Mind Association

There is No Question of Physicalism

Author(s): Tim Crane and D. H. Mellor
Source: Mind, New Series, Vol. 99, No. 394 (Apr., 1990), pp. 185-206
Published by: Oxford University Press on behalf of the Mind Association
Stable URL: http://www.jstor.org/stable/2254959 .
Accessed: 22/09/2014 06:42

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Oxford University Press and Mind Association are collaborating with JSTOR to digitize, preserve and extend
access to Mind.

http://www.jstor.org

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No QuestionofPhysicalism
TIM CRANE and D. H. MELLOR

i. Whatis physicalism?
Many philosophers are impressedby the progressachievedby physical
sciences.This hashadan especially deepeffect on theirontologicalviews:it
has mademanyofthemphysicalists. Physicalists
believethateverything is
physical:moreprecisely, thatall entities, properties, and factsare
relations,
thosewhicharestudiedbyphysics orotherphysical sciences.Theymaynot
allagreewiththespiritofRutherford's quotedremark that'thereis physics;
andthereis stamp-collecting',' buttheyall grantphysical sciencea unique
ontologicalauthority: theauthority to tellus whatthereis.
Physicalism is now almostorthodoxin muchphilosophy, notablyin
muchrecentphilosophy ofmind.But althoughofteninvoked,it is rarely
defined.
explicitly It shouldbe. The claimthateverything is physicalis not
as clearas it seems.In thispaper,we examinea numberof proposed
definitionsof physicalism and reasonsforbeinga physicalist. We will
argueboththatphysicalism lacksa clearand credibledefinition, and that
in no non-vacuous interpretation is it true.
We are concernedhereonlywithphysicalism as a doctrineaboutthe
empiricalworld.In particular, itshouldnotbe confusedwithnominalism,
thedoctrine thatthereareno universals.2 Nominalism andphysicalism are
quiteindependent doctrines. Believersin universals mayas consistently
assertas denythattheonlyproperties and relationsare thosestudiedby
physicalscience.And nominalists maywithequal consistency assertor
denythatphysicalsciencecouldprovideenoughpredicates todescribethe
world.That is the questionwhichconcernsphysicalists, not whether
physicalpredicates namerealuniversals. (We willforbrevity writeas if
theydo, butwe do notneedthatassumption.)
As we will understand it, then,physicalism is not a doctrineabout
universalsor otherabstractobjects,but abouttheempiricalworld,and
aboutminds.It saysthatmentalentities,
specifically properties,relations
and factsare all reallyphysical.The mentalis physicalism's chieftarget;
butone we thinkit does nothit.
Physicalism is a kind of monism,opposingthe dualist'sdistinction
betweentwokindsofsubstance: matter andmind.As such,itis descended
1 C. Longuet-Higgins, 'The FailureofReductionism', in C. Longuet-Higginsetal., TheNatureof
Mlind,Edinburgh, EdinburghUniversity Press,p. i6
2
Pace H. H. Field,ScienceWithout Numtibers,Oxford, Blackwell,I980 andB. Stroud'The Physical
World',Proceedingsof the AristotelianSociety, I987, p. 264.
Mind, Vol. 99 . 394 . April I990 ? OxfordUniversityPress I990

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
i86 TimCraneand D. H. Mellor
frommaterialism:the view that everythingis matter-for instance,the
view thatnothingexistsbut collectionsof atoms in the void-as opposed
say to Cartesian dualism, which held that as well as matter(extended
substance)thereis also mind (thinkingsubstance).Many physicaliststake
their doctrine to be a modern version of materialism:defendingthe
hegemonyofmodernmatteragainstthemysteriesof mentalsubstanceand
of mind/matter interaction.
But physicalismdifferssignificantly fromits materialistancestors.In its
seventeenth-century formof mechanism,forinstance,materialismwas a
metaphysicaldoctrine:it attemptedto limitphysicsa prioriby requiring
matterto be solid, inert,impenetrableand conserved,and to interact
deterministically and only on contact.But as it has subsequentlydevel-
oped, physicshas shown thisconceptionof matterto be wrongin almost
everyrespect:the 'matter'of modernphysicsis not all solid, or inert,or
impenetrable,or conserved; and it interacts indeterministically and
arguablysometimesat a distance.Faced withthese discoveries,material-
ism's moderndescendantshave-understandably-lost theirmetaphysical
nerve.No longertryingto limitthe matterof physicsa priori,theynow
takea moresubservientattitude:the empiricalworld,theyclaim,contains
just what a true completephysicalscience would say it contains.
But this raises two questions. What is physicalscience: that is, what
sciencesdoes it comprise?And whatgives it thisontologicalauthority?In
otherwords,whatentitlescertainsciencesto tellus in theirowntermswhat
the world contains-thereby entitlingthem to the physicalist'shonorific
title'physical'?
'Physical science' so construed certainly includes physics proper.
Physics is the paradigm (hence 'physical'). And chemistry,molecular
biologyand neurophysiology are also indisputablyphysicalsciences. But
not psychology,sociology, and economics. One may debate the exact
boundaryof physicalscience: but unless some human sciences,of which
psychologywill be our exemplar,lie beyond its pale, physicalism,as a
doctrineabout the mind, will be vacuous.
What puts psychologybeyondthe pale of the physical?Not the a priori
metaphysics of seventeenth-centurymaterialism,since thathas been refuted
by physicsitself.Nor thematerialist'sdenialofmentalsubstance.Psychology
can-and mostlydoes-deny thattoo: but it stilldoes notcountas a physical
science. The question of whetherthere is 'mental substance'as well as
'physicalsubstance'is an irrelevant one. For thatcontrastof substancesis
reallya contrastbetweentheircharacteristic betweenthinking,
properties: say,
and being extended. A merely thinkingsubstance is not a physically
respectableentitybecause thinkingis not a physicallyrespectableproperty.
But why not? What, if not the metaphysicsof materialism,preventsthe
empiricalpsychology of thought,and of othermentalphenomena,addingin
its own terms,as physicsdoes, to our inventory of whatthereis?

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No QuestionofPhysicalismI87

It is oftensaid that the human sciences have produced fewerresults

thanthe admittedlyphysicalsciences.Their laws are said to be fewand ill
established,and theirtheoriesto proliferate,and to predictfarless than
those of gravity,say, or of molecular biology. Perhaps this paucity of
resultsprovidesan epistemicbasis for denyingthat psychologyand the
restare physical-that is, entitledto tellus whatthereis. Perhaps theyare
just not good enough.
But that cannot be why psychologylacks the ontologicalauthorityof
physics,chemistryand the rest. There are, as we shall see, many well-
establishedpsychologicallaws. And anyway,this epistemicargumentis
the wrong way round. Those who think psychologyis epistemically
suspectdo so becauseits subject matteris not physical,and not vice versa.
In otherwords,theyhave some otherbasis fortakingphysicalscience to
exclude psychology,a basis fromwhichpsychology'sepistemicinferiority
is supposed to follow.And it is reallyquite obvious that this basis is not
epistemic. For accepting the results of psychology does not entail
accepting them as physical: on the contrary,the more such results
physicalistsaccept, the more theyreckontheyhave to explain (or explain
away) in non-psychologicalterms.
The bounds of the physicalare in factset fromthe outside. Something
about the mental is supposed to deprive psychologyof the ontological
authorityof physicsand chemistry.But what?What preventspsychology
fromtellingus in its own termswhat kinds of mentalthingsand events
thereare? There are a numberof answersto thatquestion: but none, we
shall argue, justifiesthe prima facie exclusion of psychologyfromthe
realmof the physicalwhichis needed to make physicalisma non-vacuous
doctrineabout the mind.

2. Reductionto physics
To assess physicalists'reasons fordismissingpsychologyas non-physical,
and thus ontologicallyinconsequential,we must ask what makes them
classifytheir favouredsciences as physical. What makes them count as
physicalnot only the many diversebranchesof physicsitself(mechanics,
electromagnetism, thermodynamics,gravity,and particle physics), but
also sciences like chemistryand molecularbiology?
One common answer is that these othersciences are physicalbecause
theyreduceto physics,which forpresentpurposes we may take to mean
that a physics enhanced with suitable bridge principles (to link its
vocabularyto theirs) would entail credible approximationsof all their
establishedlaws.3
Some theoriesin otherphysicalsciences have indeed been reduced to
3 See e.g. C. G. Hempel,Philosophy
ofNaturalScience,EnglewoodCliffs,
NJ,Prentice-Hall,
I966,
ch. 8.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
i88 TimCraneandD. H. Mellor
physicsin thissense,4butbyno meansall. But thoseforwhomreduction
to physicsis the touchstone of the physicaldo not proposeto do it in
practice.Theysimplyinsistthatitcan be done'in principle'. But whatis
the principle?It cannot be physicalism.These sciences cannot be
reduciblein principle becausetheyare physicalifreducibility in principle
(RIP) is supposedto tellus wheretheboundsofthephysicallie. So what
otherprinciple willtellus whichsciencescould'in principle' be reducedto
physics?
To answerthis,we mustfirstask to whatphysicstheRIP principle is
supposedto be applied:to presentphysics, or to somehypothetical future
physics?This questionposes a dilemma.For applyingthe principleto
presentphysicsentailsthatany futureextensionsof it would not be
physical:thatphysics, theparadigmphysicalscience,is alreadycomplete.
But no one believesthis.And if we applythe principleto an otherwise
unspecified futurephysics,we shallnotbe able to saywhichsciencesare
physicaluntilwe knowwhichofthemthatphysicsmustcover-whichis
just whatthe principlewas supposedto tell us. To use RIP to future
physicsto say whatthatphysicsmustcoverif it is to covereverything
physicalis obviously viciouslycircular.So thephysicalcannotbe defined
as whatis reduciblein principleto physics,eitherpresentor future.
We thinktheRIP principle's speciousappealactuallyrestson twoother
prejudices. One is theold dreamofthe'unityofscience',ofbeingable to
deriveall scientific laws fromone 'evermoreadequategrandscheme'.5
Butwe see no reasoneitherto believein or to aimforsucha scheme.The
worldevenoftheadmittedly physicalsciencescontainsa vastnumberof
verydifferent kindsofentities, propertiesand facts.That is whyso many
different sciences,usingwidelydifferent methods,are neededto study
them.No one couldthinkastrophysics and geneticsunifiedevenin their
methods, exceptunderthemostabstract ofscientific
descriptions method-
ology.And in theircontents, theydisplayno moreunitythanthatof a
conjunction. Nothingwrongwiththat:but thenwhycannotpsychology
supplyanotherconjunct?
But evenif some'unityof science'thesiswerecredible,it wouldnot
enabletheRIP principle to definethephysical.For evenphysicsproperis
notunified.Maybeit willbe someday;butevenifit is not,physicalists
will still accept gravity,quantumand electromagnetic phenomenaas
physical,to be identifiedanddescribedin theirowntermsbyindependent
physicalsciences.Similarlyforthe sciencesof chemical,biological,and
neurophysiological phenomena. So whynotforpsychology, thescienceof
mentalphenomena?
4 See M. Friedman,'TheoreticalExplanation',
in R. Healey(ed.), Reduction,
Timeand Reality,
Cambridge, CambridgeUniversity Press,198I.
5 H. Feigl, 'Physicalism,
Unityof Scienceand the Foundationsof Psychology' (1963), in his
Inquiries& Provocations,
ed. R. Cohen,Dordrecht,
Reidel,I98I, p. 315.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No QuestionofPhysicalismI89
The othersourceoftheRIP principle's appealis theidea thatthereis
reallyno moreto thingsthanthesmallestparticlestheyare madeup of.
Let us call thisthesis'microreduction',
or 'MR' forshort.6The idea is
verypersistent. Take Eddington's
twotables:hiscommonplace one,with
extension,colour,and permanence,versushis'scientific'
one,nothing but
myriadminuteparticles in emptyspace:thetablewhich'modernphysics
has by delicate test and remorselesslogic assured me . . . is the only one
whichis reallythere'.7Or morerecently, McGinn'sclaimthatsciencetells
us thatthewaythingsare is verydifferent fromthewaytheylook.The
tablethatlooksand feelsso solidis,he thinks physicstellsus,reallyfullof
holes.8
Now the studyof the smallestentitiesis indeedtraditionally called
'physics':departments of physicshave by longestablishedcustomcor-
neredthatparticular market.And thlismakesMR saythattheempirical
worldis physical,sinceit consistsonlyof its smallestparticles.We are
back with the doctrineof atoms in the void-or at least, in the
field whichcountas physicalsimplybecausetheyare microscopic.
The factthatphysics bymereconvention includesthestudyofthevery
smalldoesindeedtrivially entailthateverything extended in spaceeitheris
physicalorhassomephysical parts;andforsome,thistrivial truthis all that
physicalism means.9But forphysicalism so defined to be non-vacuous, one
mustalsotakethesesmallest thingstobe allthereis.Butwhatreasonis there
tothink this?Whyshouldwesupposetheexistence ofsub-atomic particlesto
requirethenon-existence ofatoms,molecules, tables,trees,ortennisrackets,
figsor fastfoodrestaurants-or animalsor peoplewithminds?
Proponents of MR can of coursedistinguish our non-existence from
thatof,say,unicorns. Thereareundoubtedfactswhichat leastappearto
be aboutus, whereasthereare no suchfactsapparently aboutunicorns.
Andofcourse,sincephysicsitselfalsostudiesverylargethings-galaxies,
quasars,etc. MR is also obligedto say whyfactsabout even these
admittedly physicalthingsare differentfromfactsaboutunicornson the
one handand factsaboutsub-atomicparticleson the other.WhatMR
actuallydoes say is thatall thesefacts aboutgalaxies,minds,and the
rest-reduceto factsabouttheirsub-atomic parts.So thosepartsare all
thereis,perhapsbecausewe neednotquantify overanything elsein order
tostateall thefacts-andwethinkwithQuinethatweshouldnotmultiply
entitiesbeyondquantificational necessity.'0
6 G. Schlesinger,
Methodin thePhYsicalSciences,
London,Routledge& KeganPaul, I963, ch.2.
A. S. Eddington,TheNatureofthePhysicalWorldl,
Cambridge,
CambridgeUniversity
Press,
I929, pp. XI-XIV.
8C. McGinn,TheSubjectiveView,Oxford,ClarendonPress,I983, ch.7.
9 G. P. Hellmanand F. W. Thompson,'Physicalism: Ontology,Determinationand Reduction',
JournalofPhilosophy,
I905.
10 W. V. 0. Quine,'On WhatThereIs', in hisFroma LogicalPointof View,HarvardUniversity
Press,I953.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 9go TimCraneandD. H. Mellor

Butthisappealtoreduction showsthatMR itselfneedsa strongformof

theveryRIP principle itis supposedtosupport.Anditcannothaveit.For
unlessthesciencesoftherelatively large,including psychology,reduceto
microphysics, we shall stillneed to quantifyoverentitiesdescribedin
thosesciences'terms.Butin fact,as weshallnowshow,eventhephysics of
therelatively largedoesnotreduceto microphysics. So evenifall sciences
were reduciblein principleto physics,this would not entailthatthe
smallestparticlesare all thereis: MR would be false,even if the RIP
principle weretrue.So theRIP principle cannotbe usedto supportMR.
Whatis trueis thatfactsaboutpartsoftenexplainfactsaboutwholes.
As a thesisaboutexplanation, MR is oftena goodworking hypothesis.But
it is notalwaysverified, evenin microphysics. If forexamplewe takethe
quantummechanicaldescription of a quantumensembleto be complete
(as orthodox interpretations do),thesuperposition principleentailsthatits
properties willnotbe a function onlyofthoseof itsisolatedconstituents
plus relations betweenthem.Orthodoxquantumphysicsis notmicrore-
ductive.Andsomephysicsis positively macroreductive: Mach's principle,
forexample,whichmakestheinertial massevenofmicroparticles depend
on howmatter is distributed throughout theuniverse.We realizeofcourse
thatMach'sprinciple andorthodox quantumtheory arecontroversial,and
thata futurephysicsmightwell abandonthem.But theycannotbe
abandonedbecause they conflictwith an MR entailedby modern
microphysics: since,as theyshow,it entailsno suchthing.
Andevenin themostordinary physics,MR doesnotalwayshold.It is
indeedusuallytruethatwherethepartsofsomething go,thewholething
mustgo too: thata gas samplemustgo whereits moleculesgo. But
equally,its moleculesmustgo whereit goes: sinceanythatdo not will
thereby ceaseto be its molecules.Andthatis nottheonlywayin whicha
gas's moleculesare as muchgovernedby it as it is by them.Supposefor
instancethat our sample's volumeis suddenlyhalved at a constant
temperature. If thegas is ideal,Boyle'slaw entailsthatwhenitspressure
settlesdownagainitwillbe twicewhatitwas.Thatlawdoesnotdictateall
theinterim behaviourof thesample'smolecules-exceptthatit mustbe
suchas willeventually doublethesample'spressure.That muchoftheir
behaviouris determined-andthereby explained-macroreductively by a
law governing thesampleas a whole.
So evenas a principle ofexplanation, MR doesnotalwayshold,evenin
physics.Its explanatory valuecannottherefore supportitas an ontological
thesis.A fortiori, it cannotsupportphysicalism. But it couldnotevendo
that if it were true. For no true readingof MR could entail that
macroscopic entities, and theirproperties and relations,are impugnedby
beinglinkedby laws to properties and relationsof theirsmallerparts.
They cannotbe. For iftheywere,therewouldhaveto be somesmallest
entities,without parts:thatis,a limittothesmall-scale structureofmatter.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No QuestionofPhysicalismI9I
But thereclearlyneedbe no suchlimit.So theexistenceofthecurrently
smallestknownparticlescould not be refutedby the discovery of even
smalleronesinsidethem.Butthenatomstoomustbe abletoco-existwith
theirsub-atomic parts,moleculeswiththeiratoms,andso on,up totables,
trees-andus. The existence ofanimalsandpeople,withtheirpsychologi-
cal and socialproperties and relations, cannotbe deniedjustbycrediting
themwithpartssmallenoughto matterto microphysics.
The factthatphysicstakesin theverysmallhas fostered themyththat
it is a universalsciencein a sensein whichothers-likepsychology-are
not universalbut merely'specialsciences'.11 It has fostered thismyth
becauseit makeseverything biggerthana pointhave(or havepartssmall
enoughto have) properties thatare physicalby meredefinition. In a
similarway,everything thatmoveshasphysicalproperties, suchas inertial
mass,justbecausephysicsbydefinition includesthescienceofmechanics.
Butthatdoesnotmakethesesciencesuniversal, in thesenseofencompass-
ingall theproperties and relationsof things;norbasic,in thesensethat
othersciencesmustreduceto them.In anysensethatwouldsupporta
non-vacuous definitionof thephysical,whichis whatphysicalism needs,
mechanics and microphysics areno moreuniversal or basicthanpsychol-
ogyis. They are merelythe specialsciencesof motionand of the very
small.
In short,if the phenomenaof psychologyare less ontologically
acceptablethan thoseof physicsand chemistry, it cannotbe because
psychology is irreducibleto presentor futurephysics.Reducibility to
physics, or to microphysics, is a hopelesstestoftheontological authority
ofa science:a testwhichnotevena physicalist can applyconsistently.For
as we haveseen,reducibility in practiceis neitherfeasiblenortothepoint;
whilethosewhoclaimreducibility 'in principle'eitherbegthequestionor
appealto principles, of theunityof scienceor of microreduction, which
modernphysicsitselfdenies.

3. Mental causationand intentionality

How else mightphysical(thatis, ontologicallyauthoritative)sciencebe
definedso as to excludepsychology?Perhapsby causation,whichmany
thinkis essentially
physical.Perhapsthe physicaljust is thecausal,and
whatphysicalism reallymeansis thattheempiricalworldcomprisesall
and onlythoseentities,properties, and factswhichhavecauses
relations,
oreffects.
This definition
clearlyunderliesone familiar
formulation ofthe
mind-bodyproblem:how can mentalstateshave effects in a physical
world?This questionwould not pose such a problemif it were not
assumedthatcausationis essentially
non-mental.
" See J. A. Fodor,'SpecialSciences',Synthese,
I974.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
I92 Tim Crane and D. H. Mellor

But whyshould we assume this?It is surelyobvious thatthereis plenty

of mentalcausation.Suppose you see a friend,and thiscauses you to wave
to him: how? Somethinglike this: lightis reflectedfromhim onto your
retina;impulses travelup your optic nerve; your striatecortexprocesses
the information carriedby them;you form(somehow) the beliefthatyour
friendis there;thismakesyou formthe intentionto greetyourfriend;that
makes certainthingshappen in yourmotorsystems;theycause yourarm
to rise . . .. Both physicaland mentalfactsseem equally involvedin this
chain of causation. How then can a physicalismdefined by causation
exclude these apparentlymentalcauses and effects?
It is indeed an old thoughtthatmentalcausationis hard to make sense
of,and especiallycausationlinkingthe mentalto the non-mental,because
theyseem to be so different. But whyshould thatimpressanyonewho has
learnedfromHume thatcausationnever'makes sense': thatit is alwaysa
matterof fact,not of reason?Nothingin eitherHumean or othermodern
analyses of causation forces causes to be like their effects;nor does
anythingin them stop causes and effectsbeing mental.
Take the requirementthattokencauses and effectsbe localized in space
and time, so that they can be contiguous (or, if need be, dense or
continuous)and so thatone can precede the other.Token sensationsand
even token thoughtscan certainlybe localized enough for that (since
localizing a token thoughtno more localizes its unlocalizable abstract
content than localizing a red object localizes the abstract colour red).
Nothingabout the mentalpreventspeople's tokenthoughts,feelings,and
sensationsbeing whereverand wheneverthose people are, in orderto be
wheretheycan have the immediateand thereforecontiguousmentaland
non-mentalcauses and effectswhich theyclearlyappear to have.
Other common demands on causation are also just as easily met by
mentalas by non-mentalcauses and effects:forinstance,the demand that
causes be in the circumstancessufficientfortheireffects,or necessary(or
both); or thattheymake theireffectsmore probablethantheywould have
been withoutthem; or thatcausal relationsinstantiatelaws. It is hard to
see whyany such conditionshould presentany obstacleto theexistenceof
mentalcauses and effects.
If thereis a problemwithmenul causation,it lies in intentionality,
the
mind's capacity to representaspects of the world. And intentionality is
indeed oftensupposed to preventmentalphenomenafrombeing,as such,
physical.Thus Fodor:
I supposethatsooneror laterthephysicistswillcompletethecataloguethey've
beencompiling of theultimateand irreducible of things.Whenthey
properties
do, the likesof spin,charm,and chargewill perhapsappearon theirlist.But
aboutness simplydoesn'tgo thatdeep.'2
surelywon't;intentionality
12 J. A. Fodor,Psychosemantics: ofMeaningin thePhilosophy
theProblem ofMind,Cambridge,
Mass., MIT Press,I987, p. 97.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No Questionof Physicalism I93

But in the previoussection,we have alreadydisputed the pretensionsof

physicsto provideall 'the ultimateand irreduciblepropertiesof things'.
And no one impressed by our argumentswill think it matters that
intentionality goes less 'deep' in thissense thanspin and charge.For many
non-mental(e.g. chemicaland biological)propertieswill also not figureon
the physicists'list; and if thatdoes not impugnthem or the entitiesthey
characterize,why should it impugn intentionalityor the entities it
characterizes?
But many philosophers would still agree with Fodor's subsequent
comment that 'the deepest motivationfor intentionalirrealismderives
froma certainontologicalintuition:that thereis no place forintentional
categoriesin a physicalistview of the world'. Thus Field writes:'Any
materialistwho takes beliefsand desiresat face value . . . must show that
the relationsin question are not irreduciblymental.'13 If this intuition
were correct, and there were independent reasons for accepting his
'physicalistview of the world',thenwe would indeed have reason to deny
therealityof intentionality, and thusmuch,ifnot all, of themental.But as
we shall see, the intuitionis wrong.
What is the problemof intentionality supposed to be? Intentionalstates
typicallyhave threedistinctivefeatures:(i) theyseem to be affectedby,
and to cause actions involving, distant objects or events; (ii) their
ascriptioncreatesnon-extensionalcontexts-sentences whose truth-value
may alterwhen names or descriptionsin themare replaced by othersthat
apply to the same things;and (iii) they can be about objects or events
whichdo notexist.Suppose forexamplethat(i) you read somethingabout
Santa Fe thatmakes you want to go there,which causes you to get on a
plane and do so. But (ii) you do not wantto go to themostbeautifulcityin
New Mexico, whichSanta Fe is, because you do not knowthatit is. And
(iii) you could have wantedto go to Santa Fe even if,like Eldorado, it did
not exist.
The challengewhich(i) and (iii) presentis to explain how Santa Fe can
cause you to act as you do when it is so faraway,and need not even exist.
No one believes that a city can have such effectsdirectly at such
distances-especially when it need not exist.Your actionmustbe directly
caused by some intrinsicpropertyyou actuallyhave, not by yourrelations
to distantand possiblynon-existentobjects, like Santa Fe, or to abstract
ones like the possiblyfalse propositionthatit is in New Mexico.
But this does not mean that the causal powers of tokenthoughtsand
othermentalstatescannotdepend on theircontents:theycan. All it means
is thattheymustdo so indirectly, via a mentalrepresentation,i.e. via some
intrinsicnon-relationalpropertyof the mentalstate (or of its owner). A
tokenthoughtmust have some such intrinsicproperty,correlatedsome-

13 H. H. Field,'MentalRepresentation', I978, p. 78.

Erkenntnis,

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
194 Tim Crane and D. H. Mellor

howwithitscontent, to giveit itsrightcausesand effects. An instanceof

thisproperty is, we maysay,thelocal causalsurrogate forthatcontent.
But theseintrinsicproperties could still be mental.They could be
sensations, or visualor othermentalimagesor models-whichneednot,
incidentally, be conscious.14And evenif thecompositional structureof
thought requiresthesetokensto forma correspondingly complex('syntac-
tic')structure, theycouldstillbe images-likeShepard'sshapescomposed
ofimagesofcubes.'5 So theproblemwhichthoughts poseforcausationis
notthattheyarementaland causationis not.It is thatcausationdepends
directly onlyon intrinsic properties, whereasthecausalpowersof token
thoughts dependon theircontents, whicharenotintrinsic.' 6 This indeed
showsthatthesecontents needcausalsurrogates: butnotbecausetheyare
mental,sincethesurrogates couldbe mentaltoo.
Moreover,the need forcausal surrogates is by no meansconfinedto
psychology. They are needed throughout physicalscience. It is, for
example,a standardfunctionof physicalfieldsto providelocal causal
surrogates forwhatwouldotherwise be unmediated actionat a distance.
But no one thinksthatacceptingNewtoniangravitational fieldsmeans
denyingthe physicalstatus,or the existence,of the Newtoniangravity
theymediate.
In otherpartsof physicsand chemistry, causal surrogates are needed
also to bringaboutwhatwouldotherwise haveto be backward causation.
We notedearlierthatBoyle'slaw makestheeventualpressureofan ideal
gas sample double afterits volume is suddenlyhalved at constant
temperature. Butthattokenequilibrium pressure, P, cannotdirectly affect
the non-equilibrium processeswhichlead to it, since thatwouldneed
backward causation.Moreover, P, likeSantaFe, neednotevenexist.The
sample'svolumemaybe alteredagainbeforeit reachesequilibrium: but
thiscannotaffect itsbehaviour beforethat.So thefuture P needsa causal
surrogate in thepresent to makethesamplehead forP, justas Santa Fe
needsone inyou to makeyouhead forit.
And as forthiscase, so forall systemsthattendto stableequilibria:
fromsimplependulumsto chemicaland biologicalreactionsofall kinds.
The Gibbs' potentialsof chemicalthermodynamics, forexample,are
causal surrogates forthe equilibriato whichchemicalsystemstend:17
equilibriawhose existenceand physicalstatustheycertainlydo not
impugn,any more than the fieldmechanismsof Newtoniangravity
14
MentalModels,Cambridge,
See P. N. Johnson-Laird, Cambridge Press,1983; R. N.
University
Shepardand L. A. Cooper,MentalImagesand TheirTransformations,
Cambridge,Mass.,MIT Press,
I982.
15Shepardand Cooper,op. cit.,ch. 3.
16As oneofus hasshownindetailelsewhere: andCausationofThought,
T. M. Crane,TheContent
CambridgeUniversity Ph.D. Dissertation,I989.
17 K. G. Denbigh,The Principles of ChemicalEquilibrium, Cambridge,CambridgeUniversity
Press,1955, p. 76.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No QuestionofPhysicalismI95

impugnit, or than the kineticsof gases refutesBoyle's law or shows that

halvinga gas sample's volume does not reallycause the doubling of its
pressure.Why thereforeshould physiologicalor psychologicalaccountsof
how the contentsof tokenthoughtsproduce theireffectscontradictthem,
or the causal explanationstheygive of our actions?
So much for the alleged problems posed by (i) and (iii) for the
ontologicalauthorityof intentionalpsychology.What about (ii), the non-
extensionalityof ascriptionsof intentionalmental states?This does not
exclude the mental from the physical either, since non-extensionality
occurs in physics too.'8 This is because laws entail non-extensional
conditionals.Suppose forexample thatH and K are the genes thatgive us
heartsand kidneys.The factthatwe all have both does not make 'anyone
who had gene H would have a heart'entaileither'anyonewho had gene K
would have a heart' or 'anyone who had gene H would have a kidney'.
The probabilisticlaws of modern microphysicscannot be extensional
foranotherreasontoo,because 'p(. . .) n' is notextensional:forifit were,
'a is the F' and the necessarytruth'p(a is a) = i' would entail 'p(a is the
F) = i', which it clearly does not, on any view of probability(take for
example 'F'= 'next Prime Minister').
The non-extensionality of probabilityincidentallyexplainsthatof many
singular causal instances of the contexts '. . . because . . .', even in physics.
This is because causation gives effectsprobabilities,if only subjective
ones. Probabilisticaccounts of causation make that explicit, and it is
impliciteven in deterministic accounts.'9 Effectsof sufficient causes, for
example,have probabilityi; and effectsof necessaryones would in their
absence have probabilityo. So 'E because .. .' must be non-extensional,
since thougha's being the F mightgive 'E' a contingentprobability,a's
beinga cannot.And '. . . because C' cannotbe extensionaleither,because
of its counterfactualimplications:the probabilityof a necessarytruthlike
'a is a' cannot depend on C, even if thatof a true 'a is the F' does.
These and other reasons convince us, pace Davidson and others,that
even in physicssingularcausationneverdepends on, and mostlyis not,an
extensionalcausal relationbetween particulars.20But if causal contexts
can be non-extensionalanyway, they can perfectlywell contain non-
extensional contexts like 'believes . . .', 'wants . . .', 'fears.. .', etc.: as in 'b
fearsthata is theF because a told herso' or 'b does D because she wantsa
to be theF and believeshe will be onlyifshe does D'. So we see no reason
eitherto deny the causation whichsuch sentencesobviouslyreport,or to
18 Statesof MechanicalSystems',Mfind,
B. Enc, 'Intentional 1982.
19 E.g. D. H. Mellor,'On RaisingtheChancesof Effects', and
in J. H. Fetzer(ed.), Probability
Reidel,1988;J. L. Mackie,'Causes and Conditions'(1965),in E. Sosa (ed.),
Causality,Dordrecht,
Oxford,OxfordUniversity
Causationand Conditionals, Press,1975.
(1967),in hisEssaysonActions
20 D. Davidson,'Causal Relations' andEvents, Oxford,Clarendon
Press,1980;D. H. Mellor,'The Singularly AffectingFactsof Causation',in D. Pettitet al., (eds),
Metaphysics Oxford,Blackwell,1987.
and Morality,

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
196 TimCraneand D. H. Mellor
suppose that it must be based on, or reduced to, any extensionalcausal
relation-let alone a non-mentalone thatrelatesnon-mentalparticulars.
In short,all thesupposedlyproblematicfeaturesof intentionalstatesare
as endemic to physics,and in particularto non-mentalcausation,as they
are to psychology.The notionof causationwill thusnot serveto definethe
physical(and hence ontologicallyauthoritative)sciences in such a way as
to exclude psychology.Definingthe physicalas the causal will not make
physicalisma non-vacuousdoctrineabout the mind.

4. Psychologicaland psychophysical
laws
In orderforthe issue of physicalismto be a seriousone, therehas to be a
principleddistinctionbetweenthe mentaland the physicalwhichexplains
whynon-mentalsciences have an ontologicalauthoritywhichpsychology
lacks. We have seen thatneithercausationnor reducibilityto physicscan
provide such a distinction. But perhaps laws can. The ontological
authorityof science arguablyrestson the laws it discovers,which tell us
what kinds of things there are, and what properties and relations
distinguishthem. But many agree with Davidson that the mental is
'anomalous': that strictlyspeakingthereare no psychologicalor psycho-
physicallaws.2' If that were so, psychologywould add nothingto our
ontologyof non-mentalkinds,withtheirdistinctivenon-mentalproperties
and relations.
But whyshould we deny that,forexample,'All men are mortal'(a true
psychophysicalgeneralizationlimitingthe sentienceof membersof our
species) is a law? There are some bad reasons fordenyingit, which we
shall not considerin detail. One is the idea thatlaws are necessarilytrue,
whichno generalizationsabout thementaleverare. Thus forMcGinn, for
mentaltermsto featurein laws is for'universalgeneralizationscontaining
mental terms[to be] metaphysicallynecessary'.22But, he argues (influ-
enced by Kripke's well-knownargumentagainstthe identitytheory),23no
non-analyticnecessarilytrue generalizationslink mental termseitherto
non-mentalor to othermentalterms.
Nor they do: but then none links the termsof physics to each other
either.The laws of physicsare not metaphysicallynecessary.We agree
with Davidson that laws must be 'supported by their instances' and
'supportcounterfactualand subjunctiveclaims' ('if x were F it would be
G').24 But 'All men are mortal'can clearlymeet theseconditionswithout
being a necessarytruth:the factthatsomethingwouldnotlive forever ifit
21 D. Davidson,'MentalEvents'(1970), in his Essayson Actions
and Events,Oxford,Clarendon
Press,1980.
22
C. McGinn,'Philosophical
Materialism', 1980,p. 187.
Synthese,
23 S. Kripke,
'Namingand Necessity', inD. DavidsonandG. Harman(eds),Semantics ofNatural
Languages,Dordrecht,
Reidel,1972.
24 D. Davidson,'MentalEvents',loc. cit.,p. 217.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No QuestionofPhysicalismI 97
werehumandoes notmeanit couldnot,anymorethan'if we wentwe would
go by bus' meanswe could not go by train.Nor therefore does the factthat
thelaws of physicsmeettheseconditionsshow themto be necessarytruths;
and one of us has arguedelsewherethattheyare not.25So if termshad to
featurein non-analyticnecessarylaws in order to count as physical,the
termsof physicswould not count,nevermind those of psychology.
The law thatall Fs are G entailsonlythatanythingwouldbe G ifit were
F, not thatit mustbe. (And theprobabilisticlaw thatall Fs have a chancep
of being G, whereo <p < i, does not even entail that not even if it is a
necessarytruth.)The mere possibilityof exceptionsto psychologicaland
psychophysicalgeneralizationscannot thereforestop them being laws.
And even if it could, even if laws did have to be both necessaryand
deterministic, how would one show withoutbeggingthe question that a
true generalizationreallycould have exceptions?Not just by imagining
them. We can all imagine lightgoing fasterin a vacuum than its actual
speed c. This does not show thatit reallycould go faster,stillless thatthe
constancyof c is not a law. And similarlyin psychology.Anyone can
imaginebrinetastinglike port:it can stillbe a law thatto no one withour
taste-budswould it ever taste anythinglike that.
Anotherbad reason fordenyingthe existenceof psychophysicallaws is
the so-called 'variable realisation'of mentalstates:the factthat'the range
of physical states fit to realise a given mental state can be indefinitely
various'.26 That cannot stop psychophysicalgeneralizationsbeing laws.
For if it did, therewould be hardlyany laws in physicseither.States like
masses, volumes,and temperaturesare even more variouslyrealized than
mentalstates:one can have a gramor a litreof almostanything,at any one
of an indenumerableinfinity of temperatures.27 So if variablerealization
does not rule out laws in mechanicsand thermodynamics, it can hardly
rule themout in psychology.
Nor should we be impressedby the inabilityof armchairreflectionto
excogitatepsychologicalor psychophysicalgeneralizations.Physics and
chemistryare not excogitatablea priori, and we see no reason why
psychology should be. It can take as much unobvious theory and
experimentto discoverthe psychophysicsof taste,or of vision,28or the
unconsciouspsychologyof inference,29 as to uncover(say) the biochemis-
tryof reproduction.

25
See D. H. Mellor,'Necessitiesand Universalsin NaturalLaws', in D. H. Mellor(ed.),Science,
Belief& Behaviour,Cambridge, Cambridge University Press,I980; D. M. Armstrong, Whatis a Law
ofNature?,Cambridge, CambridgeUniversity Press,I983.
26 C. McGinn,'Mental States,NaturalKinds, and Psychophysical Laws I', Proceedingsof the
Aristotelian
SocietySupplementary Volume,I978, p. 197.
27 See M. Wilson,'What is thisThing Called "Pain"?-the Philosophy of ScienceBehindthe
Contemporary Debate',PacificPhilosophical
Quarterly, I985, p. 235
28
D. Marr,Vision,San Francisco,Freeman,I982.
29 P.
N. Johnson-Laird,op. cit.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
198 TimCraneand D. H. Mellor
Stich'sfailureto excogitate
So in particular, lawsfeaturingintentional
mentalstatesdoesnotmeantherearenotany.30The obviousexplanation
ofourinabilitytostatesuchlawsin simpleandexceptionless formsis that
our intentionalpsychology is too complexand (probably)probabilistic.
But so is themeteorology of hurricanes, and thequantummechanicsof
largemolecules.Theirlaws,forthoseveryreasons,are notstatablebyus
in simpleand exceptionlessforms.No one infersfromthisthatthereare
no suchlaws;and theinference is no betterin psychology.
Davidsonhimself doesnotuse thesearguments againsttheexistence
of
psychologicallaws.His ownargument goesas follows:
(i) laws.
There are no strictpsychophysical
(2) mustinstantiate
Singularcausesand effects strictlaws.
closed'system,
(3) The mentalis nota 'comprehensive beingaffected
by
thenon-mental, whichdoes formsucha system.
strictpsychophysical
cannotinstantiate
But by (i) thesementaleffects
laws.So
(4) 'thereareno strictlawsat all on thebasisofwhichwecan predictand
explainmentalphenomena'.31
The argument failsat everystep.(i) is false,and notonlybecause'All
men are mortal'is a law. There are manymore such laws, linking
sensations-likepains,smells,tastes,and visual,auraland tactilesensa-
tions-to non-mental featuresof thosewho have them.There mustbe,
becausewholeindustries dependon them.Thinkofthelawswhichmust
underliethereliableproduction and use ofanaesthetics, scents,narcotics,
sweeteners, colouredpaintsand lights,loudspeakers, and softcushions.
And ifNewton'slawsof motionsuffice to add massesand forcesto our
physicalontology,these laws must suffice to add to it the kindsof
sensations in
thatfeature them.
But even if therewereno such psychophysical laws,thiswouldnot
undermine theontologicalauthority psychology. ifno lawslinked
of Even
thementalto thenon-mental, psychology could stillhaveits own laws,
definingitsown mental on a
ontology, par with thatofchemistry (say).For
in
as we saw section2, chemistry's ontological authority does notdepend
on itsbeingreducibleto physicsvia physicochemical laws.Nor therefore
can psychology'sontologicalauthority dependon therebeingpsychophy-
sicallaws.32
Butas we haveseen,oursensations arein factsubjecttopsychophysical
laws,whichthemselves to refuteDavidson'sdenialthat'therecan
suffice
30
S. Stich,FromFolkPsychologyto Cognitive Science,Cambridge, Mass., 1983, ch.7.
31 D. Davidson,'MentalEvents',loc. cit.,pp. 224-5.
32
Laws', Philosophical
See W. Lycan,'Psychological Topics,I98I.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No Questionof Physicalism I99

be strictlaws linkingthe mentaland the [non-mental]';33since theselaws

may verywell be 'strict'(i.e. deterministic). Davidson is admittedlymore
interestedin intentionalstates than in sensations;but the refutationstill
holds, since sensations are indisputablymental as Davidson himself
admits.3
So (i) is false. And so is (2): causes and effectsneed only instantiate
probabilistic laws.3 5 But can we not thereforemake (2) true-and
strengthen(4)-by deleting'strict' throughout?Indeed we can, and we
should: but thatwill not help Davidson, since it onlymakes(i) even more
incredible.
Nor does admittingprobabilisticlaws do anythingto rescue (3). For
whethercausation needs strict or merely probabilisticlaws, the non-
mental no more formsa 'comprehensiveclosed system'than the mental
does.
For whatdoes 'a comprehensiveclosed system'mean?For Davidson, it
means a systemof 'homonomic' laws, which 'can hope to be precise,
explicit and as exceptionlessas possible' only because they draw their
concepts'froma comprehensiveclosed theory'.The non-mentalsciences
can provide such a theory,Davidson claims; but psychologycannot. Its
generalizationsare hopelessly 'heteronomic':that is, they 'may give us
reasonto believe thereis a preciselaw at work,but one thatcan be stated
only by shiftingto a different vocabulary'.36
But thisdistinctionwillnot do, since physicsitselfis fullofheteronomic
laws. Take Newtonian mechanics,which definesNewtonian concepts of
forceand mass by sayinghow theycombineto cause acceleration.But the
laws of motionthatdo this do not forma closed theory.Indeed, without
some furtherlaw relatingforceto otherconcepts,they formno testable
theoryat all. In the theoryof Newton's Principia,the furtherlaw is the
inversesquare law of gravity.But thattheoryis not closed either.There
are manyotherkindsof force:electrical,magnetic,viscous,etc. So as a law
of net force,Newton's law of gravityis as hopelesslyheteronomicas the
laws of psychology:it can be made exceptionlessonlyby provisosinvoking
alien conceptsof electricity, etc.37 And similarlyforall the otherlaws of
force.All are trueonly as laws of kindsof forces:gravitational, electrical,
etc., which combine into net forcesby vectorialaddition. The theoryof
Newtonian mechanics is just the conjunctionof all such laws, however
diversetheirotherconcepts,withNewton's laws of motion.
Newtonianmechanicshas ofcourse been superseded,but not because it
3 D. Davidson,'MentalEvents',loc. cit.,p. 212.
3 ibid.,p. 21I; see also his 'Repliesto Essays',in B. Vermazenand M. Hintikka(eds), Essayson
Davidson:Actionsand Events,Oxford,ClarendonPress,I985, p. 246.
35 See e.g. P. Suppes, 'Davidson'sViews on Psychology as a Science',in B. Vermazen& M.
Hintikka,(eds), op. cit.;D. H. Mellor,'On RaisingtheChancesof Effects', loc. cit.
36 D. Davidson,'MentalEvents',loc. cit.,p. 2I9.
37 See N. Cartwright, How theLaws ofPhysicsLie, Oxford,ClarendonPress,I983, ch. 3.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
200 TimCraneand D. H. Mellor
was onlya conjunction: fora conjunction, as we saw in section2, has all
theunitya scienceneeds.So our'comprehensive closedtheory' canalsobe
a simpleconjunction: the conjunction of all truescientifictheoriesand
laws.But thento say,as Davidsondoes,thatthenon-mental sciencescan
supplythisconjunction on theirown is simplyto denytheexistenceof
psychological laws: whichbothbegs the questionand is refutedby the
lawsthatwe knowlinksensations to theirnon-mental causes.
So Davidson'sargument (I)-(4) quitefailsto showthatthereare no
psychophysical or psychological laws.But thisdoes notrefutehis claim
thatthereareno lawslinking intentional mentalstates.Andforthatclaim
Davidsongivesa specialargument, whichrestsmainlyon twoconnected
ideas: the 'holism' of the intentional, and the 'constitutive ideal of
rationality'.
The holism of intentionalmentalstates amountsto their being
conceptually interdependent, whichsensations arenot.The beliefthatP,
forexample,mustinhibitthebeliefthatnot-P,and also thedesirethatP
(peopledo notwantwhattheythinktheyalreadyhave).Again,neither
beliefnor desirecan cause actionon its own. To do thattheymust
combine,and different combinations can causethesameaction:I can say
'P', forexample,eitherbecauseI believeit and wantto speaktrulyor
becauseI disbelieveit and wantto lie. And thereis no doubtthatsuch
familiarrelationsbetweenbeliefs,desiresand actionsdo partlydefine
them,and thusstopanylawsinvolving thembeingwhollyindependent.
Butthesefactscannotstoptherebeingsuchlaws,becausetheytoohave
Newtonian parallels.Newtonian force(/)andmass(m) arealsoconceptually
interdependent, beingpartly defined bytherelationf=ma,whichstopslaws
involving thembeingindependent of each other.And thisrelationtoo
requiresforces andmassestocombinetoproducetheireffects (accelerations)
-and letsmanycombinations cause thesameeffect. So we can no more
infera forcef or a massm fromtheacceleration a theycausethanwe can
infera beliefor a desirefromtheactiontheycause.In short,holismalone
willnotsuffice to distinguish theintentional fromthenon-mental in a way
thatwillshowit to be anomalous-asDavidsonagainadmits.38
WhataboutDavidson's'constitutive ideal of rationality'?This is the
idea thatthe relationsbetweenbeliefs,desiresand actionsmentioned
above partlydefineor constitute (hence'constitutive') whatit is to be
rational.For instance, thefactthatthebeliefthatP willgenerally inhibit
the beliefthatnot-P is one of the holistictruismsthathelp to define
rationality:it is rationalnot to have obviouslycontradictory pairs of
beliefs.Rationalityis an idealbecausethinkers canbe moreorlessrational:
theycan failto havethetotality oftheirintentional statesstandingin all
these'rational'relations.
38
'MentalEvents',Ioc. cit.,p. 22I; 'Repliesto Essays',loc. cit.,p. 248.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No Questionof Physicalism 201

This may all be true: but again it cannot rule out psychologicallaws
since it too has a Newtonian parallel. Indeed everythingthat Davidson
says is peculiarto 'our use of theconceptsof belief,desireand therest'has
a Newtonianparallel. Here it is:
We muststandprepared,as theevidence[of accelerations inducedby gravity,
etc.] accumulates,to adjust our theory[of the forcesand masses
electricity,
involved]in thelightofconsiderations ofcogency[satisfying Newton'slaws]:the
constitutiveidealofrationality[Newton'slaws]partlycontrolseachphasein the
evolutionof whatmustbe an evolvingtheory. An arbitrary choiceoftranslation
scheme[fromaccelerations to forces]wouldprecludesuchopportunistic temper-
a rightarbitrarychoiceofa translationmanualwould
ing of theory:put differently,
be ofa manualacceptablein thelightofall possibleevidence,and thisis a choice
we cannotmake.39
We have italicizedthe two debatableanalogies.First,rationality, which
many thinkis a normativenotion, constrainingfor example, what one
oughtto believe. Well, maybe it is, but a belief'srationalitymay stillbe a
factabout it, forexample,somethingthatmakesit probablytrue;withthe
constitutiveideal simplyrequiringbeliefsto be so relatedto each other,
and to theirperceptualcauses, thatundernormalconditionsmostof them
are true. And that,far frompreventinglaws linkingthe contentsof our
beliefs to our surroundingsand to the non-mentaloperation of our
senses,positivelyrequiresthereto be some such laws (if onlyprobabilistic
ones).
Secondly, the claim that no evidence can enable us to choose a right
translationscheme: thatis, one whichcorrectlyinfersbeliefs,desires,etc.
fromtheir perceptual causes and behavioural effects.But if this is to
providea disanalogywithmechanics,it cannotjust mean thatno evidence
could entail the right theory. That is true in spades in Newtonian
mechanics,even ifforcesare observable,since everyascriptionofa mass at
any time t entails an indenumerableinfinityof net accelerationsunder
differentnet forcesat t, none of whichentailsany other,and only one of
which can be actual. How could intentionalmental states be more
underdetermined by the evidence forthem than that?
Davidson, however,thinksthat Quine's 'indeterminacyof translation'
shows that they must be.40 He says that the anomalism of the mental
For if
'tracesback' to the 'centralroleof translation'and its indeterminacy.
thereis no determinatetranslationof sentences,thereis no rightstatement
of whattheymean. So, since theirmeaningsare the contentsof the beliefs
they would express,there is no rightstatementof those either:that is,
beliefs(and a fortioriotherintentionalmentalstates)have no determinate
contents.Contents,like sentencemeanings,are not just underdetermined
39 'MentalEvents',loc. cit.,p. 223.
40 Mass., MIT Press,I960, ch.2.; D. Davidson,
W. V. 0. Quine, Wordand Object,Cambridge,
'MentalEvents',loc. cit.,p. 222.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
202 TimCraneand D. H. Mellor

by evidence:theysimplydo not exist.And if theydo not exist,they

cannothaveinstanceswhichfeature
certainly in laws.
Butwe denytheindeterminacy oftranslation,
forfamiliar
reasons.4'As
we havealreadyseen,and manyothershavepointedout,42it cannotbe
entailedby the underdetermination of theoriesby evidence,or even
Newtonianmechanicswould have no laws. Nor, withoutbeggingthe
presentquestion,can it be entailedby 'Quine's claimthattheoriesof
translationare ... underdeterminedeven by the totality of truths
expressiblein termsofphysics',43sincepsychological and psychophysical
laws are ex hypothesi not so expressible.To base the indeterminacy of
translationon thatclaimis to base it on whatwe saw in section2 is an
untenableversion-the reducibility-to-physics version-of the very
physicalism it is beingused to support.
We knowof no otherreasonto denya priorithe existenceof laws
involving intentional mentalstates.But mightnotthetotality of all true
non-mental theories be infactso comprehensive and closedas to preclude
psychological and psychophysical laws?We do notsee how.No amountof
physics,forexample,can stop mentalstatesinstantiating otherlaws as
well.The otherlawsmustofcoursebe consistent withphysics-butonly
becauseall truths mustbe consistent witheachother.Thattruism givesno
priorityto physics,whoselaws mustequallybe consistent withthoseof
psychology.
Butperhapsthisreadingof(3) maylookmoreplausibleas a thesisabout
causation,ratherthanaboutlaws:and Davidsonhimself suggeststhatthis
is howhe understands (3) whenhe saysthat'too muchhappensto affect
thementalthatis notitselfa systematic partof themental'.44
For supposephysicsdid forma comprehensive causalsystem:so that
lawsofphysicsmadeeach brainstateor bodilymovement b2 ofyoursat
anytimet2 be determined by yourbrainstatesat an earliertimet, (plus
non-mental inputbetweent, and t2). How could yourmentalstates
betweent1and t2alsoaffect b2without thesedeterministic
violating laws?
But now considera parallelcase. Suppose Kepler's laws made the
Earth'sorbitalposition P2 at t2be determined byitsposition p1 at t1(plus
itsvelocitythen,and inputfromspacebetweent, and t2).How,we might
equallyask,could theEarth'spositionsbetweent, and t2 also affect P2
without violating Kepler'slaws?Yet theymustaffect P2 ifp, does,forp,
itselfcomesbetweent2and stillearlierpositions po which,givenKepler's
laws,also determine P2. There'snothing specialabouttl.
The solutionto thispuzzle lies in the counterfactual conditional (C)

41 See e.g. R. Kirk,Translation Oxford,ClarendonPress,I986, pt.III.

Determined,
42 E.g. N. Chomsky, in D. Davidsonand J. Hintikka(eds),
'Quine's EmpiricalAssumptions',
Wordsand Objections,Dordrecht, Reidel,I969.
4
R. Kirk,op. cit. p. 136.
4 'MentalEvents',loc. cit.,p. 224.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No QuestionofPhysicalism203
whichthiscausal claim entails:ifp1 had been different,so would p2-but
po would not. In other words, what violates Kepler's laws is only (C)'s
counterfactualantecedent. (C) itself does not violate them, and nor
thereforedoes the causal claim which entailsit: indeed Kepler's laws are
what make (C), and hence the causal claim, true.
Similarly in our original case. Our mental states, intentionaland
otherwise,could-and would-affect our brain states and bodily move-
mentseven ifthe laws of physicsmade themall determinedalso by earlier
brainstates.The claim thata systemthusconstrainedby non-mentallaws
must be closed, in the sense of being unaffectableby its mental states,
simplydoes not follow-and it is not true.

5. Supervenience
We have seen that neitherlaws nor causation deprive psychologyof the
ontologicalauthorityof non-mentalsciences.But thatstillleaves one non-
vacuous interpretationof physicalism.The last refuge of the modern
physicalistis supervenience:
the thesisthatthereis no change or difference
withouta non-mentalchange or difference. Two thingswill neverchange
or differin any way withoutalso changingor differing in some non-mental
way. The physicalexcludes the mentalby being thaton whicheverything
else, includingthe mental,supervenes.
Supervenienceis strongerthan the trivialclaim that everythingex-
tendedin space has physicalparts,but weakerthanreductionism,since it
says nothing about which non-mental differencewill accompany any
mentalone: it does not entailtheexistenceofany psychophysicallaws. But
it mustbe strongerthanwe have so farindicated.For giventhe multitude
of changeable non-mentalpropertieswhich any thinghas (includingits
spatiotemporallocation), all things that change or differmentallyare
bound to change or differin fact in some non-mental respect. So
supervenience,to be serious, must mean more than that. The relevant
range of non-mentalrespects must be restricted(at least by excluding
spatiotemporallocation), and the claim must be at least subjunctive-
'Two things would never differ... '-and arguably even stronger-
'Two thingscouldneverdiffer. ..'.
However, to give superveniencea run forits money,we will take it as
weakly as we can: in its subjunctiveform,and with the relevantnon-
mentalrespectsrestrictedas littleas possible. Even so, we see no reasonto
believe it. The evidence forit cannot be empirical,since the prospectof
everfindingtwo things,complexenoughto have psychologicalproperties,
type-identicalin everyreasonablenon-mentalrespect,is extremelyslight,
to say theleast.The onlyremotelyplausibleargumentforsupervenienceis
one whichappeals to thecausal principlementionedin section3, thatthere
is no unmediatedaction at a distance.This means, as we saw there,that

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
204 Tim Crane and D. H. Mellor

tokens of beliefs and other intentionalmental states need intrinsic

propertiesto act as causal surrogatesfor their contents.And if these
propertiesare all non-mental,and sensationsare likewisedeterminedby
theirnon-mentalcauses, then superveniencemay well seem to follow.
But it does not. First, as we have already observed, the intrinsic
propertieswhichact as causal surrogatesforthe contentsof tokenbeliefs
and other intentionalstates may very well be mental. And secondly,
whethertheyare mentalor not, beliefswill still not superveneon them.
For two thinkerscould easilyhave all thesame intrinsicpropertiesand still
have differentbeliefs.This is an obvious.moralof Putnam's 'Twin Earth'
stories:45the contentof yourTwin Earth duplicate's beliefthatwateris
wet differsfromyours,because his (or her) water is XYZ and yours is
H20. And similarlyforindexicalbeliefs.If the contentof 'That's an elm'
includes the tree you look at as you thinkit, it will differfortwo people
looking at differenttrees, even if they have all the same intrinsic
properties.
The defenderof superveniencemightrespondthatthisonlyshows that
thoughtsdo not superveneon theirthinkers'intrinsicproperties.They
mightstill superveneon those plus thinkers'non-mental(e.g. spatiotem-
poral) relationsto otherthings,and those things'non-mentalproperties
(being an elm, or H20). But that is not true either,as we can see by
consideringhow thinkersmake mistakes.Suppose forexample that you
and your intrinsicallyidenticaltwin now look at the same elm, but that
althoughthismakes you thinkit is an elm, it makes him or her thinkit is
an oak. Same intrinsicproperties,same relations,same propertiesof the
thingthoughtabout: but different thoughts.
Again,the defenderof superveniencemightrespondthatin such a case
therewould alwaysbe some relevantnon-mentaldifference: ifnot in your
eyes, then in how the tree looks fromyour different viewpoints.But we
doubt this. You and yourtwinmightwell differin only mentalrespects:
forexample,in yourbeliefsabout whatelms look like-beliefs whichneed
supervene on nothingpresent or non-mental,merely on the different
mental effectstrees have had on you in the past. And we see no non-
question-beggingreasonto thinkthatthoseeffectsmustsuperveneon past
non-mentaldifferences.
On the other hand, your and your twin's past experiencesdo have
presenteffects:theymake you think'That's an elm', and yourtwinthink
'That's an oak'. And beingat a temporaldistance,theycannothave those
effectsimmediately:their effectsmust be mediated by some present
intrinsicpropertiesof you and of your twin. So perhaps your thoughts
must superveneon your intrinsicpropertiesafterall?
4 H. Putnam,'The Meaningof "Meaning"',in his Mind,Languageand Reality,Cambridge,
CambridgeUniversity by T. Burge,'Individualism
Press,1975.The moralis drawnexplicitly and
Psychology',
Philosophical
Review,1986.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
 Thereis No QuestionofPhysicalism205
Not so. For not only,as we have seen, may these mediatingproperties
themselvesbe mental,but even if theyare not, theyneed not differjust
because theirmentaleffectsdo. Causation need not,afterall, be determin-
istic,and modernphysicstells us that it oftenis not. So we have every
reason to expect some indeterminismin the causal processes of our
perception,our reasoning,and our action: this being one way in which
these processes can go wrong and make us make mistakes.But when
causationis indeterministic, causes and effectswill not superveneon each
other.In short,modernphysicsgives us reason to denythe supervenience
of thecontentsof our tokenthoughtson even themostextensivelistof our
otherintrinsicand extrinsicpropertiesand relations.
And as forthoughts,so forsensations.Their havingnon-mentalcauses
or effectswill notmakethemsupervenient.On thecontrary,iftherelevant
causation is somewhat indeterministic, sensations cannot supervene on
theirnon-mentalcauses.
Yet again, however, the defendersof superveniencemay reply that
causation, unlike supervenience,takes time-and we agree.46 Causes
always precede theireffects,whereas token thoughtsand sensationsare
only supposed to supervene on simultaneoustokens of non-mental
properties.So showingthat theydo not superveneon theirearliernon-
mentalcauses does not directlyrefutethatclaim.
But it does refuteit indirectly.For suppose an intrinsicnon-mental
propertyP causes a mental propertyM indeterministically. (Say for
examplethatone's chance of beingM at t2 is o.9 ifone has just been P (at
t1), and o.i if one has not.) Now suppose thatat t1 manypeople share all
theirintrinsicnon-mentalproperties,includingP. At t2, therefore,most
but not all of themwill be M: thatis, some pairsof people, atom-for-atom
alike at t1, will differat t2 in this mentalrespect.
Now let a and b be any such pair: at t2, a is M and b is not. What about
a's and b's intrinsicnon-mentalpropertiesat t2? Well, these may all be
determinedby a's and b's shared non-mentalstate at t1. But if so, then
theytoo will all be shared,and M will not superveneon themeither.But
M will not superveneon themanyway.For even if some relevantlaws of
physicsare indeterministic, so that a's and b's state at t1 does not make
them share all theirintrinsicnon-mentalpropertiesat t2, it still will not
stopthemdoing so. On the contrary:given enough such as and bs, some
will certainlydiffermentallyat t2 withoutdiffering in any otherway.
In otherwords,modernindeterministic physicsmustpredictthatsome
pairs of people, atom-for-atom alike in all non-mentalrespects,will differ
in some simultaneousmentalrespects:and will do so preciselybecause the
the propertiesinvolved are causally related. In short, modern physics
suggeststhat even the weakest serious formof supervenience,which is

46 D. H. Mellor,Real Time,CambridgeUniversity
Press,I98I, chs9-IO.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions
206 TimCraneand D. H. Mellor
itselftheweakestnon-vacuous
formofphysicalism, is false.Andphysical-
istscan surelynotexpecta physicalism
thatis falsified
by physicsto be
verifiedby anything
else.

6. The end
We havearguedthatno defensible of physicalism
definition willdeprive
psychology oftheontological
statusofthenon-mental sciences.In no non-
vacuoussenseis physicalism true.But thisdoesnotmeanthatwe wantto
encourage a revivalofCartesiandualism.On thecontrary, ourarguments
entailthatthereis no dividebetweenthe mentaland the non-mental
sufficientevento set physicalismup as a seriousquestion,let aloneas a
seriousanswerto it. Physicalism is the wronganswerto an essentially
trivialquestion.So it cannotbeginto helpphilosophers of mindanswer
theseriousquestionsaboutthemindand,aboveall,aboutintentionality:
whatenablessome partsof the world(us) to thinkabout otherparts,
including otherpeople(andofcourseourselves). Andto thosequestionsit
is quite obviousthatneitherdualismnor physicalism has anything to
contribute. The dualistdoesnoteventryto explainintentionality: he just
takesitforgranted, itintoexistence.
stipulating Andsayingthatmindsare
all physicalno morehelpsto explainhowsomephysicalthingscan think
thansayingthatall fleshis grasshelpsto explainthedifference between
carnivores and vegetarians.
This,therefore,shouldreallybe thelastpaper
on thesubjectof physicalism. But we fearit willnotbe.47
Department
Philosophy TIM CRANE
King's College
Strand
London WC2R 2LS
DarwinCollege D.H. MELLOR
SilverStreet
Cambridge CB3 9EU

4 The material (T. M. Crane,

in thispaperis derivedpartlyfromone author'sPh.D. dissertation
op. cit.)and partlyfrommaterial presented bytheotherauthorto meetings oftheBritishSocietyfor
thePhilosophy of Science,theOxfordUniversity Society,theCambridgeUniversity
Philosophical
PPE Workshopand of seminarsat UniversityCollege London, the University of Sussex, the
University ofCalifornia at Irvine,theUniversity
ofWisconsinat Madison,NorthWestern University
and theHistoryand Philosophy of ScienceDepartment in Cambridge.We are grateful forhelpful
comments and criticismmadebymanypeopleon thoseoccasionsand in manyprivateconversations
and correspondence.

This content downloaded from 143.167.80.218 on Mon, 22 Sep 2014 06:42:58 AM

All use subject to JSTOR Terms and Conditions