward.
Moreover,Isuggest,sinceIcongratulatedherforareason,thefactthatIcongratulatedheris
mysteriouslyrelatedtothefactthatIbelievedthatshehadwonanaward.Canweconclude
that the fact that I congratulated her is mysteriously related to the fact that she won an
award?Ithinkwecan.
Consider:whenIsaythatIcongratulatedmyfriendbecauseshewonanawardithastheright
sortofexplanatorycharacter,thefactexplainsmyactionintherightway.Likewise,whenIsay
thatTwinnylefttheroombecauseitwasfullofcarbonmonoxidethattoohastherightsortof
explanatorycharacter.Andthat is, I suggestbecausethe fact that theroom is fullofcarbon
monoxide ismysteriously related to his belief, which, in turn, ismysteriously related to his
action.Thatis,Isuggestthatwecanconcludethatourrespectivenormativereasonstoactare
mysteriously related to our respective actions because there is a chain ofmystery relations
connectingthenormativereasontotheactionandthemysteryrelationistransitive.
Wecanseethatthesecaseshavetherightsortofexplanatorycharacterbycomparingthem
withacaseinwhichthenormativereasonfailstobemysteriouslyrelatedtomyaction:mein
myroomfullofcarbonmonoxide.InthiscasewhenwesaythatIlefttheroombecauseitwas
fullofcarbonmonoxidethatdoesnothavetherightsortofexplanatorycharacter.Ofcourse
thereisperhapsasenseinwhichitistrue(seeminglyastrictlycausalsense),butthatisnotthe
senseinwhichtheexpressionwouldbeconventionallyunderstood.Whatmarksoutthesense
in which the expression would be conventionally understood from this one is, as I have
suggested,thepresenceofthisnon-causalexplanatoryrelation;themysteryrelation.
Whyisthenormativereasonnotrelatedtomyactioninthecarbonmonoxidecasebutitisto
Twinny’saction?Because, I suggest, there isnota chainofmystery relationsconnectingmy
actiontothefactthattheroomisfullofcarbonmonoxide,whilethereisachainofmystery
relationsconnectingTwinny’sactiontothatfact.
197
Moreover,whileIwillsparethereaderademonstration,Isuggestthatthesamereasoningcan
beappliedequallytoalltheothercases.
Forclaritywecansummarisethisasfollows.Firstly,somenotation:foreachcase,let‘f’stand
forthebelievedproposition;‘j’standforthejustificationthattheagenthasforbelievingit;‘b’
standforthefactthattheybelievedit;and‘a’standforwhattheagentdid.TableXIV-1sets
outthereferentsofthesesymbolsforeachcase.
Example f j b a
Awardcase Myfriendwonanaward
Ireadthatshehadwonanaward
Ibelievedthatshehadwonanaward
Icongratulatedher
CO19case Theroomwasfullofcarbonmonoxide
Itappearedtomeasthoughmyfriendwaswarningme
Ibelievedthattheroomwasfullofcarbonmonoxide
Ilefttheroom
Evacase Herhusbandwasunfaithfultoher
Itappearedtoherasthoughherhusbandwaskissinganother
woman
Shebelievedthatherhusbandwasunfaithful
Shesuedfordivorce
Climbercase Looseninghisgripwouldridhimof
danger
Itappearedtohimasthoughifheletgohewouldbefreedfromdanger20
Hebelievedthatlooseninghisgripwouldridhimof
danger
Heloosenedhisgrip
TableXIV-1:Thecomponentfactsineachexample
My suggestion is this: in any circumstance in which a chain of transitive mystery relations
would imply that the ends of the chain are mysteriously related, running through the
reasoningjustsetoutforeachoftheexamplesfindsthattheendsofthechainare,indeedso
related.Letting‘⇝’standforthemysteryrelation,TableXIV-2setsthisout.
ExamplesObservedrelations Shouldtheseobtain: Dotheyobtain
whentheyshould?f⇝j j⇝b b⇝a f⇝b? j⇝a? f⇝a?
Awardcase ü ü ü ü ü ü ü
COcase û ü ü û ü û ü
Evacase ü û ü û û û ü
Climbercase ü ü û ü ü û ü
TableXIV-2:Thetransitivityofthemysteryrelation
19CO=carbonmonoxide.20Youcouldsupplantsomethingmoreinterestinghereifyouliked–thisisjustmeanttobeindicative.
198
Ofcourse,thefactthatthemysteryrelationhappenstobetransitiveintheseexamplesdoes
notamounttoproofthat it isalwaystransitive,but isat leastevidenceforthatclaim. Inthe
nextsectionIprovideadifferentdefenceoftheclaimthatmysteryrelationsaretransitive.
7.2 Knowledgeandknowledgeaffordingjustification
My second argument for the transitivity of mystery relations is that it provides the best
accountofwhyitisthatknowledgeablebeliefsaremysteriouslyrelatedtothebelievedfacts.
To start with, consider that, to the extent that knowledge entails justified belief, 21 a
knowledgeablebeliefmustbebasedonajustificationforit.Thatbeingso,itstrikesmethatif
a belief is knowledgeable it must be based on (and therefore mysteriously related to) a
justificationthatactuallyaffordstheopportunityforknowledge.22Ifthesetwoclaimsaretrue,
then,foranyp,ifAknowsthatp,thefactthatAbelievesthatpisbasedonsomejustification
thataffordstheopportunityforAtoknowthatp.
Summarising: for any p, a knowledgeable belief that p, being based on a justification that
affords the opportunity for knowledge that p, must therefore be mysteriously related to a
justificationforthebeliefthatpthatis,inturn,mysteriouslyrelatedtothefactthatp.Thatis,
touse thenotationof theprevious section, if theagentknows that f then: f ismysteriously
relatedtoj,jismysteriouslyrelatedtob,andfismysteriouslyrelatedtob.
So,ifanagentknowsthatpthentransitivityistrueofthemysteryrelationsbetweenthebelief
thatp,ajustificationonwhichitisbasedandthefactthatp.
21Which,pickingmybattles,Iwilltakeasagiven.22Lehrer’sgypsylawyercaseisacounterexampletothisclaim(oritwouldbeifitweretrue):
Alawyerisdefendingamanaccusedofcommittingeighthideousmurders…Thereisconclusiveevidence that the lawyer's client is guilty of the first sevenmurders. Everyone, including thelawyer,isconvincedthatthemaninquestionhascommittedalleightcrimes,thoughthemanhimself says he is innocent of all. However, the lawyer is a gypsywith absolute faith in thecards.Oneeveningheconsultsthecardsabouthiscase,andthecardstellhimthathisclientisinnocent of the eighth murder. He checks again, and the cards give the same answer. Hebecomesconvincedthathisclientisinnocentofoneoftheeightmurders.Asaresulthestudiesthe evidence with a different perspective as well as greater care, and he finds a verycomplicatedthoughcompletelyvalidlineofreasoningfromtheevidencetotheconclusionthathis client did not commit the eighth murder… This reasoning gives the lawyer knowledge.Thoughthereasoningdoesnotincreasehisconviction–hewasalreadycompletelyconvincedbythecards–itdoesgivehimknowledge.(Lehrer1971,311–12)
I shareGoldman’s intuition that, ‘To the extent that I clearly imagine that the lawyer fixes his beliefsolelyasaresultofthecards,itseemsintuitivelywrongtosaythatheknows—orhasajustifiedbelief—that his client is innocent.’ (2012, 36 n8) The lawyer is in a position to know it, and depending onwhetherornotonethinksthetarotcardscountasajustification,onemightevensayhehasajustifiedbelief–butIfinditstrangetosaythatheknowsitifthepresenceoftheonlyjustificationthataffordstheopportunityforknowledgeisneitherherenortherewithrespecttohisactuallyhavingthebelief.
199
Dowehavereasontothinkthatthemysteryrelation is transitivewherever itappears?Well
wedon’thaveanyreasontodoubtit,buthereisanotherreasontothinkthatitmightbe:the
transitivity of the mystery explanation would explainwhy the belief that p is mysteriously
relatedtothefactthatpwhentheagentknowsthatp.
Althoughour intuition seems to suggest that a belief thatp is non-causally (if also causally)
explainedbythefactthatpifanagentknowsit,it’snotcleartomehowthefactthatpcould
directlyexplainanagent’sbeliefthatpinmuchthesamewayasitiswasnotcleartomehow
the fact thatp coulddirectly explainanagent’saction.23However,given thatourperceptual
experiences intermediate the explanatory relations between the world and beliefs the
transitivity of the relation that links them would ensure the connection that our intuitions
suggest between the belief and the fact itself. So, I suggest, the best explanation of why
knowledgeablebeliefsaremysteriouslyrelatedtothebelievedfactsisthatthereisatransitive
chainofmysteriousrelationsthatlinksthebelieftothefactsviathejustifications.24
7.3 Onwhomistheburdenofproof?
There areperhapsnumerouspoints in these arguments towhichone couldobject – but all
theywoulddoisunderminetheargumentsfortheclaimthatmysteryrelationsaretransitive,
theywouldnotunderminetheclaimitself.Whichleadsmetomyfinalpoint,indefenceofthe
assumption thatmystery relations are transitive: since I do not take it to be clear that the
burdenofproofissolelyonmetoprovethattheyare,itwouldneedafurtherargumentstillto
demonstrate that theyaren’t. Inparticular, sincemyview is thatourworkingassumptionof
anyparticularsortofexplanatoryrelationshouldbethatitistransitiveuntilprovenotherwise,
I take it that the burden of proof is actually upon thosewhowould deny that themystery
relationistransitive.
23Seemycriticismofdirecttheoriesofnormativereasonexplanationin§(X).24It’s worth noting that McDowell would reject this account. He notes that in the knowledge case‘appearancesareno longerconceivedas ingeneral interveningbetweentheexperiencingsubjectandtheworld.’ (1982,472)So forMcDowell it’snot true that the justification intermediatesbetweenthebeliefandthefactintheknowledgecase,ratherthefact’shaving‘madeitselfperceptuallymanifest’isenoughtodothejobonitsown.
200
8 Conclusion
Ihaveforarguedthefollowingclaims:
(M1) For any proposition,p, ifA has a justified belief thatp then the fact thatAbelievesthatpismysteriouslyrelatedtosomejustificationforit.
(M2) Foranyproposition,p,ifAknowsthatpthenthefactthatAbelievesthatpismysteriouslyrelatedtothefactthatp.
(M3) Foranypropositions,jandp,ifjaffordstheopportunityforknowledgethatpthenjismysteriouslyrelatedtothefactthatp.
(M4) IfAφsintentionallythen,forsomeproposition,p,thefactthatAbelievedthatpexplainswhyitwasprotantorationalforAtoφandismysteriouslyrelatedtothefactthatAφ’d.
I have argued that themysterious relation in each case is the same relation, and that this
relationisanon-causal,transitive,explanatoryrelation.
I shouldnote: this isnota solution to theproblemofdeviantcausal chains foranyof these
cases.Itaketheinterestingquestionsforeachcasetobewhatthisexplanatoryrelationis,why
it obtains, andwhen itobtains: I have called this themystery relationpreciselybecause the
answerstothosequestions,theinterestingquestions,remainsshroudedinmystery.AllthatI
take myself to have done here is to offer some, hopefully, bland observations about this
relation.Nonetheless,Ihopethateventheseblandobservationswillbesufficienttomakemy
case.
In the next chapter I will argue that this relation is transitive with the explanatory relation
between,forinstance,thefactthatIbelievedthatmyfriendhadwonanawardandthefact
thatitwasprotantorationalformetocongratulateher.
201
(XV)
Mysteryrelationsandwhyitisrationaltoact
InwhichIsaythatmysteryrelationsaretransitivewiththeexplanatoryrelationinvolvedinexplainingwhyit isrational. I labelthesortofexplanatoryrelationthatobtainsbetween(i)thefactthatIbelievethatitisrainingand(ii)thefactthatitisprotantorationalformetotakeanumbrella,the‘E’-relation’.Iarguethat the mystery relation is transitive with the E’-relation. I show how thisaccordswithourintuitionsinsomeoftheexamplesalreadyconsidered.
Iwanttosaythatthefactthatmyfriendwonanawardexplainswhyitwasrationalformeto
congratulateher.Ialsowanttosaythatthefactthatthetreesrustleddoesnotexplainwhyit
wasrationalforSallytorun.However,inbothcases,thereisachainofexplanatoryrelations
connectingthetwofacts.1So, if IamtosaywhatIwanttosay,asnotedin§(XIII)4, Ineeda
principled reason for saying that all the explanatory relations involved in the first case are
transitive,whiletheexplanatoryrelationsinvolvedinthesecondcasearenotalltransitive.In
thelastchapterIintroducedthemysteryrelationwithaviewtomakingthiscase.
Ihavealreadysuggestedthatthemerelycausalexplanatoryrelationbetweentherustlingof
thetreesandthefactthatSallyheardabear-likesoundisnottransitivewiththeexplanatory
relationbetweenthefactthatshebelievesthatabearischasingherandthefactthatitispro
tantorationalforhertorun.
Myargumentinthischapteristhis:themysteryrelationsbetweenthefactthatmyfriendhad
won an award and the fact that I believed that she hadare transitivewith the explanatory
relationbetweenthefactIbelievedthatshehadwonanawardandthefactthatitisprotanto
rationalformetocongratulateher.Iarguethatsomethingthatisnotafeatureofanagent’s
psychology(i.e.abeliefordesireorwhathaveyou)explainswhyitisrationalforthemtodo
some action only if it is mysteriously related to a feature of their psychology that, in turn,
explainswhyitisrationalforthemtodothataction.
Iputforwardthreeargumentsfortheclaimthatmysteryrelationsaretransitivewiththesort
ofexplanatoryrelationinvolvedin‘explainingwhyitisrational’:firstly,Iarguethattheyshare
many properties in common, and that the best explanation of why they share so many
1Thatis,thefactthatmyfriendwonanawardexplainswhyIreadthatshehadwonanaward,whichexplainswhy I believed that she hadwon an award,which explainswhy itwaspro tanto rational tocongratulateher.Likewise: thefact thatthetreesrustledexplainswhySallyheardabear-likenowise,whichexplainswhyshebelievedthatabearwaschasingher,whichexplainswhyrunningwasprotantorational.See§(XIII)4forrelateddiagrams.
202
properties isthattheyarethesamesortoftransitive,explanatoryrelation.Secondly, Iargue
thatthebestanalysisofthissortofnon-causalrelation,grounding,takesittobeofasingular,
transitivesort.Thirdly, Iarguethatthis is thebestaccountofwhywemightsay, inordinary
language,that,forinstance,itisrationalformetotakemyumbrellabecauseitisraining.
1 Explainingwhyitisrationaltoact
1.1 TheE’-relation
ThefactthatIbelievethatitisrainingexplainswhyitisprotantorationalformetotakemy
umbrella. The fact that Sally believes that a bear is chasing her explainswhy it ispro tanto
rationalforhertorun.Eachofthesecasesinvolvesaparticularsortofexplanatoryrelation–
forexpositionalconvenienceitwillhelpifwenameit.Letussaythefollowing:
Definition For any proposition p, p is E’-related to the fact that it is pro tantorationalforAtoφ ifandonlyifpexplainswhyitisprotantorationalforAtoφ.
A clarification: theE’-relation isnot a relationbetween, say, the fact that I believe that it is
rainingandtheactoftakingmyumbrella.TheE’-relationrelatesthefactthatIbelievethatitis
rainingtothefactthatitisprotantorationalformetotakemyumbrella.TheE’-relationisjust
a particular sort of explanatory relation; it iswhatever sort of explanatory relation it is that
existsbetweenthosefacts.
1.2 WhatarethepropertiesoftheE’-relation?
WhatcanwesayabouttheE’-relation?First,aswehavealreadyobserved(see§(XIII)3.2.2),it
isanon-causalexplanatoryrelation.ThefactthatSallybelievesthatabearischasingherdoes
not cause it to bepro tanto rational for her to run, and you should accept that even if you
thinkitcauseshertorun.Causationjustseemstobethewrongwaytocharacterisethissortof
relation.So,theE’-relationisnon-causal.
Second,itisatransitivesortofexplanatoryrelation.Itseemsrighttosaythatifit’sprotanto
rational for me to get some exercise, then that fact (partly2) explains why it is pro tanto
rationalformetogoswimming.Nowsupposethatitisprotantorationalformetogetsome
exercise(partly)becauseIbelievethatexercisewillliftmyspirits.Insuchcircumstancesitalso
seemsrighttosaythatthefactthatIbelievethatexercisewill liftmyspirits(partly)explains
whyitisprotantorationalformetogoswimming.
2Togetherwith,perhaps,thefactthatIbelievethatgoingswimmingisgettingexercise,etc.
203
Sowhat?Well,thefactthatIbelievethatexercisewillliftmyspirits,p,explainswhyitispro
tantorationalformetogetsomeexercise,q.Andthefactthatitisprotantorationalformeto
getsomeexercise(i.e.q)explainswhyit isprotantorationalformetoswim,r.Andthefact
thatIbelievethatexercisewillliftmyspirits(i.e.p)alsoexplainswhyitisprotantorationalfor
metoswim(i.e.r).Thus:pexplainsq,qexplainsrandpexplainsr;so,atleastinthisinstance,
theexplanatoryrelationsinvolvedsatisfytransitivity.
Now,Isubmit,suchexplanatoryrelationsmustalwaysbetransitive:ifitisrationalforAtoφ
andA believes thatψing is ameans toφing, then the fact that it is rational forA toφwill
explain why it is rational for A toψ. And whatever explains why it is rational for A toφ,
togetherwithA’sbeliefthatψingisameanstoφing,willlikewiseexplainwhyitisrationalfor
Atoψ.So,theexplanatoryrelationinvolvedinexplainingwhyit isrationaltodosomething,
i.e.theE’-relation,isatransitiverelation.
Third,totheextentthatontologicalpriorityisameaningfulconcept,thetruth-makersofthat
which explains why it is pro tanto rational to do some action are ontologicallyprior to the
rationality of actions. Consider: it’s possible that someone’s beliefs and desires/evaluative
judgements (delete or replace as appropriate)may never align in away that is sufficient to
make any action rational (perhaps theyhave veryodddesires, or normal desires, butweird
beliefs,ornormaldesiresandnormalbeliefsbut just live inadreadfully limitedworld)–the
factthatnoactionisrationaldoesnotimpingeontheirabilitytohavebeliefsanddesires.So
you can have beliefs and desires without there being any rational actions. However, the
propertyofbeingrationalcannotbeinstantiatedwithoutbeliefsanddesires.So,Isuggest,the
latterareontologicallypriortotheformer;whichistosaythattheE’-relationis(underpinned
by)arelationofontologicalpriority.
Fourth,andrelatedly,thissortofexplanatoryrelationentailsan‘invirtueof’claim.Thatis,if
thefactthatpexplainswhyitisrationalforAtoφ,thenweareseeminglyalwaysabletosay
thatitisrationalforAtoφinvirtueofthefactthatp.Forinstance,itisrationalforSallytorun
invirtueofthefactthatshebelievedthatabearischasingher.So,E’-relationsentail‘invirtue
of’claims.
2 Arelationincommon
I want to convince you that the mystery relation (a non-causal explanatory relation) is
transitivewiththeE’-relation(alsoanon-causalexplanatoryrelation).Thatis,Iwillarguethat:
(M5) Foranypropositions,p,q and r, ifp ismysteriously related toq andq is E’-relatedtorthenpisE’-relatedtor.
204
Why should you believe this? Because, I argue, these relations are both the same sort of
transitive, non-causal explanatory relations. I put forward three arguments for this claim:
firstly, the pervasive similarities between the mystery relation and the E’-relation are best
explainedby thepresenceof a common relation. Secondly, thebest accountof explanatory
relationsofthissort,grounding,takesthemtobe(i)unitary(atleastuntilprovenotherwise)
and (ii) transitive. Thirdly, accepting (M5) gives us the best explanation of why, in ordinary
language,wemightsay,forinstance,thatitisrationalformetotakemyumbrellabecauseitis
raining.
Apointofclarification:Iamonlyarguingthattheseexplanatoryrelationsareofthesame,sort
(thatis,thattheybelongtothesamefamilyof(transitive)explanatoryrelations),becausethat
isenoughformyargumentfor(M5).Itisconsistentwiththisclaimtosupposethattheyarein
fact the same explanatory relation,3however, it is likewise consistent with this claim to
suppose that they aren’t. I take no particular stance onwhether or not they are the same
explanatoryrelationjustbecauseIdon’tneedtoformyargument.
2.1 Theserelationsaresimilarbecausetheyareofacommonkind
2.1.1 Thereareahostofsimilaritiesbetweentherelations
WhatarethesimilaritiesbetweenthemysteryrelationandtheE’-relation?Wehavealready
noted that they are bothnon-causal explanatory relations. Associatedwith their bothbeing
explanatory relations comes their both being asymmetric, irreflexive and non-monotonic.
Whatelse?
First, they are both transitive relations. In the previous chapter I demonstrated that the
mystery relation is transitive. In the previous section I demonstrated that the E’-relation is
transitive.
Second, they areboth relationsof ontological priority. I have alreadynoted that thatwhich
explainswhyanaction is rationalmustbeontologicallyprior totherationalityof theaction.
What of themystery relation? The examples in the previous chapter relate, variously: facts
abouttheworld(ortheagent’sperceptionofit)tofactsaboutwhattheagentbelieves;facts
abouttheworldto justificationsforbelief;andbeliefstoactions.Presumablyanyonewho is
notanidealistandfindssomemeaninginthenotionofontologicalprioritywillagreethatfacts
3Andthatmightevenbemoreparsimoniousthanthinkingthattheyaredifferentexplanatoryrelationsthatbelongtoasinglefamily.
205
of the former kind are ontologically prior to the latter.4That is, presumably: the world is
ontologically prior to perceptual experiences of it, perceptual experiences are ontologically
priortobeliefs,andbeliefsareontologicallypriortoactions(Ihavealreadyarguedforthelast
claimintheprevioussection).Thesedon’tallneedtobetruetomakethecase,butIthinkthat
theyare.
Third, they both entail ‘in virtue of’ claims. Again, I have already noted that the E’-relation
relationalwaysentailsan‘invirtueofclaim’,andsotoodoesthemysteryrelation.Consider:it
seemsrighttosaythatSeanbelievesthathiswifehasbeenunfaithfultohim invirtueofthe
fact that shehasbeenunfaithful tohimand thatSallybelieves thatabear is chasingher in
virtueof thefactthatsheheardabear-likesound– incontrast itdoesnotseemrighttosay
thatEvabelievesthatherhusbandhasbeenunfaithfultoherinvirtueofthefactthatshesaw
himkissinganotherwoman.5
2.1.2 Theyaresimilarbecausetheyhavesomeexplanatoryrelationincommon
I think that the best explanation of these similarities is that the two relations both are the
samesortoftransitive,non-causal,explanatoryrelation.
Youmightobjecttothis.Perhapsyouarescepticaloftalkofontologicalpriority6or‘invirtue
of’7relations. In which case you will doubt that there is much in the way of similarity that
needs explaining. So be it. I, like many others,8increasingly take these to be meaningful
concepts, and I suggest that the fact that the same formofwords can be used in different
cases is, at least, aprima facie reason for thinking that there is a common relationatwork.
Since there is aprima facie reason for thinking that there is a common relation I, following
Audi,‘taketheburdenofprooftobeonthosewhothinktherearedifferentrelationsatwork
toshowwhy.’(P.Audi2012b,689)
Another objection: perhaps you say that these similarities, at best, characterise a genus of
non-causalexplanationofwhichthemysteryrelationandtheE’-relationaredifferentspecies–
theirmembershipof thegenusaccounts for their similarity,but theyare separatedby their
species-hood.Thatis,perhapsallofthesesimilaritiesderivefromtherebeingan‘E-relation’,
where an E-relation is a non-causal explanatory relation that is not transitive with other4And, I think, even some idealists can find ameaningful degreeof ontological priority of suitably re-describedfactsoftheformerkindoverfactsaboutbeliefs.5Nordoesitseemrighttosaythatshebelievesthatherhusbandhasbeenunfaithfultoherinvirtueofthefactthathehas;Eva’sbeliefisneitherknowledgeablenorjustified.6E.g.Hofweber(2009)7E.g. ‘We knowwe are in the realmofmurkymetaphysics by the presence of theweaselwords “invirtueof”.’(Oliver1996,48)8Seee.g.Rosen(2010),Fine(2012)andAudi(2012a).
206
E-relations, even if it is transitivewith itself. Nothingwould then force us to conclude that
mysteryrelationsandtheE’-rationalrelationarethesamesortofE-relation,whichiswhatis
requiredforustoadmitthetruthof(M5).
Of course, this is possible, and, indeed, the different examples are not alike in all respects.
However,whatneedstobeshownisnotthattherearedifferencesbetweenthecasesgiven,
butthatwhatdifferentiatesthemissuchthattheycannotbethesamesortofE-relation.And,
again,Itaketheburdenofproofheretobeonthosewhothinkthattheyaren’t.
2.2 Thebestaccountofsuchrelationstakesthemtobeofonekind
There is a readily available analysis of the ontological underpinnings of these explanatory
relations according to which they are the same, transitive explanatory relation: namely,
grounding.
2.2.1 Whataregroundingrelations?
Grounding is the ‘in vogue’ relation in contemporary metaphysics. Here are a few
characteristicgroundingclaims9:
1.Mentalfactsobtaininvirtueofneurophysiologicalfacts;
2.Dispositionalpropertiesaregroundedincategoricalproperties;
3.Legalfactsaregroundedinnon-legal,e.g.social,facts;
4.Morallywrongactsarewronginvirtueofnon-moralfacts;
5.Normativefactsaregroundedinnaturalfacts.
(Correia2010,251)
Assuming that grounding is abona fide relation; here are some things that are taken to be
essentialtogrounding10:
− Itisanexplanatoryrelation;11
− Itisanotmerelycausalrelation12;
− Itistransitive;
9Note: the usefulness of grounding doesn’t hang on the truth of these claims but whether or notgroundingcanbeusedtocharacterisewhattheclaimsareclaimsabout.10Seee.g.Rosen(2010),Fine(2012)andAudi(2012a).11Asanaside:Weshouldnotethatonecouldquestionwhethergroundingrelationsarethenon-causalexplanatoryrelationsthemselvesortheontologicaldeterminationrelationsthatunderpinsthem.Thatquestionis,however,largelyorthogonaltoourdiscussion.12Thoughthereisnorequirementthatgroundingrelationsexcludecausalrelations.
207
− ‘Thefactthatpisgroundedinthefactthatq’canbecharacterisedbylocutionslike‘pisthecaseinvirtueofq’;and
− Itisarelationofontologicalpriority.
Weshouldnoteanimportantdistinctionbetweenafullandapartialground.Hereisatypical
characterisation:
A is apartial ground forC ifA,on itsownorwith someother truths, is agroundofC. (Fine2012,50)
Now,althoughfullgroundsaretypicallytakentonecessitatethatwhichtheyground13,merely
partial grounds are not. For instance, the possibility of castling with one’s kingside rook is
partiallygrounded in the fact thatnopiecesobstruct themovebut that factalonedoesnot
guaranteethatonecancastlewithone’skingsiderook(forinstance,thekingmaybeincheck).
Andwhilefullgroundsmaynecessitatethatwhichtheyground,neitherfullnorpartialgrounds
needbenecessaryforthatwhichtheyground.Totakeastandardexample:thefactthatthis
ballisscarletfullygroundsthefactthatitisred,buttheformerisnotnecessaryforthelatter
(theballcouldbevermillionorruby).14
2.2.2 Theseexplanatoryrelationsaregroundingrelations
Grounding theoristsaimtoexplainwhyavarietyof relations inseeminglydifferentcontexts
exhibitthesamepropertiesbysuggestingthattheyallsharethecommon‘grounding’relation
(andthenprovidingananalysisofthatrelation).AsAudiremarks:
Such pervasive similarity among such diverse subject matters cries out for explanation. Iproposethatwhataccountsforthesimilarityissimplythatthereisasinglerelationatworkineachcase.(P.Audi2012b,689)
In§2.1.1, Iarguedthat themystery relationandtheE’-relationhaveahostofproperties in
common.Havingsetoutthepropertiesofgroundingrelationsabove,wecannowseethatthe
propertiesthattheserelationshaveincommonjustarethepropertiesofgroundingrelations.
Like grounding relations, these explanatory relations are non-causal, transitive, explanatory
relations; they can be characterised by locutions like ‘in virtue of’; they involve claims of
ontological priority; and they contribute to necessitating that towhich they relate although
theyneednotbenecessary for it toobtain.FollowingAudi’s logic, then, thebestaccountof
whytheserelationssharethosepropertiesisbecausethey,too,aregroundingrelations.
13Forexample, seeRosen (2010),Fine (2012)andAudi (2012b).There is somedissent fromthisview(forexample,seeChudnoff(2011)andLeuenberger(2014)).14The analogousness of the concepts of full and partial groundswith the concepts of full and partialexplanationinvokedthroughoutthisdiscussionispresumablyclear.
208
Moreover, the distinction between full and partial grounds readily accommodates what we
earlierobservedaboutthestrictlypartialcontributionofthefactthatSallybelievesthatabear
ischasinghertotheexplanationofwhyitisrationalforhertorun:wecansaythatthepartial
explanation partially grounds that which it explains. It similarly provides us with a ready
characterisation of the strictly partial way inwhich the fact that it is raining explainswhy I
believethatitisrainingevenwhenIknowthatitisraining(foritdoesnotdosoonitsown):
againwecansaythattheformerfactpartiallygroundsthelatter.15
So here is another reason to believe (M5): there is awell-developed analysis of the sort of
explanations that both the E’-relation and mystery relations appear to exhibit, grounding,
whichtakesthemtobothbeacommon,non-causal, transitive,explanatoryrelation.That is,
thebestavailableaccountofthesortofexplanations involved intheE’-rationalandmystery
relationsentailsthetruthof(M5).
2.3 Theirtransitivitymakessenseofordinarylanguage
A final consideration in support of (M5) is that, provided that you agreewith (M1)-(M4), it
provides thebestaccountofordinary languageexpressions inwhichnon-psychological facts
aresaidtoexplainwhyitisrationaltodosomething.
Somethingswemightreadilysay:thefactthatSallyheardabear-likesound(inawoodthat
sheknewtocontainbears)atleastpartlyexplainswhyitwasrationalforhertorun.Thefact
that my friend won an award at least partially explains why it was rational for me to
congratulateher.Itwasrationalformetotakemyumbrellabecauseitwasraining.
Aswithnormativereasonexplanationsofaction,therearethreepossibleaccountsofwhatis
going on in these sorts of explanations: either the purported explanans explains the
explanandumelliptically,directlyorindirectly.Now,Isuggestthatthesameargumentsagainst
theelliptical anddirectaccountsofnormative reasonexplanationalsoapply to theelliptical
anddirect accountsof theexplanationofwhy it is rational (see§(X)). In contrast, provided
thatyouagreewithmyaccountofmysteryrelations,theclaimthattheyaretransitivewiththe
E’-relation (i.e. (M5)) furnishes us with an account of these explanations that is thoroughly
natural.
15Ourreadinessoftotalkintermsofgroundsforbelieforgroundsforactionis,perhaps,furthergristtothismill.
209
3 Whennon-psychologicalfactsexplainwhyitisrational
Recall the challenge set out in §(XIII)4: I need a principled account of why it is that, for
instance, theexplanatoryrelations fromthebelieved fact to therationalityof theactionare
transitiveinthe‘award’casebutnotinthe‘carbonmonoxide’case.Whatisthataccount?
It is this: something that isn’tadirect featureofanagent’spsychologycanexplainwhy it is
rational for them to do some action only if it is mysteriously related to a feature of their
psychologythatexplainswhythatactionisrational.Inparticular,amerelycausalexplanatory
relationisnotsufficient.
3.1 Someexamples
Itwill helpmake theaccount clear ifwe revisit someof thedifferent cases considered.The
diagrams below set out three cases. I have labelled the explanatory relations as follows:
instancesofthemysteryrelationaremarked‘⇝’;instancesoftheE’-relationaremarked‘E’’;
andmerelycausalrelationsaremarked‘c’.
FigureXV-1:Theexplanatoryrelationsintheawardcase.
FigureXV-2:Theexplanatoryrelationsinthecarbonmonoxidecase
FigureXV-3:TheexplanatoryrelationsintheEvacase
Iwill run througheachof theseexamples in turn. In theawardcase (FigureXV-1) there isa
chainofnon-causalexplanatoryrelationsofacommonsortthat linksthefactthatmyfriend
won an award to the fact that it was pro tanto rational for me to congratulate her. In
particular:thefactthatmyfriendwonanawardismysteriouslyrelatedtothefactthatIread
Ireadthatshehadwonan
award
Myfriendwonanaward
Ibelievedthatshewonan
awardexplains
Congratulatingherwasprotantorationalexplains explains
⇝ ⇝ E’
Itappearedasthoughmyfriendwaswarningme
ofCO
TherewasCOintheroom
IbelievedthattherewasCOintheroomexplains
Leavingwasprotantorational
explainsexplains
⇝c E’
Itappearedtoherasthoughherhusbandwaskissinganotherwoman
Eva’shusbandwas
unfaithful
Shebelievedthather
husbandwasunfaithfulexplains
Divorcinghimwasprotanto
rationalexplainsexplains
E’⇝ c
210
thatshehadwonanawardinthenewspaper,whichismysteriouslyrelatedtothefactthatI
believed that she had won an award, which is E’-related to the fact that it was pro tanto
rationalformetocongratulateher.
Now, since themystery relation is transitive, and transitivewith theE’-relation (from (M5)),
thatmeansthatthefactthatmyfriendwonanawardisE’-relatedtothefactthatitwaspro
tanto rational for me to congratulate her; that is: the fact that my friend won an award
explainswhyitwasprotantorationalformetocongratulateher.
Incontrast,inthecarbonmonoxidecase,thereisnochainofnon-causalexplanatoryrelations
linkingthefactthattherewascarbonmonoxideintheroomtothefactthatitwasprotanto
rational formeto leave it.Thefactthattherewascarbonmonoxide intheroomis(atbest)
merelycausallyrelatedtothefactthatitappearedtomeasthoughmyfriendwaswarningme
aboutthecarbonmonoxide.However,thefactthatitappearedtomeasthoughmyfriendwas
warningmeaboutthecarbonmonoxideismysteriouslyrelatedtothefactthatIbelievedthat
therewas(becauseIbasedmybeliefonthewaythingsappearedtobe),whichisE’-relatedto
thefactthatitwasprotantorationalformetoleave.So,again,giventhetransitivityofthese
relations,thefactthat itappearedtomeasthoughtherewascarbonmonoxideintheroom
doesexplainwhyitwasprotantorationalformetoleave.16
Finally,inEva’scase,thereislikewisenochainofnon-causalexplanatoryrelationslinkingthe
factthatherhusbandwasunfaithfultothefactthatitwasprotantorationalforhertodivorce
him.WhilethatfactismysteriouslyrelatedtothejustificationEvahasforbelievingit(i.e.that
it appeared toher as thoughhewas kissing anotherwoman), since that justification is only
merelycausallyrelatedtoherbelief, thenon-casualexplanatorychaindoesnotextendfrom
thefactintheworld,throughthejustification,totherationalityoftheaction.Thatis,neither
the fact that Eva’s husband was unfaithful to her, nor the fact that it appeared to her as
thoughhewaskissinganotherwomanexplainswhyitwasprotantorationalforhertodivorce
him.
16Itwillpresumablyalsobeclearhowthesamelineofreasoningshouldleadustotheconclusionthatthe rustling of the trees does not explain why it is (pro tanto) rational for Sally to run. That is, therustling of the trees is merely causally related to her hearing a bear-like sound, so the explanatorybetweenthetwofacts isnottransitivewiththeE’-relation,as itwouldneedtobe inordertoexplainwhyheractionwas(protanto)rational.
211
Thetablebelowsummarisesthesecases:
Example Thebelievedfact/justificationforbelievingitExplainswhyitwasprotantorationaltoact?
Evacase
f Herhusbandwasunfaithfultoher û
jItappearedasthoughherhusbandwaskissing
anotherwoman û
COcasef Theroomwasfullofcarbonmonoxide û
j Itappearedasthoughmyfriendwaswarningme ü
Awardcase
f Myfriendwonanaward ü
j Ireadthatshehadwonanawardinthenewspaper ü
TableXV-1:Asummaryofwhatexplainsandwhatdoesn'tineachcase
4 Conclusion
Ihavearguedthatsomethingthatisnotadirectfeatureofanagent’spsychologycanexplain
why their action is pro tanto rational only if it is mysteriously related to a feature of their
psychologythat,inturn,explainswhytheiractionisprotantorational.Ihavearguedforthis
on the basis that themystery relation is transitivewith the E’-relation,whilemerely causal
relationsarenot.IconsidersomeobjectionstothisproposalintheAppendixtothischapter.
Thenext, and final, chapter combines the insightsof thisdiscussion to setoutmy theoryof
reasons.
Appendix
A.1 Objections
Objection1 Thechallengewasforyoutoshowwhyitwasthat(i)thefactthatyourfriend
wonanawardexplainedwhy itwasrational foryoutocongratulateher,despite (ii) the fact
thattheroomwasfullofcarbonmonoxidedidn’texplainwhyitwasrationalforyoutoleave.
Your‘answer’isthatitisamystery,butthat’snosortofanswer!Youarejustre-labellingthe
problemthatwasalreadydiagnosed.Whatprogresshasreallybeenmade?Don’twewantto
knowwhythisrelationshipispresentin(i)andnotin(ii)?
Response Well, here is some progress that has beenmade: we’ve identified that the
mysteryrelationispartofsomefamilyofexplanatoryrelations,wheremembersofthatfamily
havecertainproperties.Thatisatleastthestartingpointforasystematicinvestigationofwhy
therelationshipispresent.
212
More generally, however, I amnot trying to answer the question ofwhy the relationship is
present.ThequestionIamtryingtoansweriswhytheexplanatorychainin(i)istransitiveand
whyitisn’tin(ii).Myansweristhatthereisachainof(mysterious)explanatoryrelationsin(i)
that there isn’t in (ii)and that thesemystery relationsare transitivewith theE’-relation (i.e.
(M5)).IfyouacceptthesetwoclaimsthatisenoughtoanswerthequestionthatIamtryingto
answer.
Thequestionastowhy there isamysteryrelationispresent in(i)but isn’t in(ii) isadeeper
(more distal) and perhaps more interesting question than the one I am answering. But
answering it is beyond the scope ofwhat I need to do tomakemy case. In identifying the
mystery relation as an explanatory relation, and showing that it is transitive with the
E’-relation,andshowingthatitobtainsin(i)butnotin(ii), Ihaveananswertothechallenge
posed,andIdon’tneedmorethanthat.
Objection2 If grounding-based formulationsofphysicalismare true, thenyouare forced
intotheabsurdconclusionthat,say, the fact thatSally’sbrain is instateBexplainswhy it is
rationalforhertorun.
Muchofyourargumentfor (M5)wasbasedontheclaimthatthebestexplanationofall the
grounding-likepropertiesthatthemysteryrelationandtheE’-relationsshareisthattheyboth
involveacommon,explanatory relation. Itwas for that reason thatyousuggested that they
weretransitive,so,yousuggest:
(a) Allexplanatoryrelationsthatexhibitgrounding-likepropertiesaretransitivewitheachother.
Grounding-based formulations of physicalism (e.g. Correia 2010; Kroedel and Schulz 2016)
holdthatallmental factsaregrounded inphysicalones.Now,grounding-basedformulations
of physicalism take the relation betweenmental and physical facts to exhibit exactly those
propertiesthatyousaidthattheE’-relationandmysteryrelationexhibit.17Thus:
(b) Allmental factsstand inaparticularexplanatory relation tosomephysical factsandthatexplanatoryrelationexhibitsgrounding-likeproperties.
Thus, from (a) and (b), togetherwith your claim that the E’-relation exhibits grounding-like
properties, we infer that the explanatory relation between the mental and the physical is
transitivewiththeE’-relation.
17Forinstance:Therelationisanon-causalexplanatoryrelation.Itcanbecharacterisedbythe‘invirtueof’locution.Etc.
213
So,somephysicalfact(e.g.thefactthatSally’sbrainisinstateB)non-causallyexplainsthefact
shebelievesthatitisrationalforhertorun.AndsincethatistransitivewiththeE’-relation,we
canconcludethatthefactthatSally’sbrainisinstateBexplainswhyitisprotantorationalfor
hertorun.Butthatisabsurd!
Response IagreethatitisabsurdtosaythatfactsaboutSally’sbrainstateexplainwhyit
wasprotantorationalforhertorun.So,topreservemytheoryinthefaceofthisconclusionI
mustrejecteither(a)or(b).18
Asitis,I’mnotconvincedthat(b)istrue–inparticular,it’snotcleartomethatfactsaboutan
agent’sbrainstateexplaintheirmentalstate.However,theclaimthatreductionrelationsare
explanatory relations is true in a number of popular construals of physicalism beyond just
grounding-basedformulations19–sorejectingitisnotwithoutitscosts.
Isthereawaytorendermytheoryconsistentwithsuchconstrualsofphysicalism,byrejecting
(a),thatdoesn’talsounderminemyargumentfor(M5)?Isuggestthatthereis.
Recall that inmyargument for (M5), I considered thepossibility that theE’-relationand the
mystery relation involve different species of a genus of non-causal explanatory relations, E-
relations,andthatitistheirbelongingtothatgenusthataccountsforthepropertiestheyhave
in common. The argumentwas that E’-relations andmystery relations need not involve the
samerelationtoexhibitthesameproperties–beingmembersofthesamefamilyissufficient.
MyresponsetothatpossibilitywasAudi’s:thattheburdenofproofisonthosewhotakethe
relationstobedifferenttoshowthattheyaredifferent.20
With respect tomystery relations, I can thinkofno compelling reasonas towhywe should
think that the sort of explanation involved is entirely different to the sort involved in the
E’-relation.However,unlikethemysteryrelation,totheextentthattherelationbetweenthe
mentalandthephysicalreally isanon-causalexplanatoryrelation(i.e.giventhatweassume
(b)),Ithinkthatwehavereasontothinkthatitdoesnotinvolvethesamesortofnon-causal
explanatoryrelationastheE’-relation.Thatis,Ithinkthatinthiscasewedohaveareasonfor
thinkingtheyinvolvedifferentsortsofexplanation,butinthecaseofthemysteryrelation,we
don’t.
Whatisthereasonwehave?Iarguethatthesortofexplanatoryrelationthatismeanttoexist
between the mental and the physical is not transitive with the sort of explanation the
18Atleastsolongasweassumethat,onanygrounding-basedversionofphysicalism,Sally’sbrainstateisthe‘ground’ofhermentalstate.19ThisiswhatCrane(2000)calls‘conceptualreduction’(asopposedto‘ontologicalreduction’).20Schaffer(2009,377)makessimilarremarks.
214
E’-relation involves–andthatthis isareasonforthinkingthattheydonot involvethesame
relation.
Letusintroduceadistinction(whichismeanttobeintuitive,butwhichwillbesubstantiated
with examples) between two sorts of non-causal explanation: vertical and horizontal.21Say
thatverticalexplanatoryrelations,Ev-relations,areexplanatoryreductionrelations(ifthereare
suchthings), likethosebetweenthementalandthephysical.Saythathorizontalexplanatory
relations,Eh-relations,arelikethoseinvolvedinE’-relationsandmysteryrelations.WhatIwill
need to demonstrate is that horizontal explanatory relations are not transitivewith vertical
explanatoryrelations(i.e.that(a)isfalse).
Tostart:consider that the fact that this sliceofcake is thebiggestexplainswhy itwouldbe
impolitetotakeit.This,Isubmit,isahorizontalexplanatoryrelation–thefactthatthecakeis
biggestmakestaking it impolite inthesamewaythatabeliefthat it’scoldexplainswhyit is
rationaltoturnontheheating.
Nowgiventhatreduction isanexplanatoryrelation,asthe ‘conceptualreduction’physicalist
supposes, factsabout themicrophysicalpropertiesof the slice (and theother slices)explain
whyitisthelargestslice.Isuggestthatthisexplanatoryrelation,quareductionrelation,isthe
samesortofverticalexplanatoryrelation,i.e.theEv-relation,astherelationbetweenmental
andphysicalfacts.
But,Isuggest,itisoddtosaythatfactsaboutthemicrophysicalstructureofthesliceofcake
areapartoftheexplanationofwhyit’simpolitetotakeit.Thatis,wecannotinferfrom(i)the
fact that facts about the microphysical structure of the cake (partly) explain why it is the
largestslice;and(ii) thefactthatthefactthat it isthe largestslice(partly)explainswhy it is
impolite to take it; to (iii) the conclusion that facts about themicrophysical structureof the
cake(partly)explainwhyitisimpolitetotakeit.Inotherwords,thetransitivityofexplanation
breaksdown.Now,sincetheEh-relationisatransitiverelation,wehaveareasontothinkthat
theEv-relation is not transitivewith theEh-relation, so they aredistinct sortsof explanatory
relation.
This particular argument relies on the reduction of macro-physical to micro-physical facts
being relevantly analogous to the reduction of mental facts to physical ones. However, I
21Ihaveappropriatedandre-purposedthisterminologyfromJaegwonKim(2003).Kimtalksintermsof‘verticaldetermination’and‘horizontaldetermination’–whilstverticaldeterminationisclosetowhatIcharacterise as ‘vertical explanation’, Kim’s notionof horizontal determination is explicitly causal. ForthatreasonIdistinguishbetweenverticalexplanationandhorizontalexplanation.
215
cannotseehowtransitivitywithEh-relationscouldfailfortheformer,(asIhavearguedthatit
does)butsucceedwiththelatter,soIhaveinferredthatitdoesn’t.22
In contrast, as I have argued above, the transition from amystery relation to anE’-relation
seeminglydoes involvea transitive formof explanation: the claimsweendupmaking ifwe
taketherelationtobetransitivethere(e.g.thefactthatitiscoldexplainswhyit’srationalfor
metoputtheheatingon)seemcommonplaceincomparisontothepreposterousnessofthe
claimswemakeifweassumethatverticalexplanationsarecontinuouswithhorizontalones.
MypointisjustthatwhileEv-relationsare‘transitivewith’eachother,theyarenot‘transitive
with’Eh-relations.
Thus,Ithinkthatthereisacrediblewayofrejecting(a)thatdoesnotunderminemyargument
for (M5), so that the proponent of grounding-based formulations of physicalism (and
conceptual reduction more generally) can accept my theory without arriving at the absurd
conclusion that Sally’s brain state explainswhy it is rational forher to run. Theonly cost to
such a physicalist is that they must admit that there are at least two kinds of grounding
relation that, despite sharing a host of properties, are not transitive with one another.
However,this isnotacost IneedpaysinceIanywaydon’tthinkthatreductionrelationsare
explanatory(thatis,mypreferredresponsetothisobjectionistoreject(b)).
Objection3 Deviantcausalchainsaffectmanyotheranalysesofcausation,forinstance:
Itwouldnotworktosaythattheheatoftheovencooksthechickenifandonlyiftheheatoftheovencausesthechickentobeinacookedstate.Theheatoftheovenmighttriggersomemicrowaveactivityelsewherewhichcausesthechickentobeinthatstate;inthiscasetheheatwouldnothavecookedthechicken.(Stout2010,161)
That being so, the mystery relation is presumably not only restricted to the cases you’ve
considered.Supposingthatthemysteryrelationdifferentiatesdeviantfromnon-deviantcases,
wemightsaythis:iftheovencooksthechickenthenthefactthattheovencookedthechicken
ismysteriouslyrelatedtothefactthatthechickenreachedacookedstate.
So,let’simagineatwicedeviantcarbonmonoxidecase.Insteadofcausingyoutohallucinate
in the ‘normal’way, thecarbonmonoxidecausesa creature in the roomtohallucinate that
youareattacking it, and,unbeknownst toyou, it injectsyouwithhallucinatoryvenom.That
thenmakesyouhallucinatethatyourfriendiswarningyouaboutthecarbonmonoxideandso
on.
22Indeed,thefactthatitisabsurdtoclaimthatthefactthatSally’sbrainisinstateBexplainswhyitispro tanto rational for her to run is clear evidence that the explanatory relations involved are nottransitive.
216
In theoriginalexample thecarbonmonoxidemadeyouhallucinate inanon-deviantway. In
thisnewcaseitmadeyouhallucinateinadeviantway.Butifwhatdifferentiatesdeviantcases
fromnon-deviantonesisthemysteryrelation,thenthatmustmeantthat,inthenon-deviant
case,thefactthattherewascarbonmonoxideintheroom ismysteriouslyrelatedtothefact
that itappearedtoyouas thoughyour friendwaswarningyouabout thecarbonmonoxide.
That being so, given the transitivity of themystery relation, the fact that therewas carbon
monoxideintheroomoughttoexplainwhyitwasrationalforyoutoleaveit.
Response First,Ineverofferedthemysteryrelationasananalysisofwhatdifferentiates
deviantcausalchainsfromnon-deviantonesinanycontext.Itwasspecificallyrestrictedtothe
casesconsidered.
Second,Idonotthinkitshouldbeextendedtoothercases.Whatdistinguishesdeviantcausal
chainsfromnon-deviantonesismysteriousinthementalcasebecauseitinvolvesthemental
(or,at least, therepresentational). In thecaseofcookingthechicken,while Idon’thavethe
solutiontotheproblem,Idon’tthinkthereisthesamefundamentallymysteriousproblemat
work.Whatever themystery relation is, it is, I suggest, to do with the rational faculties of
agents,whereas,itseemstome,thequestionofdeviantcausalchainsinmerecausalcasesisa
merequestionofmechanism.
Perhaps you don’t find this response very compelling. I’m afraid I don’t have a more
compellingone;shouldthisprovetobeaninsuperabledifficultythenIshouldhavetolookfor
someotheranalysisastowhyfactsabouttheworldcanexplainwhyouractionsarerational.
However,Istruggletofindthisobjectioninsuperable.
217
(XVI)
Anewtheoryofreasons
InwhichIsetoutmytheoryofreasons.Idiscusswhatexplanatoryrationalismsaysabout theapplicationofeachreasonexpressiontothecasewhere I takemyumbrellahavingseenthatitisraining.Ishowhowexplanatoryrationalismsolves theproblems facedbyother theories. I suggest that thebest theoryofreasons is a pluralist theory of reason that combines explanatory rationalismand favourism; I call this theory ‘new pluralism’. I show how explanatoryrationalism enables new pluralism to meet the main challenge to pluralisttheories.
In §(V), I argued that ifweare to solveall of theproblemsdiscussed in§§(II)-(VI) thenwe
needanewfamilyofclaimsaboutreasons.In§(VI),Isuggestedthatexplanatoryrationalism
was the new family of claims thatwe needed, however, I noted, The Explanatory Exclusion
Problempresentedasignificantchallengeforit.Theintermediatingchaptershavearguedthat
TheExplanatoryExclusionProblemisnottheproblemitseemstobe,andthatit istherefore
noobstacletoexplanatoryrationalism’ssolvingtheproblemsdiscussedin§§(II)-(V).
Now it is time to discuss howexplanatory rationalism solves theseproblems. The answer is
perhapsobvious:explanatoryrationalismsolvestheseproblemsbyrejectingthetroublesome
viewsthatgaverisetothem;thatis,explanatoryrationalismrejectsfavourism,psychologism
anddeliberativism.Moregenerally,explanatoryrationalismisnotsusceptibletosimilarsorts
ofproblemsbecauseitisconsistentbothwiththeideathatagentsalwaysactforpsychological
reasons, andwith the idea that they sometimes also act for normative reasons. Because of
this,Iargue,explanatoryrationalismisthebestunivocalaccountofwhatitistobeareason.
However,youwillrecallfrom§(V)thatIthinkthatourtheoryofreasonsoughttobepluralist
becauseIsharethetwosenses intuition.Tothatend, Ipresentnewpluralism: Isuggestthat
onesenseofwhatitistobeareasonisexplanatoryrationalistandtheothersenseisfavourist.
Thatismytheoryofreasons.
In what follows I revisit explanatory rationalism, and consider what it says about what my
reasonswerewhenIsawthatitwasraining,andconsequentlytookmyumbrella.Ithenshow
howexplanatory rationalismsolves theproblemsthataffectother theories.Finally, I setout
newpluralismandshowhowitaddressesthemainchallengetopluralisttheories.
218
1 ExplanatoryRationalism:Revisited
Recallwhatexplanatoryrationalismhastosayabouteachreasonexpression:
Reasonexpression Explanatoryrationalism
Foranyp,pisareasonforAtoφ… …ifandonlyifpexplainswhyitisprotantorationalforAtoφ.
Foranyp,pisareasonforA’sφing… …ifandonlyifpexplainswhyitisprotantorationalforAtoφandpmakesA’sφing,insomerespect,worthdoing.
Foranyp,pisareasonAhastoφ… …ifandonlyifpexplainswhyitisprotantorationalforAtoφ.
Foranyp,pisA’sreasonforφing… …ifandonlyifpexplainswhyitisprotantorationalforAtoφandexplains(intherightway)whyAφ’d.
TableXVI-1:Explanatoryrationalism
Itmayhelpputthistheoryintocontextifweconsideranexample.Beforedoingso,however,
thereissomethingthatperhapsstillneedstobemadeexplicit:whatitmeanstoexplaininthe
rightwaywhysomeoneacted.ForthepurposeofthisdiscussionIwillassumethatifsomefact
ismysteriously relatedtothe fact that theagentdidwhat theydid, then itexplains it in the
rightway.Idiscussthisassumptionfurtherin§A.2oftheAppendixtothischapter.1
Thatclarificationhavingbeenmade,letusre-considerthefollowingexample:Ilookoutofthe
windowandseerain.Iknowthatit’sraining,soItakemyumbrellawhenIleavethehouse.
1.1 Theexplanatoryrelationsinvolved
What are the explanatory relations in this example? Well, the fact that it was raining is
mysteriously related to the fact that it appeared to me as though it was raining, which is
mysteriouslyrelatedtothefactthatIbelievedthatitwasraining,whichismysteriouslyrelated
to the fact that I tookmyumbrella. Furthermore, the fact that Ibelieved that itwas raining
explainswhyitwasprotantorationalformetotakemyumbrella.Diagrammatically:
1Asanaside:itisworthalsonotingthatthisassumptiondoesnotprecludethepossibilitythatacausalexplanatoryrelationisalsonecessaryforareasontoexplainanactionintherightway:inparticular,itmightbethat(atleastwhenitcomestoaction)acausalexplanatoryrelationbetweenareasonandanactionisanecessary(butnotsufficient)conditionforamysteryrelation.Inwhichcaseexplainingwhysomeonedidwhattheydid intherightway involvesbothacausalexplanatoryrelationandamysteryrelation.
219
FigureXVI-1:Theexplanatoryrelationsinvolvedwhenitwasraining
So, given the transitivity of themystery relationwith the E’-relation, the following facts all
(partially)explainwhyitwasprotantorationalformetotakemyumbrella:thefactthatitwas
raining, the fact that itappeared tomeas though itwas rainingand the fact that Ibelieved
thatitwasraining.And,giventhetransitivityofthemysteryrelation,thesesamefactsareall
alsomysteriouslyrelatedtothefactthatItookmyumbrella.
1.2 ThereasonstherewereandthereasonsIhadtotakemyumbrella
Accordingtoexplanatoryrationalism,explainingwhyitisprotantorationalforsomeonetodo
someactionisbothnecessaryandsufficientforbeingareasonforthemtoactandforbeinga
reasonthattheyhavetoact.2So,thefactthatitwasraining,thefactthatitappearedtomeas
thoughitwasrainingandthefactthatIbelievedthatitwasrainingwereallreasonsformeto
takemyumbrellaandreasonsthatIhadtotakemyumbrella.
1.3 Thereasonsfortakingmyumbrella
Accordingtoexplanatoryrationalism,theconditionsforbeingareasonfor(oragainst)doing
something are different from the conditions for being a reason to (or not to) do it. In
particular,explanatoryrationalismmaintainsthatareasonfordoingsomethingmustbothbe
areasontodoitandmakeit,insomerespect,worthdoing.
Thefactthatitisrainingisareasonformetotakemyumbrellaanditmakestakingit,insome
respect,worth doing, so it is a reason formy takingmyumbrella.However, the fact that it
appearedtomeasthoughitwasraining,andthefactthatIbelievedthatitwasrainingdonot
make takingmyumbrella, in any respect,worthdoing,3so, despite being reasons forme to
takemyumbrella,theyaren’treasonsfortakingmyumbrella.
2Notethatsince,accordingtoexplanatoryrationalism,theconditionsforbeingareasonthereistoactarethesameasbeingareasonthatonehastoact,explanatoryrationalismtakesthesetwoexpressionstobecoextensive–thatis,topickoutthesamekindofreason.Thusexplanatoryrationalismentails(F8)(reasonsagentshavetoactareareasonsforthemtoact).3Recallthediscussionofthispointin§(I)4.2inparticular.
Itappearedtomeasthoughitwasraining
Itwasraining
Ibelievedthatitwasrainingexplains
Itookmyumbrella
explains explains
⇝ ⇝ ⇝
Itappearedtomeasthoughitwasraining
Itwasraining
Ibelievedthatitwasrainingexplains
Takingmyumbrellawasprotantorationalexplains explains
⇝ ⇝ E’
220
Ifyouareconcernedthatitsoundsoddtosaythatsomethingisareasontodosomethingbut
notareasonfordoingit,pleaserefertothediscussionof§A.3oftheAppendixtothischapter.
1.4 Myreasonfortakingmyumbrella
Finally,explanatoryrationalismholdsthatthereasonforwhichanagentactsisanythingthat
bothexplainswhyitwasprotantorationalforthemtoact4andexplainswhytheyacted(inthe
rightway),whichistosaythatanyreasonforanagenttoactthatexplainswhytheyacted(in
therightway)istheirreasonforacting.5
Ihaveassumedthatthemysteryrelationissufficientforexplainingwhysomeoneactedinthe
rightway.Thismeansthatanyreasonforsomeagenttodosomeactionthatismysteriously
relatedtothefactthattheydiditisamongsttheirreasonsfordoingit.
Ihavenotedthatthefollowingfactswereallbothreasonsformetotakemyumbrellaandare
mysteriously related to the fact that I took it: the fact that it was raining, the fact that it
appearedtomeasthoughitwasraining,andthefactthatIbelievedthatitwasraining.Thus
thesewereallreasonsforwhichItookmyumbrella.
Ihavesaidthatthefactthat itwasrainingwasamongstmyreasonsfortakingmyumbrella.
Consider also that the fact that it was raining was a normative reason for me to take my
umbrella:so,Iactedforanormativereason.Sometheoristsholdthatoneactsforanormative
reasononly if one knows it (as I did in this case); in §A.5 of theAppendix to this chapter I
considerhowexplanatoryrationalismcouldexplainwhythatshouldbeso.
1.5 Asummary
TableXVI-2providesa summaryofwhatexplanatory realismhas to sayaboutwhich reason
expressionsapplytowhichfactinthisexample.
4Animplicationofthisformulationisthatanagentcoulddosomethingforareasonthatwasprotantobutnot all things considered rational for them todo. I discuss this implication further in §A.4of theAppendixtothischapter.5 It is worth noting that the formulation of explanatory rationalism thus entails the prima faciereasonableclaims(F7)and(F9)(theclaimsthatconnectactingforareasonwiththerebeingareasontoact)aswellas(D2)(theclaimthattheagent’sreasonforactingalwaysexplainstheiraction).
221
Theputativereasons
Areasonformetotakemyumbrella
Areasonformy
takingmyumbrella
AreasonIhadtotake
myumbrella
AreasonforwhichItookmy
umbrella
Itwasraining ü ü ü ü
It appeared tome as thoughitwasraining ü û ü ü
Ibelievedthatitwasraining ü û ü ü
TableXVI-2:Anapplicationofexplanatoryrationalism
The facts considered above do not exhaust the reasons that there are in this example,6
howevertheyshouldhopefullybeindicativeofthesortsofclaimsthatexplanatoryrationalism
makes. For reference, I provide the same table for several of the examples considered in
previouschaptersin§A.1oftheAppendixtothischapter.
2 Solvingtheproblems
Explanatory rationalism solves the problems considered in §§(II)-(VI) by rejecting the
problematic theses that gave rise to them. That is, explanatory rationalism rejects all of the
followingclaims:
(FAV) Foranyp,p isareasonforAtoφonly ifA’sφing, is insomerespect,worthdoing.
(PSY) Foranyp,ifAφsforthereasonthatpthenpisafeatureofA’spsychology.
(DEL1) Foranyp,ifpisaconsiderationinlightofwhichAφsthenpisA’sreasonforφing.
(DEL2) Foranyp,ifpisA’sreasonforφingthenpisaconsiderationinlightofwhichAφs.
Manywill find explanatory rationalism’s rejectionof these viewsunpalatable. Indeed, itwill
seemtomanythatatleastsomeoftheseclaimsarethemselvesprimafaciereasonable.Ido
nottakethattobethecase–ratherIthinkthatanyresistancetorejectingtheseclaimscomes
fromone’stheoreticalcommitmenttothem,andnotfromtheirinherentplausibility.
Showing that explanatory rationalism can solve the problems considered in §§(II)-(VI) is, of
course,notademonstrationthatit isconsistentwithalloftheprimafaciereasonableclaims
set out in those chapters. Indeed, it is possible that there are otherprima facie reasonable
6Forinstance,thefactthatmyumbrellawouldkeepmedryisalsoareasonformetotakeit(giventhatIknowthatitwould),andareasonforwhichItookit.
222
claimsouttherethat,whencombinedwiththosethatIhaveconsidered,createnewproblems
thatarespecifictoexplanatoryrationalism.
Against that possibility I have two responses: firstly, it is for the sceptic to generate such
claims,notformetoprovethatnoneexist(howcouldIprovethat?).Secondly,andperhaps
more compellingly, explanatory rationalism is to some extent inoculated from the
conventionalformoftheseproblemsbecauseitacknowledgesboththatagentsalwaysactfor
psychological reasons, and that agents also can, and often do, act for normative reasons.
Becauseitcanreconciletheseseeminglycompetingtheses, it is immunefromproblemsthat
arisefromdenyingeither,and,forthatreason, Isuggestthat it isthebestavailableunivocal
theoryofreasons.
3 Newpluralism
Explanatoryrationalismmaybethebestavailableunivocaltheoryofreasons,however,Ithink
that our theory of reasons ought to be pluralist. Recall that in §(V) I introduced the ‘two
senses’intuition:IsuggestedthatthereisasenseinwhichSallyranforareasonandasensein
whichshedidn’t;similarly,IsuggestedthatifIbelievethatthereismilkathomeeventhough
there isn’t, then there is a sense inwhich I don’t have a reason to buymilk and a sense in
which I do. In light of this observation I suggest thatwe should not stopwith the univocal
accountofreasonsprovidedbyexplanatoryrationalismonitsown;instead,weoughttoadopt
apluralisttheoryofreasons.
The sense in which Sally runs for a reason is the sense in which running intentionally,
deliberatelyandpurposefully is running fora reason. Likewise, the sense inwhich Ihaveno
reasontobuymilkisthesenseinwhich,ifitisnotrationalformetodosomething,Ihaveno
reasontodoit.OverthepreviouschaptersIhavearguedthatthebestwaytocharacterisethe
senseoftheseexpressionsforwhichtheseclaimsaretrueisexplanatoryrationalism.
Incontrast,thesenseinwhichSallydoesnotrunforareasonisthesenseinwhichthereisno
reasonforher torunbecausenogoodwillcomefromherrunning; it is innorespectworth
doing.Likewise,thesenseinwhichIhaveareasontobuymilkisthesenseinwhichonehasa
reasontodosomethingbecausedoingthatthing is, insomerespect,worthdoing.Thus,the
senseinwhichSallydoesnotrunforareason,andthesenseinwhichIhaveareasontobuy
milkischaracterisedbythefavouristfamilyofclaimsaboutreasons.
So,totheextentthatonesharesthe‘twosenses’ intuition, Isuggestthatthebesttheoryof
reasonstakesonesenseofeachreasonexpressiontobeexplanatoryrationalist,andtheother
sensetobefavourist.Thisisthenewpluralisttheoryofreasons.
223
ReasonexpressionSenseA
(Explanatoryrationalism)
SenseB
(Favourism)
Foranyp,pisareasonforAtoφ…
…ifandonlyifpexplainswhyitisprotantorationalforAtoφ.
…ifandonlyifpmakesA’sφing,insomerespect,worthdoing.
Foranyp,pisareasonforA’sφing…
…ifandonlyifpexplainswhyitisprotantorationalforAtoφandpmakesA’sφing,insomerespect,worthdoing.
…ifandonlyifpmakesA’sφing,insomerespect,worthdoing.
Foranyp,pisareasonAhastoφ…
…ifandonlyifpexplainswhyitisprotantorationalforAtoφ.
…ifandonlyifpmakesA’sφing,insomerespect,worthdoing.
Foranyp,pisaA’sreasonforφing…
…ifandonlyifpexplainswhyitisprotantorationalforAtoφandexplains(intherightway)whyAφ’d.
…ifandonlyifpmakesA’sφing,allthingsconsidered,worthdoingandexplains(intherightway)whyAφ’d.
TableXVI-3:NewPluralism
While I think that new pluralism is the best way to capture our myriad intuitions about
reasons, I think that explanatory rationalism is the de facto sense of reason expressions in
ordinarylanguage.ItseemstometobemoreofastraintoinsistthatIhaveareasontobuy
milkevenifIbelievethatIhaveplenty,thantoinsistthatSallyhasareasontorunalthoughno
bearischasingher.Indeed,Ifeelmoreinclinedtoqualifytheformer–Imightsaythatthereis
asenseinwhichIhaveareasontobuymilk–andIamsoinclined,Ithink,becausethesensein
whichIhaveareasontobuymilkisnottheconventionalsenseofwhatitistohaveareason
(mutatismutandisforSally).7
4 Thechallengeforpluralism
Recallthefollowingprimafaciereasonableclaim:
(S1) Whenever we give an agent’s reason for acting,whatever the sense of the
expression used, we explain their action in a way that makes them seem
rational.
In §(V)7, I noted that (S1) creates a problem for pluralism. The problemwas as follows: a
theoryofreasonsthatispluralistwithrespecttothereasonsforwhichanagentacts,takesthe
‘agent’sreasonforacting’expressiontohavetwodifferentsenses.Inparticular,suchatheory
holds that the reason-relation for each sense (i.e. the relation between the action and the
7Ofcourse,thegreatvirtueofexplanatoryrationalismisthatthereareoccasionsonwhichsomethingisa reason that I have todo something inboth the favourist and theexplanatory rationalist sense. Forinstance,whenIknowthatitisraining,thefactthatitisrainingisareasonformetotakemyumbrellainboththefavouristandtheexplanatoryrationalistsense.
224
reasoninvirtueofwhichitisareason)isdifferent.Thechallengeforpluralisttheoriesisthus
to explain how it is that, if the reason-relations are different for the different senses, it is
nonetheless thecase thatwhenwegiveanagent’s reason foracting,whatever the senseof
theexpressionused,weexplaintheiractioninawaythatmakesthemseemrational.Thatis,if
theexpression‘theagent’sreasonforacting’hastwodifferentsenses,howis(S1)true?
New pluralism’s answer to this question relies on the following observation: whenever an
agentactsforareasoninthefavouristsense,theyactforthesamereasonintheexplanatory
rationalistsense.Toseewhythisisso,recallwhatfavourismaboutthereasonsforwhichwe
actclaims:
- Favourismaboutthereasonsforwhichweact:Foranyp,p isaA’sreasonforφingifandonly ifpmakesA’sφing,all things considered,worthdoingandexplains (in therightway)whyAφ’d.8
Ifsomethingmakesanagent’sactionallthingsconsideredworthdoingthenitalsomakesitin
somerespectworthdoing,so it isanormativereasonforthemtodo it. Ihavesuggested, in
previouschapters,thatanormativereasontodosomeactionexplainswhattheagentdidin
therightwayonly if it ismysteriouslyrelatedtoabeliefthat, inturn,bothexplainswhythe
agent’s action is pro tanto rational and is mysteriously related to the fact that they
(intentionally)didit.Nowrecallthatabeliefthatexplainsbothwhyitisprotantorationalfor
some agent to do something and why they did it is the agent’s reason for acting in the
explanatoryrationalistsense.
Thus,anormativereason,p,istheagent’sreasonforactinginthefavouristsenseonlyifitis
mysteriouslyrelatedtothefactthattheagentbelievesthatp,wherethefactthattheagent
believes thatp is theagent’s reason foracting in theexplanatory rationalist sense.But, as I
haveargued,ifpismysteriouslyrelatedtothefactthattheagentbelievesthatp,andthefact
thattheagentbelievesthatp is, intheexplanatoryrationalistsense, their reasonforacting,
then the fact thatp isalso their reason foracting in theexplanatory rationalist sense.Thus,
wheneveranagentactsforareasoninthefavouristsense,theyactforthesamereasoninthe
explanatoryrationalistsense.
8Whilethisthesisisrarely, ifever,explicitlyadvocatedbyfavourists,Itakeittobeatleastimpliedbytheir views. For instance,many favouristshold thatanagentacts foranormative reasononly if theyknowit (seethediscussionsofboth§(X)4and§A.5oftheAppendixtothischapterforsomerelateddiscussionoftheseaccounts).Now,whetherornottheyacceptmyindirecttheoryofnormativereasonexplanation,suchfavouristsstillwanttoholdthatanormativereasonmustexplaintheagent’sactionintherightway if it istoexplaintheiraction–andthat isallthat isrequiredforthemtoagreewithmycharacterisation.
225
And when an agent acts for a reason in the explanatory rationalist sense, their reason for
actingexplainsbothwhytheyactedandwhyitwasprotantorationalforthemtoact.Sothisis
whygivinganagent’sreasonforactingalwaysexplainstheiractioninawaythatmakesthem
seemrational (orat leastprotanto rational):becausetheirreasonforactingalwaysexplains
whytheydidwhattheydidandwhyitwas(atleastprotanto)rationalforthemtodoit.
Newpluralismthusmeetsthechallengethat(S1)createsforallpluralisttheoriesbyinsisting
that whenever something is an agent’s reason for acting, it is a reason for acting in the
explanatory rationalist sense, but sometimes it is also the agent’s reason for acting in the
favourist sense. New pluralism thus provides a plausible account of the truth of (S1)whilst
maintainingitspluralistcredentials.
5 Conclusion
Explanatory rationalism provides us with a univocal account of what it is to be a practical
reasonthatdoesnotsufferthefailingsofmostcontemporarytheoriesofreasons.Thegreat
virtueofexplanatoryrationalism is thusthat itdoesnotgenerallycompelus tomakeclaims
about reasons that are strange, counterintuitive or prima facie paradoxical, unlike most
contemporary theories of reasons. In this respect, it is, I suggest, superior to favourism,
deliberativism,andpsychologism.
Moreover, I have suggested that new pluralism, a theory that combines explanatory
rationalism and favourism, is the best theory of reasons; it combines all the virtues of
explanatoryrationalismwithawaytosatisfy the ‘twosenses’ intuition,and ithappilymeets
themajorchallengetopluralisttheories.
226
Appendix
A.1 Somemoreexamples
Forreference,TableXVI-4showswhatexplanatoryrationalismhastosayaboutwhichreason
expressions apply towhich facts/propositions in someof theexamples considered in earlier
chapters.
Example Theputativereasons AreasonforAtoφ
AreasonforA’sφing
AreasonAhadtoφ
A’sreasonforφing
Award
Myfriendwonanaward ü ü ü ü
Ireadthatshehadwonanawardinthenewspaper ü û ü ü
Ibelievedthatshehadwonanaward ü û ü ü
Sally
AbearwaschasingSally û û û û
Sheheardabear-likesound ü û ü ü
Shebelievedthatabearwaschasingher ü û ü ü
Eva
Herhusbandwasunfaithful û û û û
It appeared to her as thoughher husband was kissinganotherwoman
û û û û
She believed that he had beenunfaithful ü û ü ü
Climber
Loosening his grip would ridhimofdanger ü ü ü û
It appeared as thoughlooseninghisgripwouldridhimofdanger
ü û ü û
He believed that loosening hisgripwouldridhimofdanger ü û ü û
TableXVI-4:Otherexamplesforexplanatoryrationalism
A.2 Explainingintherightway
Throughoutthischapter,Iassumedthatifsomefactismysteriouslyrelatedtothefactthatthe
agentdidwhattheydid,thenitexplainsitintherightway.Thissectionprovidesmorecontext
tothatassumption.
227
Accordingtoexplanatoryrationalism,somethingisthereasonforwhichanagentdoessome
actiononlyifitisareasonforthemtodoit9anditexplainswhytheydiditintherightway.But
whatdoesitmeantosaythatitexplainsitintherightway?
Well,recallthecaseoftheclimberwhowassounnervedbyhisresolutiontodrophisfriend,
that he loosened his grip on his friend unintentionally (i.e. not for a reason).10According to
explanatoryrationalismtherewasareasonforhimtoloosenhisgrip,whichwas,interalia,the
factthathebelievedthatdoingsowouldridhimofdanger.11Moreover,giventhat itcauses
himtoloosenhisgrip,thereisperhapsa(causal)senseinwhichthefactthathebelievedthat
looseninghisgripwouldridhimofdangerexplainswhyheloosenedhisgrip.Thatbeingso,a
reasonforhimtoloosenhisgripexplainswhyheloosenedhisgrip,butitwasnonethelessnot
hisreasonforlooseninghisgrip.12Whynot?Becauseitdoesnotexplainwhyheloosenedhis
gripintherightway.13
Whatisthe‘rightway’ofexplaining?AsIsuggestedin§(XIV),therightwayofexplainingan
agent’sactionisnotmerelycausal.Toelaborate:whiletherightwaymightbepartlycausal(I
willtakenoviewonthat), it isnotmerelycausal.Inparticular,Ihavesuggestedthatifafact
explainswhyanagentactedintherightwaythenitismysteriouslyrelatedtothefactthatthe
agentactedastheydid.14
Whattheseremarksestablishisthatamysteryrelationbetweenareasonthereistoactand
thefactthattheagentactedastheydidisnecessaryforthereasontoexplaintheiractionin
the rightway. For the purpose of this discussion, I assumed that it is sufficient; so that if a
reasonthatanagenthastoactismysteriouslyrelatedtothefactthattheyactedastheydid,
then it is their reason for acting. This assumption does not have a significant impact onmy
argument;itjustmakestheexpositionlessinvolved.
9I.e.itexplainswhyitisprotantorationalforthemtodoit.10See§(XIV)4.11Itwasareasonforhimtoloosenhisgripbecauseitexplainswhyitwasprotantorationalforhimtoloosenhisgrip.12Thisisclearlyimpliedbythefactthathedoesnotloosenhisgripforareason.13Cf. ‘What distinguishes actionswhich are intentional from thosewhich are not? The answer that Ishall suggest is that they are the actions to which a certain sense of the question ‘Why?’ is givenapplication; the sense is of course that in which the answer, if positive, gives a reason for acting.’(Anscombe1957,§5)14Thatis,Ihavesuggestedthatthedifferencebetweenafactthatmerelycausallyexplainswhyanagentactedandonethatexplainswhytheyactedintherightway isthatitisonlyinthelattercasethatthefactismysteriouslyrelatedtothefactthattheagentdidwhattheydid.
228
A.3 Theproblemofreasonsforacting
Unlike any other theory of reasons (tomy knowledge) explanatory rationalismdistinguishes
betweenthekindofreasonpickedoutbytheexpressions‘areasontoact’and‘areasonfor
acting’.Itinsiststhatwhileareasontoactisjustanythingthatexplainswhyanagent’saction
isprotantorational,areasonforactingissomethingthatisbothareasontoactandcountsin
favouroftheagent’sdoingit.Aconsequenceofthisviewisthatsomethingcouldbeareason
foranagenttodosomeactionwithoutbeingareasonfortheirdoingit.
While thismay sound paradoxical, inwhat follows Iwill argue that this is actually the least
worstresponsetoaproblemthataffectsalltheoriesofreasons.
A.3.1 WhatreasontherewasforSallytorun
Totheextentthatweacceptthat it isprimafacie reasonabletoclaimthatSally’sreasonfor
runningisthatsheheardabear-likesound(i.e.(F6))andthatanagent’sreasonforactingmust
beareasontherewasforthemtosoact(i.e.(F9)), it followsthatthefactthatSallyhearda
bear-like sound (in a wood that she knew to contain bears) was a reason for her to run.
Indeed,itseemstome,ifyouhearsomethingthatsoundslikeabearinawoodthatyouknow
tocontainbears,thenthatisagoodreasonforyoutorun.Thus,Isuggest:
(R5) ThefactthatSallyheardabear-likesoundisareasonforhertorun.
A.3.2 Reasonstoactandreasonsforacting
AsIhavealreadynoted, itseemsreasonabletosaythat ifsomethingisareasontodosome
actionthenitisareasonfordoingit,thus:
(R6) Foranyp,ifpisareasonforAtoφthenpisareasonforA’sφing.
A.3.3 Reasonsforactingcountinfavourofactions
In§(I)1.3,Inotedthatallreasonsforactingcountinfavouroftheactionsforwhichtheyare
reasonsandthatallreasonsagainstsomeactioncountagainstdoingthataction.In§(I)4.4,I
argued further that a reason for some action counts in favour of that action in virtue of its
beingareasonforit;thatis,Iarguedthatthe‘for’prepositiononitsownisenoughtogiveit
thatmeaning.15Thus:
(R7) Foranyp,ifpisareasonforA’sφingthenpcountsinfavourofA’sφing.
15In§(I)4.4,Iarguedfurtherthatareasonforsomeactioncountsinfavourofthatactioninvirtueofitsbeing a reason for it; that is, I argued that the ‘for’ preposition on its own is enough to give it thatmeaning–Ileftitopenwhetherornotareasonforactingalsocountsinfavourofanactioninvirtueofitsbeingareason.Similarremarksapplytoreasonsagainstsomeaction.
229
A.3.4 Countinginfavourof
AsIhavealreadyargued,16themostnatural interpretationofwhatit istocountinfavourof
someactionistomakeit,insomerespect,worthdoing.17Thus:
(R8) For anyp, p counts in favour ofA’sφing if and only ifpmakes A’sφing, in
somerespect,worthdoing.
A.3.5 Whatmake’sSally’srunningworthdoing
As I established in §(II)4, Sally’s running is, in no respect,worth doing, so nothingmakes it
worthdoing.And,inparticular:
(R9) ThefactthatSallyheardabear-likesounddoesnotmakeherrunning,inany
respect,worthdoing.
A.3.6 TheReasonsforActingProblem
TheReasonsforActingProblemisthis:thefactthatSallyheardabear-likesoundisareason
forhertorun,soitisareasonforherrunning,soitcountsinfavourofherrunning,soitmakes
her running, in some respect, worth doing, but it doesn’tmake her running in any respect
worthdoing!Explicitly,thefollowingclaimsaremutuallyinconsistent:
(R5) ThefactthatSallyheardabear-likesoundisareasonforhertorun.
(R6) Foranyp,ifpisareasonforAtoφthenpisareasonforA’sφing.
(R7) Foranyp,ifpisareasonforA’sφingthenpcountsinfavourofA’sφing.
(R8) For anyp, p counts in favour ofA’sφing if and only ifpmakes A’sφing, in
somerespect,worthdoing.
(R9) ThefactthatSallyheardabear-likesounddoesnotmakeherrunning,inany
respect,worthdoing.
Thisisaproblemthatalltheoriesofreasonsface(notehowalloftheseclaimsareprimafacie
reasonable, although they are mutually inconsistent). The response of favourists and
deliberativiststothisproblemistoreject(R5).Psychologismaboutreasonsforactingrejects
(R7).18Kearns and Star (2008) interpret what it is to ‘count in favour of acting’ as ‘being
evidence thatoneought to soact,’ so theywouldpresumably reject (R8). I knowofnoone
16Recallthediscussionofthispointin§(I)4.2inparticular.17Where what it is for an action to be worth doing is to be determined, except for the claim thatwhetherornotsomeactionisworthdoingforsomeagentisindependentofantheircognitivestates.18Psychologistswouldprobablyalsoreject(R5).
230
who would reject (R9). For reasons that I have set out at various points over the previous
chapters,Idon’tfindanyoftheseoptionspalatable.
Incontrast,explanatoryrationalismrejects(R6).Itakethistobetheleastworstoftheoptions
available.Does it soundoddtosay thatsomethingcouldbea reasontoactwithoutbeinga
reasonforacting?Somewhat.However,Ithinkthattheoddnessofsayingthisdissipateswhen
particular examples are considered: it does not soundodd tome to say that the fact that I
believethatitisrainingisareasonformetotakeanumbrellabutnotareasonfortakingan
umbrella.
A.4 Actingprotantorationally
According toexplanatory rationalism, it ispossible thatanagentcoulddosomeaction fora
reasoneven though thatactionwasmerelypro tanto rational,andnotall thingsconsidered
rational. That is, explanatory rationalism suggests that evenwhen an agent does one thing,
despite judging somethingelse tobeall things consideredworthdoing instead, they still do
whattheydoforareason.Isthisright?
Well,considerthefollowingexample:
Amanwalkinginaparkstumblesonabranchinthepath.Thinkingthebranchmayendangerothers,hepicks itupandthrowsit inahedgebesidethepath.Onhiswayhomeitoccurstohim that the branch may be projecting from the hedge and so still be a threat to unwarywalkers. He gets off the tram he is on, returns to the park, and restores the branch to itsoriginalposition…It iseasyto imaginethatthemanwhoreturnedtotheparktorestorethebranchtoitsoriginalpositioninthepathrealizesthathisactionisnotsensible.Hehasamotiveformovingthestick,namely,thatitmayendangerapasser-by.Buthealsohasamotivefornotreturning,whichisthetimeandtroubleitcosts.Inhisownjudgement,thelatterconsiderationoutweighs the former; yet he acts on the former. In short, he goes against his own bestjudgement.(Davidson2004,172&174)
ThemaninDavidson’sexampledoesactforareason:hethinksthatthestickmayendangera
passer-by. It seemstomethat the fact that theaction is ‘less than fully rational’,19doesnot
impingeupon itshavingbeendone fora reason.20But thismeans thatweshouldbecareful
aboutsayingthatifoneactsforareasonthentheyactrationally.HereiswhatIsuggest:
- Anagentactsprotantorationallyifandonlyiftheirreasonfordoingitisaprotantoreasontodoit(i.e.somethingthatexplainswhytheiractionisprotantorational).
19TouseParfit’s(2011,34)term.20Othersarealsoofthisview,for instance: ‘Theincontinentmanholdsonecoursetobebetter(forareason)andyetdoessomethingelse(alsoforareason).’(Davidson2001b,34);‘Akrasia,orweaknessofthewill, occurswhen, in the face of conflicting reasons for and againstX-ing someonemakes an all-things-consideredjudgementthatheoughtnottoX,butX’sanywayanddoessoforareason,namely,for whatever the reason in favour of X-ing was (which was included in the basis of the all-things-consideredjudgement).’(Hurley1992,130)
231
- Anagentactsallthingsconsideredrationallyifandonlyiftheirreasonfordoingitisanallthingsconsideredreasontodoit(i.e.somethingthatexplainswhytheiractionisallthingsconsideredrational).
A.5 Actingforanormativereasonandknowledge
Several theorists hold that an agent acts for a normative reason only if they know it.21In
§(XIV)2, Iarguedthatamysteryrelationbetweenthe fact thatpandthe fact thatanagent
believesthatp isanecessaryconditionontheagent’sknowingthatp.Now,supposefurther
thatitisalsoasufficientcondition(i.e.ifthefactthatpismysteriouslyrelatedtothefactthat
anagentbelievesthatpthentheagentknowsthatp).
If that were true, then explanatory rationalism would provide us with an account ofwhy
knowinganormativereasonshouldbenecessaryforactingonit:becauseitisonlyifanagent
knowsanormative reason that it canmysteriously explain their action– and it is only their
reasonforactingifitmysteriouslyexplainstheiraction.Letmeelaborate.
Anormativereasontodosomeactionmustbemysteriouslyrelatedtotheagent’sbeliefinitif
it istoexplaineitherwhy itwasprotanto rational fortheagenttodothataction,or(inthe
right way) why they did it 22 ; that is, a normative reason to do some action must be
mysteriouslyrelatedtotheagent’sbeliefinitifitistobetheagent’sreasonforacting.Andif
wesupposethatthemysteryrelationissufficientforknowledge,thenanormativereasonwill
onlybemysteriouslyrelatedtoanagent’sbeliefinitiftheagentknowsthatnormativereason.
So this iswhyanagent canonly act for anormative reason if they know it: because if they
don’t know it then, interalia, itwon’texplainwhy theydid it in the rightway, and itwon’t
explainwhyitwasprotantorationalforthemtodoit.
21(E.g.Unger1978;Hyman1999,2015;Hornsby2008;McDowell2013)22Excludingweirdcases(e.g.whereanormativereasontodosomeactionalsohappenstobeafeatureoftheagent’spsychology,whichanywayexplainswhytheagent’sactionisprotantorational).
232
BibliographyAlvarez,Maria.2010.KindsofReasons.OxfordUniversityPress.
———.2013. ‘ExplainingActionsandExplainingBodilyMovements’. InReasonsandCauses:CausalismandAnti-CausalisminthePhilosophyofAction,editedbyGiuseppinaD’Oro,141–59. History of Analytic Philosophy. Houndmills, Basingstoke, Hampshire ; NewYork:PalgraveMacmiilan.
———. 2016a. ‘Reasons for Action: Justification, Motivation, Explanation’. In The StanfordEncyclopedia of Philosophy, edited by Edward N. Zalta, Summer 2016.http://plato.stanford.edu/archives/sum2016/entries/reasons-just-vs-expl/.
———.2016b.‘ReasonsforAction,ActingforReasons,andRationality’.Synthese,January,1–18.
Anscombe,G.E.M.1957.Intention.HarvardUniversityPress.
Audi,Paul. 2012a. ‘AClarificationandDefenseof theNotionofGrounding’. InMetaphysicalGrounding, edited by Fabrice Correia and Benjamin Schnieder, 101–21. Cambridge:CambridgeUniversityPress.
———. 2012b. ‘Grounding: Toward a Theory of the in-Virtue-of Relation’. The Journal ofPhilosophy109(12):685–711.
Audi, Robert. 2001.The Architecture of Reason: The Structure and Substance of Rationality.Oxford ;NewYork:OxfordUniversityPress.
Broome, John. 2006. ‘Reasons’. InReason and Value: Themes from theMoral Philosophy ofJoseph Raz, edited by R. Jay Wallace, Philip Pettit, Samuel Scheffler, and MichaelSmith,29–55.OxfordUniversityPressonDemand.
———. 2013.Rationality throughReasoning. The Blackwell/Brown Lectures in Philosophy 4.Chichester,WestSussex ;Malden,MA:WileyBlackwell.
Chudnoff,Elijah.2011.‘WhatShouldaTheoryofKnowledgeDo?’Dialectica65(4):561–79.
Collins,ArthurW.1997.‘ThePsychologicalRealityofReasons’.Ratio10(2):108–23.
Comesaña, Juan, andMatthewMcGrath. 2014. ‘Having False Reasons’. In Epistemic Norms,editedbyClaytonLittlejohnandJohnTurri,59–79.OxfordUniversityPress.
Correia,Fabrice.2010.‘GroundingandTruth-Functions’.LogiqueetAnalyse53(211):251–279.
Crane,Tim.2000.‘Dualism,Monism,Physicalism’.Mind&Society;Heidelberg1(2):73–85.
Dancy,Jonathan.2000.PracticalReality.NewYork:OxfordUniversityPress.
———.2004.‘TwoWaysofExplainingActions’.InAgencyandAction,editedbyJohnHyman,25–42.Cambridge:CambridgeUniversityPress.
———.2006. ‘Acting intheLightoftheAppearances’. InMcDowellandHisCritics,editedbyCynthiaMacdonaldandGrahamMacdonald,121–34.BlackwellPublishingLtd.
———.2008a. ‘OnHowtoActDisjunctively’. InDisjunctivism,editedbyAdrianHaddockandFionaMacpherson.OxfordUniversityPress.
———.2008b. ‘Arguments from Illusion’. InDisjunctivism,editedbyAlexByrneandHeatherLogue,116–35.TheMITPress.
———.2011.‘ActinginIgnorance’.FrontiersofPhilosophyinChina6(3):345–57.
233
———. 2012. ‘Response to Mark Schroeder’s “Slaves of the Passions”’. Edited by MarkSchroeder.PhilosophicalStudies:AnInternationalJournalforPhilosophyintheAnalyticTradition157(3):455–62.
———.2014.‘OnKnowingOne’sReason’.InEpistemicNorms,editedbyClaytonLittlejohnandJohnTurri,80–96.OxfordUniversityPress.
Darwall, Stephen. 2003. ‘Desires, Reasons, and Causes’. Philosophy and PhenomenologicalResearch67(2):436–443.
Davidson, Donald. 2001a. ‘Actions, Reasons and Causes (1963)’. In Essays on Actions andEvents,2nded,3–21.Oxford :NewYork:ClarendonPress ;OxfordUniversityPress.
———.2001b.‘HowIsWeaknessoftheWillPossible?(1969)’.InEssaysonActionsandEvents,2nded,21–42.Oxford :NewYork:ClarendonPress ;OxfordUniversityPress.
———.2001c. ‘Intending (1978)’. InEssaysonActionsandEvents, 2nded, 83–102.Oxford :NewYork:ClarendonPress ;OxfordUniversityPress.
———. 2001d. ‘A Coherence Theory of Truth and Knowledge (1987)’. In Subjective,Intersubjective,Objective,137–58.OxfordUniversityPress.
———.2004. ‘Paradoxesof Irrationality (1982)’. InProblemsofRationality,169–88.Oxford :NewYork:ClarendonPress ;OxfordUniversityPress.
Davis, Wayne A. 2003. ‘Psychologism and Humeanism’. Philosophy and PhenomenologicalResearch67(2):452–59.
———.2005.‘ReasonsandPsychologicalCauses’.PhilosophicalStudies122(1):51–101.
Everson,Stephen.2009. ‘WhatAreReasonsforAction?’ InNewEssaysontheExplanationofAction, edited by Constantine Sandis, 22–47. Houndmills, Basingstoke, Hampshire;NewYork:PalgraveMacmillan.
FIDE. n.d. ‘Fide Laws of Chess Taking Effect from 1 July 2017’. World Chess Federation.https://www.fide.com/component/handbook/?id=207&view=article.
Fine,Kit.2012. ‘GuidetoGround’. InMetaphysicalGrounding,editedbyFabriceCorreiaandBenjaminSchnieder,37–80.Cambridge:CambridgeUniversityPress.
‘For, Prep. and Conj.’ n.d. OED Online. Oxford University Press.http://www.oed.com/view/Entry/72761.
Garrard,Eve,andDavidMcNaughton.1998. ‘MappingMoralMotivation’.EthicalTheoryandMoralPractice1(1):45–59.
Gettier,EdmundL.1963.‘IsJustifiedTrueBeliefKnowledge?’Analysis23(6):121.
Gibbons,John.2010.‘ThingsThatMakeThingsReasonable’.PhilosophyandPhenomenologicalResearch81(2):335–61.
Gjelsvik, Olav. 2007. ‘Are There Reasons to Be Rational?’ Homage à Wlodek: PhilosophicalPapersDedicatedtoWlodekRabinowicz,142–147.
Goldman,AlvinI.1967.‘ACausalTheoryofKnowing’.TheJournalofPhilosophy64(12):357.
———.2012.ReliabilismandContemporaryEpistemologyEssays.OxfordUniversityPress.
Hall,N.2004.‘TwoConceptsofCausation’.CausationandCounterfactuals,225–276.
Harman, Gilbert H. 1970. ‘Knowledge, Reasons, and Causes’. The Journal of Philosophy 67(21):841–55.
234
Hecht, Stephen S. 2006. ‘Cigarette Smoking: Cancer Risks, Carcinogens, and Mechanisms’.Langenbeck’sArchivesofSurgery391(6):603–13.
Heuer,Ulrike.2004.‘ReasonsforActionsandDesires’.PhilosophicalStudies:AnInternationalJournalforPhilosophyintheAnalyticTradition121(1):43–63.
Hieronymi, Pamela. 2011. ‘XIV—Reasons for Action’. Proceedings of the Aristotelian Society(Hardback)111(3pt3):407–427.
Hofweber, Thomas. 2009. ‘Ambitious, Yet Modest, Metaphysics’. InMetametaphysics: NewEssaysontheFoundationsofOntology,editedbyDavidJohnChalmers,DavidManley,andRyanWasserman,347–83.Oxford ;NewYork:ClarendonPress.
Hornsby,Jennifer.2007.‘KnowledgeinAction’.ActioninContext,285–302.
———. 2008. ‘A Disjunctive Conception of Acting for Reasons’. In Disjunctivism, edited byAdrianHaddockandFionaMacpherson,244–61.OxfordUniversityPress.
Hurley,S.L.1992.NaturalReasons:PersonalityandPolity.OxfordUniversityPress.
Hyman, John. 1999. ‘How Knowledge Works’. The Philosophical Quarterly (1950-) 49(197):433–51.
———.2011.‘ActingforReasons:ReplytoDancy’.FrontiersofPhilosophyinChina6(3):358–68.
———.2015.Action,Knowledge,andWill.OxfordUniversityPress.
Kearns,Stephen,andDanielStar.2008.‘Reasons:ExplanationsorEvidence?’Ethics119(1):31–56.
———.2009.‘ReasonsasEvidence’.InOxfordStudiesinMetaethics,editedbyLandau.Vol.4.
Kim, Jaegwon. 1988. ‘Explanatory Realism, Causal Realism, and Explanatory Exclusion’.MidwestStudiesInPhilosophy12(1):225–39.
———. 1989. ‘Mechanism, Purpose, and Explanatory Exclusion’. Philosophical Perspectives3:77–108.
———. 1993. Supervenience andMind: Selected Philosophical Essays. Cambridge Studies inPhilosophy.NewYork,NY,USA:CambridgeUniversityPress.
———. 2003. ‘Blocking Causal Drainage and Other Maintenance Chores with MentalCausation’.PhilosophyandPhenomenologicalResearch67(1):151–76.
———. 2008. Physicalism, or Something near Enough. 3rd print., And 1st paperback print.PrincetonMonographsinPhilosophy.Princeton:PrincetonUniv.Press.
Korcz, Keith Allen. 2015. ‘The Epistemic Basing Relation’. In The Stanford Encyclopedia ofPhilosophy,editedbyEdwardN.Zalta,Fall2015.MetaphysicsResearchLab,StanfordUniversity.https://plato.stanford.edu/archives/fall2015/entries/basing-epistemic/.
Kroedel, Thomas, and Moritz Schulz. 2016. ‘Grounding Mental Causation’. Synthese 193(6):1909–23.
Lehrer,Keith.1971.‘HowReasonsGiveUsKnowledge,ortheCaseoftheGypsyLawyer’.TheJournalofPhilosophy68(10):311–13.
Leuenberger,Stephan.2014.‘GroundingandNecessity’.Inquiry57(2):151–74.
Lewis,David. 1987. ‘Causal Explanation’. InPhilosophical PapersVolume II, 314–240.OxfordUniversityPress.
235
Littlejohn, Clayton. Forthcomming. ‘How and Why Knowledge Is First’. In Knowledge First,editedbyA.Carter,E.Gordon,andB.Jarvis.OxfordUniversityPress.
———.2012.JustificationandtheTruth-Connection.CambridgeUniversityPress.
Locke, Dustin. 2015. ‘Knowledge, Explanation, and Motivating Reasons’. AmericanPhilosophicalQuarterly52(3).
Lord, Errol. 2008. ‘Dancy on Acting for the Right Reason’. Journal of Ethics and SocialPhilosophy2(3):1–7.
———.2010. ‘HavingReasonsandtheFactoringAccount’.PhilosophicalStudies149(3):283–96.
———. 2015. ‘Acting for the Right Reasons, Abilities, and Obligation’. In Oxford Studies inMetaethics,Volume10,editedbyRussShafer-Landau,26–52.OxfordUniversityPress.
Markovits,J.2010.‘ActingfortheRightReasons’.PhilosophicalReview119(2):201–42.
———.2011. ‘WhyBean InternalistaboutReasons?’ InOxfordStudies inMetaethics,editedbyRussShafer-Landau,256–80.OxfordUniversityPress.
McCain, Kevin. 2012. ‘The Interventionist Account of Causation and the Basing Relation’.PhilosophicalStudies:AnInternationalJournalforPhilosophyintheAnalyticTradition159(3):357–82.
McDonnell,Neil.2015.‘TheDevianceinDeviantCausalChains:TheDevianceinDeviantCausalChains’.Thought:AJournalofPhilosophy4(3):162–70.
McDowell, John. 1982. ‘Criteria, Defeasibility, and Knowledge’. Proceedings of the BritishAcademy68:455–79.
———. 2013. ‘Acting in the Light of a Fact’. In Thinking About Reasons, edited by DavidBakhurst,BradHooker,andMargaretOliviaLittle,13–28.OxfordUniversityPress.
Mele,AlfredR.2007.‘ReasonologyandFalseBeliefs’.PhilosophicalPapers36(1):91–118.
Mill, John Stuart. 1863. Utilitarianism. The Electric Book Company.http://www.myilibrary.com?id=124074.
Mitova,Veli.2015.‘TruthyPsychologismaboutEvidence’.PhilosophicalStudies172(4):1105–26.
———.2016.‘ClearingSpaceforExtremePsychologismaboutReasons’.SouthAfricanJournalofPhilosophy35(3):293–301.
O’Brien,Lilian.2015.‘BeyondPsychologismandAnti-Psychologism’.EthicalTheoryandMoralPractice18(2):281–95.
Oliver,Alex.1996.‘TheMetaphysicsofProperties’.Mind105(417):1–80.
Olson,Jonas,andFransSvensson.2005.‘RegimentingReasons’.Theoria71(3):203–214.
O’Shaughnessy,Brian.1973.‘Trying(astheMental"PinealGland")’.TheJournalofPhilosophy,365–386.
———. 1980. The Will: A Dual Aspect Theory. Cambridge [Eng.] ; New York: CambridgeUniversityPress.
Parfit,Derek. 2001. ‘Rationality andReasons’.ExploringPractical Philosophy: FromAction toValues,17–39.
———.2011.OnWhatMatters:VolumeOne.OxfordUniversityPress.
236
Pautz, Adam. 2010. ‘Why Explain Visual Experience in Terms of Content?’ In Perceiving theWorld,editedbyBenceNanay,254–309.OxfordUniversityPress.
Plantinga,Alvin.1993.Warrant:TheCurrentDebate.OxfordUniversityPress.
Pollock, John L., and Joseph Cruz. 1999. Contemporary Theories of Knowledge. Rowman &Littlefield.
Raz, Joseph.1999a.EngagingReason:OntheTheoryofValueandAction.Oxford;NewYork:OxfordUniversityPress.
———.1999b.PracticalReasonandNorms.OxfordUniversityPress.
———. 2009. ‘Reasons: Explanatory and Normative’. In New Essays on the Explanation ofAction, editedbyConstantine Sandis, 184–202.Houndmills, Basingstoke,Hampshire;NewYork:PalgraveMacmillan.
———.2011.FromNormativitytoResponsibility.OxfordUniversityPress.
Rosen, Gideon. 2010. ‘Metaphysical Dependence: Grounding and Reduction’. InModality:Metaphysics,Logic,andEpistemology,editedbyBobHaleandAvivHoffmann,109–36.OxfordUniversityPress.
Ruben, David-Hillel. 2004. Explaining Explanation. London: Routledge, Taylor & Francis e-Library.
———. 2008. ‘Disjunctive Theories of Perception and Action’. In Disjunctivism: Perception,Action,Knowledge,editedbyAdrianHaddockandFionaMacpherson,227–42.OxfordUniversityPress.
Russell,Bertrand.1917.MysticismandLogic,andOtherEssays.London :G.Allen&Unwin,.
Ryle,Gilbert.1949.TheConceptofMind.London:Hutchinson.
Sandis, Constantine. 2012. The Things We Do and Why We Do Them. London: PalgraveMacmillanUK.
———. 2013. ‘Can Action Explanations Ever Be Non-Factive?’ In Thinking About Reasons,edited by David Bakhurst, Brad Hooker, and Margaret Olivia Little, 29–49. OxfordUniversityPress.
Saporiti,Katia.2007. ‘ASeemingSolutiontoaSeemingPuzzle inExplainingAction’.Action inContext,303.
Scanlon, Thomas. 1998.WhatWe Owe to Each Other. Cambridge, Mass.: Belknap Press ofHarvardUniversityPress.
———.2014.BeingRealisticaboutReasons.OxfordUniversityPress.
Schaffer, Jonathan.2009. ‘OnWhatGroundsWhat’. InMetametaphysics:NewEssayson theFoundations of Ontology, edited by David John Chalmers, David Manley, and RyanWasserman,347–83.Oxford ;NewYork:ClarendonPress.
Schnieder,Benjamin.2011. ‘ALogicfor“Because”’.TheReviewofSymbolicLogic4(03):445–65.
Schroeder,Mark.2007.SlavesofthePassions.Oxford;NewYork:OxfordUniversityPress.
———.2008. ‘HavingReasons’.PhilosophicalStudies:An International Journal forPhilosophyintheAnalyticTradition139(1):57–71.
Setiya,Kieran.2007.ReasonswithoutRationalism.Princeton,N.J.:PrincetonUniversityPress.
Smith,Michael.1987.‘TheHumeanTheoryofMotivation’.Mind96(381):36–61.
237
———.1994.TheMoralProblem.PhilosophicalTheory.Oxford,UK ;Cambridge,Mass.,USA:Blackwell.
———. 1998. ‘The Possibility of Philosophy of Action’. In Human Action, Deliberation andCausation, edited by Jan Bransen and Stefaan Cuypers, 17–41. Kluwer AcademicPublishers.
———. 2004. ‘The Structure of Orthonomy’. InAgency and Action, edited by John Hyman,165–94.Cambridge:CambridgeUniversityPress.
Sosa,Ernest.2015.JudgmentandAgency.OxfordUniversityPress.
Stout,Rowland.1996.ThingsThatHappenBecauseTheyShould:ATeleologicalApproach toAction.OxfordPhilosophicalMonographs.NewYork:OxfordUniversityPress.
———.2009.‘WasSally’sReasonforRunningfromtheBearThatSheThoughtItWasChasingHer?’InNewEssaysontheExplanationofAction,editedbyConstantineSandis,48–61.Houndmills,Basingstoke,Hampshire;NewYork:PalgraveMacmillan.
———.2010. ‘DeviantCausalChains’. InACompaniontothePhilosophyofAction,editedbyTimothyO’ConnorandConstantineSandis,159–65.Oxford,UK:Wiley-Blackwell.
Stoutland,Frederick.1998. ‘TheRealReasons’. InHumanAction,DeliberationandCausation,editedbyJanBransenandStefaanE.Cuypers,43–66.PhilosophicalStudiesSeries77.SpringerNetherlands.
———.2007.‘ReasonsforActionandPsychologicalStates’.ActioninContext,75.
Turri,John.2009.‘TheOntologyofEpistemicReasons’.Noûs43(3):490–512.
Unger,Peter.1978.Ignorance:ACaseforScepticism.OxfordUniversityPress.
Vogelstein,Eric.2012.‘SubjectiveReasons’.EthicalTheoryandMoralPractice15(2):239–57.
Way,Jonathan,andDanielWhiting.2016.‘ReasonsandGuidance(Or,SurprisePartiesandIceCream)’.AnalyticPhilosophy57(3):214–35.
———. 2017. ‘Perspectivism and the Argument from Guidance’. Ethical Theory and MoralPractice20(2):361–74.
Wedgwood,Ralph.2002.‘InternalismExplained*’.PhilosophyandPhenomenologicalResearch65(2):349–69.
Whiting, Daniel. 2014. ‘Keep Things in Perspective: Reasons, Rationality and the A Priori’.JournalofEthicsandSocialPhilosophy8:i.
Williams,Bernard.1981. ‘InternalandExternalReasons’. InMoralLuck:PhilosophicalPapers1973-1980,101–13.CambridgeUniversityPress.
Yablo,Stephen.2008.‘WideCausation’.InThoughts:PapersonMind,Meaning,andModality,275–306.OxfordUniversityPress.