https://ebookmass.com/product/neurocognitive-mechanismsexplaining-biological-cognition-gualtiero-piccinini/
Instant digital products (PDF, ePub, MOBI) ready for you
Download now and discover formats that fit your needs...
Physical computation : a mechanistic account Piccinini
https://ebookmass.com/product/physical-computation-a-mechanisticaccount-piccinini/ ebookmass.com
Cognition Marvin Chun
https://ebookmass.com/product/cognition-marvin-chun/
ebookmass.com
eTextbook 978-0205985807 Cognition
https://ebookmass.com/product/etextbook-978-0205985807-cognition/
ebookmass.com
Emergency Imaging of At-Risk Patients: General Principles 1st Edition Michael N. Patlas
https://ebookmass.com/product/emergency-imaging-of-at-risk-patientsgeneral-principles-1st-edition-michael-n-patlas/
ebookmass.com
Weird Fiction: A Genre Study Michael Cisco https://ebookmass.com/product/weird-fiction-a-genre-study-michaelcisco/ ebookmass.com
Fighting the Opioid Epidemic: The Role of Providers and the Clinical Laboratory in Understanding Who is Vulnerable 1st Edition Amitava Dasgupta
https://ebookmass.com/product/fighting-the-opioid-epidemic-the-roleof-providers-and-the-clinical-laboratory-in-understanding-who-isvulnerable-1st-edition-amitava-dasgupta/ ebookmass.com
The No-Code Startup Emma Reilly
https://ebookmass.com/product/the-no-code-startup-emma-reilly/
ebookmass.com
What to Believe Now: Applying Epistemology Contemporary Issues 1st Edition – Ebook PDF Version
https://ebookmass.com/product/what-to-believe-now-applyingepistemology-contemporary-issues-1st-edition-ebook-pdf-version/ ebookmass.com
Small Habits for a Big Life: Release the self-sabotaging behaviour that is holding you back, one habit at a time
Rebecca Ray
https://ebookmass.com/product/small-habits-for-a-big-life-release-theself-sabotaging-behaviour-that-is-holding-you-back-one-habit-at-atime-rebecca-ray/ ebookmass.com
Use and Reuse of the Digital Archive 1st ed. 2021 Edition
https://ebookmass.com/product/use-and-reuse-of-the-digitalarchive-1st-ed-2021-edition-potts/
ebookmass.com
NeurocognitiveMechanisms Neurocognitive Mechanisms ExplainingBiologicalCognition GUALTIEROPICCININI GreatClarendonStreet,Oxford,OX26DP, UnitedKingdom
OxfordUniversityPressisadepartmentoftheUniversityofOxford. ItfurtherstheUniversity’sobjectiveofexcellenceinresearch,scholarship, andeducationbypublishingworldwide.Oxfordisaregisteredtrademarkof OxfordUniversityPressintheUKandincertainothercountries ©GualtieroPiccinini2020
Themoralrightsoftheauthorhavebeenasserted FirstEditionpublishedin2020
Impression:1
Allrightsreserved.Nopartofthispublicationmaybereproduced,storedin aretrievalsystem,ortransmitted,inanyformorbyanymeans,withoutthe priorpermissioninwritingofOxfordUniversityPress,orasexpresslypermitted bylaw,bylicenceorundertermsagreedwiththeappropriatereprographics rightsorganization.Enquiriesconcerningreproductionoutsidethescopeofthe aboveshouldbesenttotheRightsDepartment,OxfordUniversityPress,atthe addressabove
Youmustnotcirculatethisworkinanyotherform andyoumustimposethissameconditiononanyacquirer
PublishedintheUnitedStatesofAmericabyOxfordUniversityPress 198MadisonAvenue,NewYork,NY10016,UnitedStatesofAmerica
BritishLibraryCataloguinginPublicationData Dataavailable
LibraryofCongressControlNumber:2020947366
ISBN978–0–19–886628–2
Printedandboundby CPIGroup(UK)Ltd,Croydon,CR04YY
LinkstothirdpartywebsitesareprovidedbyOxfordingoodfaithand forinformationonly.Oxforddisclaimsanyresponsibilityforthematerials containedinanythirdpartywebsitereferencedinthiswork.
Acknowledgments vii ListofFigures xi Introduction1
1.LevelsofBeing6
2.Mechanisms,MultipleRealizability,andMediumIndependence38
3.FunctionalMechanisms67
4.MechanisticFunctionalism89
5.TheFirstComputationalTheoryofCognition:McCulloch andPitts’ s “ALogicalCalculusoftheIdeasImmanentin NervousActivity” 107
6.ComputationandInformationProcessing128
7.MechanisticModelsofCognition:FromAutonomytoIntegration156
8.TheCognitiveNeuroscienceRevolution182
9.TheComputationalTheoryofCognition205 10.TheChurch–TuringFallacy225
11.TheResilienceoftheComputationalTheoryofCognition244
12.NeuralRepresentation258
13.NeuralComputation297
14.ComputationandtheFunctionofConsciousness317 Bibliography 351 Index 393
Acknowledgments MyundergraduatestudiesattheUniversityofTurin,duringtheearly1990s, exposedmetotheclassicdebateonthefoundationsofcognitivescience.Isthe brainacomputer?Doescognitioninvolvecomputationoverrepresentations?Is themindthesoftwareofthebrain?Therewereargumentsonbothsides,and Iwantedtosortthemout.
WhileingraduateschoolattheUniversityofPittsburgh,duringthelate1990s andearly2000s,Iembarkedonaresearchprogram: first, figureoutwhatphysical computationisbydrawingfromcomputerscienceandcomputerengineering; then,applythatunderstandingtothebrainandanswerthefoundationalquestions aboutcognition.Forgoodmeasure,Iincludedacarefulstudyoftheoriginof computationaltheoriesofcognition.Thatwaswaytoomuchforadissertation. Mydissertationincludedsomehistoricalresearchandanaccountofphysical computation(Piccinini2003a).Ileftthefoundationalquestionsforfuturework. Myearlymechanisticaccountofphysicalcomputationwas,roughly,that physicalcomputationisthemanipulationofdigitsbyafunctionalmechanism inaccordancewitharule(Piccinini2007a).Ithoughtthisaccountwouldallowme torefutetheComputationalTheoryofCognition(CTC)onceandforall.Iwrotea paperattemptingthat.Myargumentwasthatcomputationisdigital;neural activityisnotdigital;therefore,neuralactivityisnotcomputational.ButInever publishedthatpaperbecauseIeventuallyrealizedIwasmissingsomething.
Iknewthereusedtobeanalogcomputers,andtheyhadbeenusedasan alternativemodelofbrainactivity.Ialsoknewthatanalogcomputersarequite differentfromdigitalcomputers,arenotprogram-controlled(quaanalog),arenot universalinAlanTuring’ssense,existedbeforedigitalcomputers,andusedtobe calleddifferentialanalyzers.Theyonlystartedtobecalledanalog “computers” after digitalcomputersbecamepopular.So,Iarguedthatanalog “computers” weren ’tcomputersproperlysocalledafterall(Piccinini2008a).Atleasttwo eventspushedback.
First,thelateJonathanMillsaskedmetocollaborateonexplicatinganalog computation.AlthoughheandIneverwroteanythingtogether,heintroduced metothegroundbreakingresearchonanalogcomputationthathewasdoingat IndianaUniversity,ofwhichIwasunaware.Independently,inJuly2007,Marcin Milkowskiwrotetomeandpointedoutthatsomecomputerscientistswerereviving analogcomputation;herecommendeda “humblerapproach” thanmine.
Iwonderedwhetherthosewhostudyanalogcomputation,andperhapsothers whostudyotherunconventionalmodelsof(so-called)computationthatarenot
digital,mightbepickinguponsomethingthatdigitalcomputation,analog “computation,” andothertypesof “computation” haveincommon.Fromamathematicalstandpoint,theyareallwaysofsolvingmathematicalproblems.ButIwas afteranaccountof physical computation.Whatdothedisparatephysicalprocesses thatimplementdifferenttypesofcomputationhaveincommon?
Thereceivedviewisthatcomputationhastodowithrepresentation.ButIhad argumentsthat,infact,computationdoesnotrequirerepresentation(Piccinini 2004a,2008b).Ialsoknewthatthenotionsoffunctionandmultiplerealizability aretooweaktounderwritearobustnotionofcomputation(Piccinini2004b, 2007b).Atsomepoint,IrememberedreadinganinsightfulpaperinwhichJustin Garson(2003)usesthenotionofmediumindependencetocharacterizeinformation.(JohnHaugeland(1985)usesasimilarnotiontocharacterizeautomatic formalsystems,butIonlyfoundoutaboutthatmuchlater.)Ire-readJustin’ s paperandfoundthatmediumindependencewasjustwhatIneeded.Afterthat, Igeneralizedmyaccountofphysicalcomputationtocoveranalogcomputation andotherunconventionalmodelsofcomputation.Theresultingaccountisthat physicalcomputationisthemanipulationofmediumindependentvehiclesbya functionalmechanisminaccordancewitharule(PiccininiandScarantino2011, Piccinini2015).
Giventhisbroaderunderstandingofphysicalcomputation,myattitudeabout thebrainandcognitionswitched.Inowhadthematerialsforacogentargument infavorof CTC withatwist.WithhelpfromneuroscientistSonyaBahar,IrewrotemyearlypaperagainstCTCtoarguethat,ononehand,neurocognitive processesarecomputationsbecausetheyaremediumindependentand,onthe otherhand,neuralcomputationsareneitherdigitalnoranalog theyaresui generis(PiccininiandBahar2013).
Thatseemedimportantbut,toanswerthefoundationalquestions,muchmore neededtobedone.Ineededaproperontologicalfoundation,anadequateaccount offunctions,awaytointegratepsychologyandneuroscience,anassessmentof otherargumentsproandconCTC,andaplaceforrepresentationandconsciousness.Aftermanymorecollaborationsandpublicationsandmuchadditional research,thisbookistheresult.ItpresentsacomprehensivedefenseofCTC, updatedfortheeraofcognitiveneuroscience,withsurprisesforbothdefenders andcriticsoftraditionalformsofCTC.
WaymorepeoplehavehelpedmethanIcanthankindividually.Icollectively thankmyteachersaswellasthemanyaudiencestowhomIpresentedversionsof theseideas.Ilearnedalotfromthem.
Myworkbuildsonthosewhocamebeforeme.BesidesthoseIalreadymentionedandmycollaborators,thosewiththegreatestphilosophicalinfluenceon thisprojectincludeBillBechtel,ChristopherBoorse,PaulandPatricia Churchland,JackCopeland,CarlCraver,RobertCummins,DanielDennett, FredDretske,FrankieEgan,JerryFodor,GilbertHarman,JohnHeil,Jaegwon
Kim,RuthMillikan,HilaryPutnam,JonathanSchaffer,OronShagrir,Wilfried Sieg,andStephenStich.Onthescienceside,thosefromwhomIlearnedthemost includeWarrenMcCulloch,AllenNewell,WalterPitts,ClaudeShannon,Herbert Simon,AlanTuring,andJohnvonNeumann.
Forcommentsonmanypapersandilluminatingdiscussionsovermanyyears IowespecialthankstoNealAnderson,KenAizawa,TreyBoone,CarlCraver, CoreyMaley,AndreaScarantino,OronShagrir,MarkSprevak,JohnHeil,Justin Garson,andEricThomson.Forcommentsonaspectsofthiswork,Ithanks DarrenAbramson,AnibalAstobiza,BriceBantegnie,SergioBarberis,BillBechtel, PeterBradley,GiovanniCamardi,GlennCarruthers,DavidChalmers,Mazviita Chirimuuta,MichelleCiurria,JudithCrane,RobertCummins,TonyDardis,Joe Dewhurst,FrankieEgan,ChrisEliasmith,IlkeErcan,NirFresco,RachelFrancon, ErikFunkhouser,CarlGillett,StuartGlennan,MahiHardalupas,InmanHarvey, EricHochstein,PhyllisIllari,AnneJacobson,DavidM.Kaplan,DanielKramer, ArnonLevy,BillLycan,PeterMachamer,JackMallah,CoreyMaley,Diego Marconi,MarcinMilkowski,KevinMorris,AlyssaNey,AbelWajnermanPaz, AlessioPlebe,RussPoldrack,TomPolger,CharlesRathkopf,MichaelRescorla, SarahRobins,WaldemarRohloff,RobertRupert,Anna-MariRusanen,Kevin Ryan,MatthiasScheutz,SusanSchneider,WhitSchonbein,AlexSchumm,Paul Schweizer,SamScott,OronShagrir,LarryShapiro,KentStaley,TerrenceStewart, JackieSullivan,BrandonTowl,CharlesWallis,DanWeiskopf,GeneWitmer, MarkZeise,manyanonymousreferees,andothersIcan’trecall.Specialthanks toMichaelBarkasi,MarkCouch,andtheanonymousrefereesforOUPfor insightfulcommentsonthebookmanuscript.
ThankstoCrystalBrown,FrankFaries,MirindaJames,ElliottRisch,andJames Virtelforeditorialassistance.
TheIntroduction,Chapter1,andChapter3arenew.
Chapter2includesasubstantiallyrevisedandexpandeddescendantof PiccininiandMaley2014.Sections2.1,2.3,and2.5arenew.
Chapter4includesasubstantiallyrevisedandexpandeddescendantofMaley andPiccinini2013.Sections4.1.1and4.1.2arenew.
Chapter5isarevisedandexpandeddescendantofpartsofPiccinini2004c.
Sections6.2–6.8ofChapter6arearevisedandexpandeddescendantof Piccinini2017a.Section6.9isareviseddescendantofasectionofPiccinini 2009.Sections6.1and6.10–6.12arenew.
Chapter7includesarevisedandexpandeddescendantofBooneandPiccinini 2016a.Sections7.1–7.3and7.7arenew.
Chapter8isareviseddescendantofBooneandPiccinini2016b.
Sections9.1and9.2ofChapter9includerevisedversionsofportionsof Piccinini2016.MostofSection9.2isarevisedandexpandeddescendantofa sectionofPiccininiandBahar2013.Sections9.3–9.5areareviseddescendantof partsofPiccinini2009.
Chapter10isasubstantiallyreviseddescendantofPiccinini2007c.
Chapter11isasubstantiallyrevisedandexpandeddescendantofPiccinini 2010a.
Chapter12isasubstantiallyreviseddescendantofThomsonandPiccinini 2018.
Chapter13isarevisedandexpandeddescendantofmuchofPiccininiand Bahar2013.
Chapter14includesasubstantiallyreviseddescendantofPiccinini2010b. Sections14.1and14.9arenew.
Itwasacollectiveeffort.Iamdeeplygratefultomyco-authorsonsomeofthe articlesfromwhichsomeofthechaptersderive SonyaBahar,TreyBoone,Corey Maley,andEricThomson twophilosophersandtwoneuroscientists.They graciouslyagreedtoletmeuseourjointworkinthisbookanddeservecredit formuchofwhatiscorrectinthechaptersderivedfromit.Myresearchprogram benefitedimmenselyfromeachcollaboration.
ThismaterialisbaseduponworksupportedbytheNationalScience FoundationunderGrantsNo.SES-0216981,SES-0924527,andespeciallySES1654982.IalsoreceivedresearchsupportfromaUniversityofMissouriResearch BoardAward,a2006NEHSummerSeminaratWashingtonUniversityinSt. Louis,aUniversityofMissouriResearchGrant,agrantfromtheInternational StudiesandProgramsattheUniversityofMissouri St.Louis,anAdelleand ErwinTomashFellowship,anAndrewMellonPredoctoralFellowship,anda RegioneSardegnaDoctoralScholarship.Anyopinions, findings,conclusions, andrecommendationsexpressedinthisworkarethoseoftheauthoranddo notnecessarilyreflecttheviewsofthesefundinginstitutions.
ThankstoPeterMomtchiloffandthewholeteamatOUPforshepherdingmy manuscriptthroughthepublicationprocess.
Deepthankstomyfriendsandfamilyfortheirloveandsupportduringallthese years.I’mgratefultomypartnerMichelleanddaughtersViolet,Brie,andMartine forbringingsomuchmeaningandjoytomylife.
ListofFigures 1.1Possiblerelationsbetweenhigher-levelpropertiesandlower-levelproperties26
1.2Waysofrepresentingawholeanditsparts33
1.3Wholesareinvariantsovercertainchangesinparts,propertiesofwholes areaspectsoftheirlower-levelrealizers35
2.1Theoverlapbetweenlower-levelproperty1andlower-levelproperty2isa higher-levelpropertyrealizedbybothproperties1and2,andsoforth39
2.2Higher-levelpropertiesareaspectsoftheirlower-levelrealizers.49
2.3Theintersectionsoflower-levelpropertiesarevariablyrealizable higher-levelproperties53
2.4Differenttypesofhigher-levelproperties63
5.1DiagramsofMcCullochandPittsnets114
5.2Netexplainingtheheatillusion115
6.1Typesofdigitalcomputationandtheirrelationsofclassinclusion147
6.2Typesofcomputationandtheirrelationsofclassinclusion150
7.1Graph-theoreticrepresentationofalateralinhibitioncircuit175
9.1SomeprominentformsofCTCandtheirrelations223
12.1Retinalrepresentationofthevisualworld270
12.2Mainvisualpathway271
12.3TopographicrepresentationofmotioninmonkeyareaMT273
12.4Binocularrivalrydemonstration275
12.5Workingmemory:behavioralandneuronalperspectives277
12.6Frommusclestoactionmaps283
12.7Movement fieldinM1285
12.8SpaceofM1representationaltheories286
12.9Sensorycancellationinweakelectric fish291
12.10Forwardmodelsandefferencecopy293
13.1Analog,digital,andneuralsignals299
Introduction Thisbookdefendsaneurocomputationaltheoryofcognitiongroundedina mechanistic,functionalist,egalitarianontology.Iarguethat biologicalcognitive capacitiesareconstitutivelyexplainedbymultilevelneurocognitivemechanisms, which performneuralcomputationsoverneuralrepresentations.Providinga scientificexplanationofcognitionrequiresunderstandinghowneurocognitive mechanismswork.Therefore,thescienceofcognitionoughttoincludeneurosciencetoadegreethattraditionalcognitivesciencewasnotexpectedto.Scientists onthegroundhavebeenworkingonthisforawhile.Psychologyisbecoming moreandmoreintegratedwithneuroscience.
ThepictureIdefendstandsoppositetobothtraditionalreductionism(typeidentitytheory)andanti-reductionism(autonomyofpsychology).Contraryto traditionalreductionism,neuralcomputationsandrepresentationsarenotidenticaltotheirlower-levelrealizers.Contrarytotraditionalanti-reductionism, neuralcomputationsandrepresentationsarenotentirelydistinctfromtheir realizers.Instead,neuralcomputationsandrepresentations bothtypesand tokens are aspects oftheirrealizers.
ThepictureIdefendstandsoppositetobothcomputationalchauvinism(computationandrepresentationareproprietarypsychologicalnotions)andantirealism(computationsandrepresentationsaremeremannersofspeaking). Contrarytocomputationalchauvinism,computationsandrepresentationsare propertiesofthenervoussystem.Contrarytoanti-realism,computationsand representationsarereal,causallyefficaciousproperties asrealasanyother propertiesofthenervoussystem.
Iwillbeginbyprovidingaccountsoflevelsofcompositionandrealization, mechanisms,functions,computation,andmultilevelneurocognitivemechanisms. Thisontologicalfoundationwillallowmetoimproveonexistingversionsof functionalismandmakeclearwhatthecomputationaltheoryofcognitiondoes anddoesnotsay.Afterthat,Iwilladdresssomefallaciesandobjectionstothe computationaltheoryofcognition.Finally,Iwillprovideempiricalevidencethat neurocognitivesystemsperformsuigeneriscomputationsoverneuralrepresentations.Iwillconcludebyclarifyingtherelationbetweencomputationand consciousnessandofferinganoncomputationalfunctionalismaboutphenomenal consciousness.
By capacity,Imeanacausalpowersubjecttonormativeevaluation apower thatcanbemanifestedcorrectlyorincorrectly.
NeurocognitiveMechanisms:ExplainingBiologicalCognition.GualtieroPiccinini,OxfordUniversityPress(2020). ©GualtieroPiccinini. DOI:10.1093/oso/9780198866282.001.0001
By cognitive capacities,Imeancapacitiessuchasperception,memory,reasoning,emotion,language,planning,andmotorcontrol.Cognitivecapacitiesexplain someofthemostinterestingbehaviorsexhibitedbyphysicalsystems.
By biological cognition,Imeancognitioncarriedoutbybiologicalorganisms morespecifically,organismswithaspecializedcontrolorgan.Thespecialized controlorganofearthlyorganismsisthenervoussystem.Somephilosophers arguethatplantshavecognitivecapacities(Calvo2016).Plants,fungi,andeven bacteriaareexquisitelyadaptedtotheirenvironments.Yettheylackaspecialized organforcontroloftheirorganismicfunctionsbasedonprocessinginformation thatresultsfromintegratinginputsfrommultiple,physicallydistinctsources.My topicisthecognitivecapacitiesoforganismswithaspecializedcontrolorganthat integratesmultiplesourcesofinformation.
Therearealsoartifactssuchasrobotsanddigitalassistantsthatpossess computationalandinformationprocessingcapacities,andthereareanalogies betweensuchartificialsystemsandbiologicalcognizers.Thisbookfocuseson biological cognitionexhibitedbyorganismswithanervoussystem.
Somearguethattherealizersofcognitivestatesandprocessesinbiological organismsincludenotonlythenervoussystembutalsosomethingsoutsideit (e.g.,Wilson1994).Iwillmostlyignorethispossibilitytosimplifytheexposition; thisdoesnotaffectmyarguments.
By constitutiveexplanation,Imeanexplanationofacapacityofasystemin termsofthesystem’scausalstructure.Iwillarguethatconstitutiveexplanationis providedby mechanisms thatis,pluralitiesofcomponents,componentfunctions,andorganizationalrelationsthat,collectively,possessthecapacityand produceitsmanifestations(Chapters3,7).
Typically,componentsofmechanismsarethemselvesmechanismswhose capacitiesareexplainedmechanistically.Dittoforthecomponents’ components. Thismultilevelmechanisticstructurerequiresanontologicalfoundation.Iargue thatall levels areequallyreal:neitherhigherlevelsnorlowerlevelsaremore fundamentalthanoneanother.Higher-levelobjectsareinvariantsunderthe additionandsubtractionofsomeparts,whilehigher-levelpropertiesareaspects oftheirlower-levelrealizers.Adequateconstitutiveexplanationrequiresidentifying thehigher-levelpropertiesandorganizationalrelationsthatproduceacapacity. Therefore,adequateconstitutiveexplanationrequiresidentifyingappropriate higher-levelobjectsandtheirrelevantproperties(Chapter1).
Capacitiessuchascognitionarecausalroles thatis,theyarehigher-level propertiesdefinedsolelybytheirregulareffectsundercertainconditions.Causal rolesareoften multiplyrealizable,meaningthattherearedifferentkindsof mechanismthatcanperformthesamecausalrole.Inaddition,somemultiply realizablecausalrolesare mediumindependent,meaningthateventheirinputs andoutputsaremultiplyrealizable(Chapter2).
Causalrolesareone,broadnotionoffunction.InChapter3,Iarguethatthe notionofcausalroleistoobroadtodojusticetomechanisticexplanationof organismsandartifacts.Thecapacitiesoforganismsandartifactsareexplainedby causalrolesthatmakeregularcontributionstothegoalsoforganisms.These contributionsare teleologicalfunctions (Chapter3).
Onthisbasis,Iproposeanimprovedformulationof functionalism:theview thatthemindisthefunctionalorganizationofthebrain.Iarguethatfunctional organizationshouldbeunderstoodmechanistically,asencompassingnotonly causalrelationsbetweeninternalstatesbutalsothecomponentsbearingthestates, theirfunctions,andtheirorganizationalrelations(Chapter4).
Functionalismisacloseallyofthecomputationaltheoryofcognition,which was firstproposedbyWarrenMcCullochandWalterPittsin1943.Theycreateda neuralnetworkformalismforrepresentingneuronalactivityinasimplifiedand idealizedway,andtheyarguedthatneuralnetworksperformdigitalcomputations.Theirideahadanimmenseimpactonthe field(Chapter5).
Somemechanisms,suchastheneuralnetworksdevisedbyMcCullochand Pitts,havethespecialteleologicalfunctionsof computing and processinginformation.Computingisprocessingmedium-independentvehiclesinaccordancewitha rule,whileinformationprocessingistheprocessingofvehiclesthatcarryinformation.Computationalvehiclesmayormaynotcarryinformation,andinformationprocessingmayormaynotbedonebycomputing(Chapter6).
Manyauthorshavearguedthatthereareconstitutiveexplanations whether computationalornot thatarenotmechanistic.Iarguethatsuchputatively nonmechanisticconstitutiveexplanationsare aspects ofmechanisms.Adequate constitutiveexplanation includingcomputationalexplanation involvesmechanisms(Chapter7).
Thescientificstudyofneurocognitivemechanismsiscognitiveneuroscience. Toalargeextent,cognitivesciencehasalreadyturnedintocognitiveneuroscience. Thisistheintegratedstudyofhowmultilevelneurocomputationalmechanisms thatprocessneuralrepresentationsexplaincognition(Chapter8).
The ComputationalTheoryofCognition (CTC)isthetheorythatcognitive processesarecomputations,orcognitionisexplainedcomputationally.Since computationisamechanisticprocess,CTCisamechanistichypothesis.Since theorganofbiologicalcognitionisthenervoussystem,CTCforbiological organismsistheclaimthatneurocognitiveprocessesarecomputations.Iargue thatCTCinitsgenericformulationiscorrectforacoupleofreasons.First,the mainvehiclesofneuralprocessing spiketrains arefunctionallysignificant thanksprimarilyto firingratesandspiketiming,whicharemediumindependent. Second,neuralsignalscarry,integrate,andprocessinformationfromphysically differentsources,whichrequirestransducingthemintoshared,mediumindependentvehicles(Chapter9).
AssessingCTCrequiresunderstandingnotonlythereasonsforit,butalsowhat is not areasonforit.ArangeofputativeargumentsforCTCarebasedonthe Church–Turingthesis,thethesisthatthefunctionsthatarecomputableinan intuitivesensearepreciselythosefunctionsthatarecomputablebyTuring machines.Iarguethattheseargumentsarefallacious:theChurch–Turingthesis doesnothelpestablishCTC(Chapter10).
AssessingCTCalsorequiresunderstandingputativeobjections.Therearetwo classesofobjections.Insufficiencyobjectionsmaintainthatcognitioninvolves X(forsomeX)andcomputationisinsufficientforX.CandidateXsinclude consciousness,intentionality,embodiment,embeddedness,dynamics,andmathematicalinsight.Ireplythatinsuf ficiencyobjectionsdonotundermineCTC;at most,theyshowthatsomethingelseisneeded,inadditiontocomputation,to explaincognition.Iemphasizethatanadequatecomputationalexplanationof biologicalcognitioninvolvescomputationsthatareembodiedandembedded. ObjectionsfromneuralrealizationarguethatneuralprocessesinvolveY(for someY)andcomputationdoesnotinvolveY;therefore,neuralprocessesarenot computations.IreplythatnoneoftheputativeY’sunderminesCTC(Chapter11).
Althoughcomputationcanoccurintheabsenceofinformationprocessing,this isnotwhathappensinthenervoussystem.Neuralprocessescarry,integrate,and processinformationintheserviceofteleologicalcontrolfunctions.Informational Teleosemanticsistheviewthatrepresentationsarepreciselystatesthatcarry informationintheserviceofteleologicalfunctions.Thus,Informational Teleosemanticsappliedtotheneurocognitivesystementailsthattheneuralstates thatcarryinformationintheserviceofcontrolfunctionsare neuralrepresentations.Iarguethatneuralrepresentationsarenotonlyrealbutroutinelyobserved byneuroscientists(Chapter12).
Onelong-standingdisputeaboutneuralcomputationiswhetheritisdigitalor analog.Iarguethatitisneither,atleastinthegeneralcase.Instead,neural computationissuigeneris.Therefore,theoriesofcognitionshouldtakeinto accountwhatisknownaboutneuralcomputation(Chapter13).
The ComputationalTheoryofMind (CTM),asIusethisterm,isastrongerview thantheComputationalTheoryof Cognition.CTMcoversthewholemind both cognition and consciousness.Thus,CTMsaysthatthewholemindhasacomputationalnature,orthatallthereistothenatureofmentalstates,including consciousstates,isbeingcomputationalstates.Sincecomputationisamechanistic process,CTMisamechanistichypothesis.Itisthecomputationalversionof mechanisticfunctionalism.Contrarytoacommonassumption,thealternative toCTMisnotjustthetype-identitytheoryofmind.Thereisalsoanoncomputationalversionoffunctionalismaboutconsciousnessthatdeservestobeexplored (Chapter14).
Biologicalcognition,oratleastbiologicalcognitioninorganismswitha centralizedcontrolsystemthatintegratesmultiplesourcesofinformation,turns
outtobeneuralcomputationoverneuralrepresentations.Neuralcomputationis asuigeneriskindofcomputation neitherdigitalnoranalog thatneurocognitivemechanismsperform.Neuralrepresentationsaresimulationsoftheorganism, itsenvironment,andtheirinteractionthatneurocognitivemechanismsconstruct. Neuralcomputationsprocessneuralrepresentations.Theresultofthisprocessis biologicalcognition.
Neurocognitivemechanismsspanmultiplelevelsoforganization,fromsingle neuronsuptothewholenervoussystem.Ateachneurocognitivelevel,neural computationsprocessneuralrepresentations.Exceptforthelowestlevel,each neurocognitivelevelisrealizedbythelevelbelowit.Exceptforthehighestlevel, eachneurocognitivelevelrealizesthelevelaboveit.Thus,lower-levelneural representationsandcomputationsrealizehigher-levelneuralcomputationsand representations.
Needlesstosay,explaininghowcognitivelyendowedorganismsbehave requiresconsideringthedynamicalcouplingamongnervoussystem,body,and environment.Inotherwords,explaininghoworganismsbehaverequiresconsideringhownervoussystemsareembodiedandembeddedintheirenvironment. Consideringallofthisiswhatcognitiveneurosciencedoes.Nevertheless,nervous systemscontributesomethingdistinctivelycognitivetotheexplanation:theperformanceofneuralcomputationsoverneuralrepresentations.Thatisthefocusof thisbook.
Althoughthechaptersaredeeplyintertwined,Ioutlinethemainpointofeach chapterinits firstsection exceptforChapter1.Areaderwhowantsthegistof thebookquicklycanreadChapter1plusthe firstsectionoftheotherchapters.
LevelsofBeing 1.1Parts,Wholes,andtheirProperties Thisbookarguesthatcognitionisexplainedbymultilevelneurocognitive mechanisms.The firststeptowardsthisview oranyviewinvolvinglevelsof compositionandrealization,forthatmatter isclarifyingwhatlevelsareandhow theyarerelated.Thischapterlaysoutanaccountoflevelsonwhichtherestofthe bookwillbuild.
Thewaywetalkaboutthingsandtheir proper¹partsgivesrisetopuzzles.For example,ahammerhastwoparts:ahandleandahead.Ifyouhaveahammer, howmanyobjectsdoyouhave?Ifyouanswerone,youarenotcountingthehead andthehandle.Ifyouanswertwo,youarenotcountingthewholehammer.Ifyou answerthree,youseemtobecountingthesameportionofrealitytwice.Noneof theanswersareentirelysatisfying.Andthentherearethehammer’smolecules, atoms,andsubatomicparticles.Howmanythingsarethere?
Puzzleslikethismakeitdifficulttounderstandhowourdiscourseaboutwholes fitswithourdiscourseaboutparts.Bothlaypeopleandscientiststalkabout differentsortsofobjects,someofwhicharepartofothers.Wetalkaboutpeople, nervoussystems,brains,neuralnetworks,neurons,neurotransmitters,andsoon. Wedescribepropertiesthattheseobjectshaveandtheirmanifestations:people walk,brainsdevelop,andneuronssendactionpotentials.
Wealsodescribe relations betweenobjects.Forconvenience,inthischapter Iwilloftenfollowthecommonconventionofusingtheterm “property” toinclude relations.Forinstance,Iwilltreat AbeinginsideB asaspatialpropertypossessed byAandBcollectively.So,Iwilloftentalksimplyofobjectsandproperties, withtheunderstandingthatrelationsarealsoincluded.Nothingsubstantive hingesonthis.
Somerelationsoccurbetweenobjectsthatarewhollydistinctfromone another objects atthesamelevel:peopletalktooneanother,brainsareencased inskulls,andneuronsattachtooneanother(viasynapses).Otherrelationsoccur betweenobjectsthatarepartofoneanother.Forexample,synapsesarepartof
¹Inmereology theformalstudyofthepart–wholerelation itisconvenienttodefine “part” so thateverythingispartofitself,and “properpart” asapartthatisnotidenticaltothewhole.Bycontrast, inordinarylanguage,weusuallypresupposethatapartisnotidenticaltothewhole.Fromnowon, Iwillfollowordinaryusageanduse “part” tomeanproperpart.
NeurocognitiveMechanisms:ExplainingBiologicalCognition.GualtieroPiccinini,OxfordUniversityPress(2020).
©GualtieroPiccinini. DOI:10.1093/oso/9780198866282.001.0001
neurons,whichtogetherformneuralnetworks;nervoussystemsarepartially madeoutofbrains,whicharepartiallymadeoutofnetworks,amongother structures. Beingpartof somethingand beingmadeoutof somethingarerelations thatoccurbetweenobjects atdifferentlevels ofbeing.
Sortingoutwhatthereis,whatpropertiesobjectshave,andhowtheyrelateis thebusinessofscience.Sortingoutwhat sorts ofobjectsthereareaswellaswhat sorts ofpropertiesandrelationstheyhaveisthebusinessofmetaphysics.Scientists relyonexperimentingandtheorizing.Metaphysiciansrelyonscientificresults andconceptualconsiderations.Byusingthesetools,metaphysiciansattemptto organizeourconceptssowedon’tgetconfusedorsaythingsthatmakelittlesense. Thisbookbeginswithmetaphysicsandprogressesfromtheretowardsthephilosophyofmindandcognitiveneuroscience.
Iwillusetheterm “composition” fortherelationbetweenwholesandtheir parts:partscomposeawhole;awholedecomposesintoitsparts.Iwillusetheterm “realization” fortherelationbetweenpropertiesofawholeandpropertiesof itspartstakencollectively,orbetweenitspartstakencollectivelyandallofthe parts’ partstakencollectively:higher-levelpropertiesarerealizedbylower-level properties;lower-levelpropertiesrealizehigher-levelproperties.² Propertiescomeintypesandtokens(instances).Forexample,eachplanetin thesolarsystemisanoblatespheroid.Thatis,eachplanetinstantiatesthe property type ofoblatespheroidicity.Equivalently,eachplanethasitsown token,orinstance,ofoblatespheroidicity.Wecantalkabouttyperealizationor tokenrealization.Iwillmostlyfocusontyperealization,withoutmakingitexplicit everytime.MuchofwhatIsayappliestotokenrealizationaswell.
AsIusetheseterms,bothcompositionandrealizationareasymmetric,irreflexive,andtransitive.³Botharerelationsofsynchronicmetaphysicalnecessitation:onceyouhavethepartswiththeirpropertiesandrelations,younecessarily havethewholewithitspropertiesandrelations.Necessitationdoes not entail dependence!Bothcompositionandrealizationareoften assumed toberelationsof ontologicaldependence,meaningthatwholesandtheirpropertiesontologically dependon(aregroundedin,areposteriorto)theirpartsandtheirproperties.Not onlyamI not makingthisdependenceassumption;Iwillsoonargue against it. Thereareanumberofpuzzlesthatarisefromourtalkaboutthingsatdifferent levelsofbeing.Tokeepthediscussionmanageable,let’sfocusononeespecially difficultandtheoreticallyimportantpuzzle:thepuzzleofcausalexclusion.I’ll
²Thereisadisputeastowhether,whenpropertyPisrealizedbypropertyQ(whereQcouldalsobe apluralityofproperties),PandQbelongtothesameobjectXorQbelongstoX’sparts.Iwilldiscuss thisinChapter2,whereIarguethat,inthesenseofrealizationthatmattersmosttothemetaphysicsof science,QbelongstoX’sparts.Inthischapter,I’mjust defining “realization” astherelationthatholds betweenthepropertiesofawholeandthepropertiesofitsparts,orbetweenitspartstakencollectively andalloftheparts’ partstakencollectively.
³Asymmetry:xRy → notyRx.Irreflexivity:notxRx.Transitivity:(xRyandyRz) → xRz.Onthe featuresofrealization,cf.Baysan2015.
proposeaframeworkforthinkingaboutlevelsofbeingthatsolvesthecausal exclusionpuzzleandexplainshowdifferentlevels fittogether.Beware:aproper treatmentofthesetopicswouldrequireabookofitsown;inthischapterIonly haveroomforabriefsketchthatwillhelpusthroughtherestof this book.
1.2CausalExclusion SupposethatbrainBiscomposedofabunchofneurons,glia,othercells,and extracellular fluid,whichI’llcollectivelylabelNN.Forsimplicity,Iwillreferto NNsimplyasneurons.Differentpeopledisagreeabouttherelationshipbetween wholesandtheirparts e.g.,betweenBandNN.Somethinktheyareidentical, othersthattheyaredistinct,yetothersthatoneofthetwoisgroundedinthe other.We’llsuspendjudgmentfornow.
ConsideravoluntaryactionEperformedbyanorganism,andconsidertwo causalclaimsthatareentailedbywhatisfoundinanyneurosciencetextbook:
(1)BcausesE.
(2)NNcauseE.
Whenwemakethiskindofcausalclaim,weusuallyassumethatasystem causesaneffectinvirtueofbeinginaparticularstateorhavingaparticular property.Iassumethat beinginastate isthesameas possessingaproperty,sowe cantalkaboutstatesandpropertiesinterchangeably.
Totaketheroleofpropertiesintoaccount,let’sreformulate(1)and(2)interms ofbrainstateSB andneuralstateSNN,suchthatSNN realizesSB:
(3)SB causesE.
(4)SNN causesE.
Again,differentpeopledisagreeabouttherelationshipbetweenpropertiesof wholesandpropertiesoftheirparts e.g.,betweenSB andSNN.Somethinkthey areidentical,othersthattheyaredistinct,yetothersthatoneofthetwois groundedintheother.Again,we’llsuspendjudgmentfornow.
Whenwemakethiskindofcausalclaim,wegenerallyassumethateachcauseis sufficientfortheeffectundertherelevantbackgroundconditions.Thatistosay,if SB causesE,then givenrelevantbackgroundconditions theoccurrenceofSB is enough,allbyitself,tobringaboutE.Bythesametoken,ifSNN causesE,then givenrelevantbackgroundconditions theoccurrenceofSNN isenough,allby itself,tobringaboutE.
ThepuzzleofcausalexclusionarisesbecauseifSNN isalreadyenoughtocause E,thereseemstobenothingleftforSB todo.Supposethatweestablish(4) thata
neuralstatecausesavoluntaryaction.Whywouldwesaythatabrainstatealso causesthataction,ifwealreadyestablishedthataneuralstateisenoughtocause thatverysameaction?Theproblemisentirelysymmetrical.Supposeweestablish (3) thatacertainbrainstatecausesacertainaction.Whyaddthataneuralstate alsocausesthataction,ifwealreadyestablishedthatabrainstateisenoughtocause thatverysameaction?Ifoneofthetwoisenoughtoproducetheaction,itseems thattheotheriseitheridenticaltoitorshouldbedispensedwith.Butwhichisit? Theproblemgeneralizes.Ipickedneuronsandbrainsamongmanyotherlevels thatIcouldhavepicked.Thesamepuzzlecanberunwithneuralsystems,neural networks,molecules,atoms,andsoon nottomentionwholeorganisms.Thisis thepuzzleofcausalexclusion:ifanylevelisenough,allbyitself,tocauseaneffect suchasE,thenotherlevelsseemeitherdispensableoridenticaltotheonecausally efficaciouslevel(Kim1998,2005;Merricks2001).Yetbothlaypeopleandscientistsattributecausalefficacytomultiplelevelsofbeing,andwedon’tseemtotake alllevelstobeidenticaltooneanother.Whatgives?
1.3Overdetermination Oneanswertothecausalexclusionpuzzleisthatdistinctlevelscausethesame effectatthesametime.Thistheoreticaloptioniscalledoverdetermination.To understandoverdetermination,considertheclassicexampleofa firingsquad, eachofwhosemembers firesadeadlybullet.Giventheoperativebackground conditions,eachbulletissuf ficient,allbyitself,tokillaperson.Inthiscase,the presenceofasuf ficientcauseiscompatiblewithmanyothersuf ficientcausesfor thesameeffect.Byanalogy,levelsofbeingmightbelikesoldiersina firingsquad: eachlevelproducesthesameeffectindependentlyoftheothers(Loewer2007; Pereboom2002;Schaffer2003;andSider2003).Mightthisbeasolutiontothe causalexclusionpuzzle?
Theredoesseemtobeasimilaritybetweenthetwosituations.Inbothcases, multiplethingsoccurringatthesametimearesaidtobecausallysuf ficientforthe sameeffect.Yettherearealsoprofounddissimilarities.
Incasesofordinarycausaloverdetermination,suchasthe firingsquad,the differentcausescanbeindependentlyobservedandtracked.Forexample,wecan checkthedifferentriflesbeforethesquadshootsandseeiftheycontainlive ammunition.Aftertheshooting,wecansearchfordistinctbulletsandseewhere theyendedup.Thissortofempiricalinvestigationisnotanoptionwhendifferent levelsofbeingareinplay.Wecan’tinvestigatetheeffectsofabrainindependently ofthoseoftheneuronsthatmakeupthatverybrain.Wecan’tobserveor manipulatethestateofthebrain,separatelyobserveormanipulatethestateof theneuronsthatmakeupthatsamebrain,andseeiftheyhavedistinctcausal powersanalogoustothepowersofdistinctbullets.Therefore,wecannotcollect
anyevidencethattherearedistinctcausalpowersatdistinctlevelsthatare sufficientforthesameeffect.Withoutevidence,itisbothontologicallyprofligate andepistemicallyirresponsibletopositthem.
Adefenderofoverdeterminationmayreplythatthedisanalogybetweenmultileveloverdeterminationandordinary firing-squad-type overdeterminationis immaterial.Distinctlevelsofbeing(wholesandtheirparts,higher-levelproperties andtheirrealizers)aremetaphysicallyconnectedtooneanotherinawaythat distinctbulletsarenot.Giventhetightconnectionbetweenlevels,adefenderof overdeterminationmightinsistthatdistinctlevelsmaywellbecausallysufficient forthesameeffect atleastinaloose-enoughsenseof “ cause. ”
Thisreplymisunderstandsthestakesoftheproblem.Thequestionisnot whetherasuitablywatered-downnotionofcausationcanunderwriteclaimsto theeffectthatdistinctlevelscausethesameeffect.Thequestioniswhetheritis legitimatetopositdistinctcausalpowersthatarecausallysufficientforthesame effectinsituationswhereitisimpossible,asamatterofprinciple,tocollect empiricalevidenceoftheirdistinctexistence.
Evenifwesetasideepistemicconsiderations,itmakesnosensetopositthat thingsthatstandinarelationofcompositionorrealizationhavedistinctcausal powersforthesameeffect.Theobjectsthemselves awholeanditsparts arenot even(wholly)distinct.Thepartsmakeupthewhole;thewholeconsistsofthe parts.Bythesametoken,higher-levelpropertiesandtheirrealizersaretoo intimatelyconnectedtohavedistinctcausalpowersforthesameeffect.Once youhavetherealizer,youalsohavetherealizedproperty;andonceyouhavethe realizedproperty(token),youalsohaveitsrealizer.Howcouldthingsthatareas intimatelyrelatedasawholeanditsparts,orahigher-levelpropertyandits realizer,cometohavedistinctcausalpowers,eachsufficientforthesameeffect (cf.Bernstein2016)?LaterIwillelucidatethemetaphysicalconnectionsbetween wholesandtheirpartsaswellashigher-levelpropertiesandtheirrealizers,arguing thatsuchconnectionsareinconsistentwithoverdetermination.
Insummary,theoverdeterminationsolutiontothepuzzleofcausalexclusion positsmultipledistinctcausesforthesameeffectthatarepossessedbyawhole anditsparts,orbyahigher-levelpropertyanditsrealizers.Thesearemorecauses thanarenecessarytoexplainphenomena,withoutanyevidencethattheyarethere oranywaytoinvestigatethemseparately.Becauseofthis,overdeterminationisa desperateoption.Perhapsweshouldconsideritasalastresortifallotheroptions fail.Let’spondermoreappealingalternatives.
1.4Reduction Thesimplestsolutiontothepuzzleofcausalexclusionwouldbetoidentifythe one genuine causeofaneffectandreducealltheotherputativecausesto that one.
Forexample,supposeweestablishedthatneuralstatesarethegenuinecauseof voluntaryactions.Ifwecouldreducebrainstocollectionsofneurons,andbrain statestoneuralstates,thenwewouldconcludethatthereisnorealconflict between(1)and(2),orbetween(3)and(4),because(1)reducesto(2),and(3) reducesto(4).ThereisonlyonegenuinecauseofE neuralstates.
Ifwearegoingtoconsiderthissortof reductionism,weneedtosaywhatwe meanbyreduction.Thisisavexedquestioninbothmetaphysicsandthephilosophyofscience(vanRielandvanGulick2019).Forpresentpurposes,Iwill adoptthesimplestandclearestnotionof ontological reduction:identitywitha direction.Thatistosay,XreducestoYifandonlyifXisstrictlyidenticaltoYand YismorefundamentalthanX.Speci fically,BreducestoNNifandonlyifB=NN andNNismorefundamentalthanB,andSB reducestoSNN ifandonlyifSB =SNN andSNN ismorefundamentalthanSB.Thedifficultieswithreductionariseforany reasonablenotionofreduction;wemightaswellkeepthingsmanageableby focusingonreductionasidentityplusdirection.⁴
Itshouldgowithoutsayingthatreductionintherelevantsenseisnotjusta matterofwhatispartofwhat.Everyoneagreesthatneuronsarepartsofbrains, moleculesarepartsofneurons(aswellasbrains),andeveryphysicalobjectispart ofthewholeuniverse.Yetreasonablepeopledisagreeaboutwhetherlargethings reducetosmallthings,smallthingsreducetotheuniverse(e.g.,Schaffer2010),or neitherreducestotheother.
Assumereductionismholds.The firstquestionis, which directiondoesreductiongo?Dobrainsreducetoneuronsorviceversa?Doneuralstatesreduceto brainstatesorviceversa?We’vealsoseenthattherearen’tjusttwolevels there aremany.Weneedtosaywhereneuralsystemsandtheirproperties,peopleand theirproperties,moleculesandtheirproperties,andsoforth fitwithinthe reductionistpicture.Foranyoneoftheselevels,eitheritisthereductionbase foralltheothers,oritreducestowhicheverlevelallothersreduceto.For reductiontowork,weneedto findtheone fundamental leveltowhichallother levelsreduce.Idoubtwecan.
Inrecentdecades,perhapsthemostpopularsolutionistopickthelevelof elementaryfermionsandbosonsasthefundamentalontologicallevel,towhich everythingelsesupposedlyreduces.Thisdoeshavesomeappeal atleasttothose, likeme,withatasteforparticlephysics.Physiciststhemselvescallelementary fermionsandbosonsthe fundamentalparticles.Maybetheparticlesthatare physically fundamentalarealso ontologically fundamental?⁵
⁴ AdifferentproposalbyHemmoandShenker(2015;Shenkerunpublished)isthatXreducestoYif andonlyifXisan aspect ofY.Their flatphysicalism,whichwasdevelopedindependentlyofthepresent work,goesinthedirectionoftheegalitarianontologyIdefendhereandismorecompatiblewithitthan traditionalreductionism.
⁵ Thepreciseontologicalstatusoffermionsandbosons,andofquantummechanicalsystemsin general,iscontroversial.Debatesincludewhetherfermionsandbosonsareindividualobjects(French
Beforeansweringthatquestion,letmepointoutthatphysicalfundamentality does not entailontologicalfundamentality.Somethingisphysicallyfundamental, roughly,justincaseithasnophysicalparts nopartsthataredistinguishablevia physicaloperationsortheories.Beingphysicallyfundamentalisawesomeand important:theoriesthataretrueofphysicallyfundamentalparticlesaretrueof everythingphysical.Nevertheless,physicalfundamentalityisnotthesameas ontologicalfundamentality.
Somethingis ontologically fundamentaljustincaseeverythingelsesynchronicallyandasymmetrically depends onit.Theontologicallyfundamentalstuff grounds alltheotherstuff.Theontologicallyfundamentalstuffisontologically prior totheotherstuff.Thesearethemetaphorsusedbymetaphysicians;theyare noteasytopindown.
Ontologicaldependenceiseasytograspwhentime flowsandnewobjectsand propertiescomeintoexistence.Forinstance,ifagroupofpeopleformsanewclub, theclubontologicallydependsonthetemporallypriorexistenceofthegroupof people.Thereasonisobvious:withoutpeople,noclubcouldbeformed.Thesame ideaisalsoeasytograspwhenwedealwitharbitraryfusionsofobjectsor derivativeproperties.Forexample,theobjectformedbycombiningtheStatue ofLibertyandtheTajMahaldependsontologicallyontheexistenceofthe StatueofLibertyandtheTajMahal;thepropertyofbeingamileawayfrom theStatueofLibertydependsontologicallyontheexistenceandlocationofthe StatueofLiberty.
Itislessclearwhatbeingontologicallymorefundamentalmeanswhenwe consider,atoneandthesametime,awholeversusitsparts,orawhole’sproperties versusitsparts’ properties,whenthewholeanditspropertiesareindividuated independentlyofthepartsandtheirproperties.
Inspiteofthislackofclarityaboutwhatitmeanstobeontologicallyfundamental,lotsofphilosopherstalkaboutonelevelbeingmorefundamentalthan another,oronelevelbeingabsolutelyfundamental.Ultimatelythelackofclarity aboutontologicalfundamentalitywillnotmattertousbecauseIwillarguethat, givenanyrelevantnotionoffundamentality,nolevelofbeingismoreontologicallyfundamentalthananyother.Fornow,though,let’spretendwehaveatleasta faintgraspofwhatontologicalfundamentalitymeansandproceedtoexaminethe prospectsofthisidea.Couldtheontologicallyfundamentallevelbethephysically fundamentallevel?
Elementaryfermionsandbosonsarethebuildingblocksoftheuniverse:they havenophysicalpartsandtheyconstituteeveryphysicalobject.Inaddition, 2015)andwhetherquantummechanicalsystemsadmitofuniquedecompositions(Healey2013). Addressingthesedebatesfallsoutsidethescopeofthischapter.Forpresentpurposes,Iassumethat fermionsandbosonscanbeindividuatedbyacombinationofinvariantsandobservables(Castellani 1998).
accordingtotheStandardModelofparticlephysics,thepropertiesandinteractionsoffermionsandbosonsconstitutivelyexplainallphysicalphenomena. Thatincludes,presumably,allhigher-levelphenomenasuchasbrainstatescausingvoluntaryaction.TheStandardModelisbyfarthemostsuccessfulandwellconfirmedscientifictheoryeverdevised.That’simpressive.Thus, metaphysical atomism reducinghigherlevelsofbeingtofundamentalphysicalparticlesand theirproperties seemslikeagoodreductionistoption.Butagain,ontological fundamentalitydoesnot follow fromphysicalfundamentality.
Anotherpopularreductionistoptionistopickthewholeuniverseasthe fundamentalontologicallevel,towhicheverythingelsereduces(Schaffer2010). Ifwelookatthehistoryofphilosophy,this monistic solutionmaybemorepopular thanatomism.Oneadvantageofmonismisthat,bydefinition,theuniverse includeseverythingwithinitself allobjectsandproperties.Ifwetaketheuniversetobeontologicallyfundamentalandreduceeverythingelseto it,surelywe arenotleavinganythingout.
Anothermetaphysicaladvantageofmonismisthatitislesssubjecttothe vagariesofphysicaltheorythanmetaphysicalatomism.Somephysicistshopeto replacetheStandardModelwithsomethingevenmorephysicallyfundamental, whichwillunifyallphysicalforces.Toperformthisunification,theoretical physicistsexplorethepossibilitythatthereareentitiesevenmorephysically fundamentalthanfermionsandbosons.Iftherearesuchfundamentalphysical entities,thenperhapseverythingreducesto them.Butwhatiftherearenotrue atoms noindivisiblephysicalentities?Whatifeveryphysicalentityhassmaller andsmallerpartsallthewaydown?Thisseemstomakereductionimpossible,for lackofareductionbase.Monism theviewthateverythingreducestothewhole universe doesnotseemtofacethissortofrisk(cf.Schaffer2010).
Still,monismshareswithatomismtheproblemthatpickinganylevelasa reductionbaseseemsarbitrary.Asidefromthepuzzleofcausalexclusion,it’sunclear whyanylevelshouldbemoreontologicallyfundamentalthananyother.Andthe puzzleofcausalexclusionisentirelyneutralonwhichlevelismorefundamental.It couldbethewholeuniverse,theleveloffermionsandbosons,thelevelofbrainsand othermid-sizeobjects,thelevelofmolecules(cf.Bickle2003),oranylevelin between.Inadditiontothisarbitrariness, reductionismseemstobestuckwithlack oftestability:evenifthereisafundamentallevel,it’sunclearhowto findit.
Anotherdrawbackisthatreductionismgivestheimpressionthatsomesciences aremoreimportantthanothers.Specifically,whicheverscienceinvestigatesthe fundamentallevel thegenuinecauses seemstohaveanedgeovertheother sciences.Afterall,othersciencesarejuststudyingstuffwhosecausalpowers dependonreducingittothefundamentalstuff.Humannaturebeingwhatitis, manypeopledon’tlikethattheirfavoritesubjectmatter,towhichtheydevote theirworkinghours,isjustderivativeonsomethingmorefundamental,whichis studiedbysomeoneelse.Plus,ifeverythingreducestothefundamentallevel,so
