Original Paper UDC 165.62: 165.12 Received October 16th, 2007 Bruce J. MacLennan Department of Electrical Engineering & Computer Science, University of Tennessee, 1123 Volunteer Blvd. USA-Knoxville, TN 37996–3450 [email protected]Consciousness: Natural and Artificial Abstract Based on results from evolutionary psychology, we discuss important functions that can be served by consciousness in autonomous robots. These include deliberately controlled action, conscious awareness, self-awareness, metacognition, and ego consciousness. We distinguish intrinsic intentionality from consciousness, but argue it is also important to un- derstanding robot cognition. Finally, we explore the Hard Problem for robots (i.e., whether they can experience subjective awareness) from the perspective of the theory of protophe- nomena. Keywords autonomous robot, awareness, consciousness, evolutionary psychology, the Hard Problem, intentionality, metacognition, protophenomena, qualia, synthetic ethology 1. Introduction There are many scientific and philosophical problems concerning conscious- ness, but in 1995 David Chalmers proposed using “the Hard Problem” to refer to the principal scientific problem of consciousness, which is to understand how physical processes in the brain relate to subjective experience, to the feeling of being someone. As he put it, “It is widely agreed that experience arises from a physical basis, but we have no good explanation of why and how it so arises” (1995). Unfortunately, the scientific investigation of experience is impeded by the unique epistemological status of consciousness, for while scientific observation is based on specific experiences, consciousness is the ground of all possible experience (MacLennan, 1995). Chalmers called on re- searchers to face up to the Hard Problem, and Shear (1997) collects a number of papers responding to his challenge. Of course, neither Chalmers nor I intend to suggest that all the other prob- lems connected with consciousness are “easy”; indeed, some of them are as difficult as any in neuropsychology. However, they may be approached us- ing ordinary scientific methodology, as developed in cognitive science and neuroscience, and so in this sense they are “less hard” than the Hard Problem. They have in common that, at least in principle, they can be solved in terms of neural information processing and control, without reference to any associ- ated subjective experience. In this article I will begin by considering some of these “less hard” problems, especially in the context of robot consciousness, which provides a useful point of contrast and comparison to these problems in the context of humans and other animals. Then I turn to the Hard Problem in the contexts of both natural and artificial intelligence.
33
Embed
Consciousness: Natural and Artificialbmaclenn/papers/CNA-published.pdf · Consciousness: Natural and Artificial Abstract Based on results from evolutionary psychology, we discuss
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
AbstractBased on results from evolutionary psychology, we discuss important functions that can be served by consciousness in autonomous robots. These include deliberately controlled action, conscious awareness, self-awareness, metacognition, and ego consciousness. We distinguish intrinsic intentionality from consciousness, but argue it is also important to un-derstanding robot cognition. Finally, we explore the Hard Problem for robots (i.e., whether they can experience subjective awareness) from the perspective of the theory of protophe-nomena.
Therearemanyscientificandphilosophicalproblemsconcerningconscious-ness,butin1995DavidChalmersproposedusing“theHardProblem”torefertotheprincipalscientificproblemofconsciousness,whichistounderstandhow physical processes in the brain relate to subjective experience, to thefeelingofbeingsomeone.Asheputit,“Itiswidelyagreedthatexperiencearisesfromaphysicalbasis,butwehavenogoodexplanationofwhyandhowitsoarises”(1995).Unfortunately,thescientificinvestigationofexperienceisimpededbytheuniqueepistemologicalstatusofconsciousness,forwhilescientificobservationisbasedonspecificexperiences,consciousnessisthegroundofallpossibleexperience(MacLennan,1995).Chalmerscalledonre-searcherstofaceuptotheHardProblem,andShear(1997)collectsanumberofpapersrespondingtohischallenge.Ofcourse,neitherChalmersnorI intendtosuggestthatall theotherprob-lemsconnectedwithconsciousnessare“easy”;indeed,someofthemareasdifficultasanyinneuropsychology.However,theymaybeapproachedus-ingordinaryscientificmethodology,asdeveloped incognitivescienceandneuroscience,andsointhissensetheyare“lesshard”thantheHardProblem.Theyhaveincommonthat,atleastinprinciple,theycanbesolvedintermsofneuralinformationprocessingandcontrol,withoutreferencetoanyassoci-atedsubjectiveexperience.InthisarticleIwillbeginbyconsideringsomeofthese“lesshard”problems,especiallyinthecontextofrobotconsciousness,whichprovidesausefulpointofcontrastandcomparisontotheseproblemsinthecontextofhumansandotheranimals.ThenIturntotheHardProbleminthecontextsofbothnaturalandartificialintelligence.
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial402
2. Less Hard Problems
Functional Consciousness
Oneof thedifficulties in the scientific studyofconsciousness is that evenpsychologistsandphilosophersusethetermwithavarietyofinterrelatedandoverlappingmeanings(e.g.,Block,1995).InthissectionIwillconsiderseve-ralofthesenotionsandthe“lesshard”problemsassociatedwiththeminthecontextofanimalsandrobots.Whatisconsciousnessgoodfor?Whyhasitevolved?Isthereanyreasonweshouldwantrobotstobeconscious?Toanswerthesequestions,weneedtounderstandthefunction,thatis,thepurpose fulfilled,bybiologicalconscious-ness.Inbiology,questionsofthefunctionofanorganorprocessareansweredbyinvestigatingitsadaptivevalue,thatis,byaskingwhatselectiveadvan-tageithasconferredinthespecies’environment of evolutionary adaptedness(EEA),whichistheenvironmentinwhichthespeciesevolvedandtowhichitisadapted.Tothisend,comparativestudiesbetweenspeciesareoftenin-formative.Evolutionary psychologyreferstotheapplicationofevolutionarybiologytopsychologicalquestions,andIwilluseasimilarapproachtoad-dressthe“lesshard”problemsofrobotconsciousness.1
One of the functions of consciousness is to control what is referred to asvoluntary action,but toavoid irrelevant issuesof“freewill”, it isperhapslessconfusingtocallitdeliberately controlled action.Muchofoureverydayactivityisautomatically controlled,thatis,thedetailedsensorimotorcontrolisunconscious.Examples includewalking,feedingandwashingourselves,anddrivingacarunderordinaryconditions.Undersomeconditions,however,ourcontrolofouractionsbecomesveryconsciousanddeliberate.Thismayberequiredwhenconditionsareabnormal(e.g.,walkingwhenyouaredizzyorcrossingice,drivinginbadweatheror traffic),orwhenwearelearninganewskill (which, therefore, isnotyetautomatic).Forexample,anunex-pectedsensationduringautomaticbehaviorcantriggeranorientingresponseandabreakdownintheautomatizedbehaviorsothatitmaybeplacedundermoredeliberate(“voluntary”)control.Forexample,whenwalkingaleggetscaughtorstuck,orthegroundisinfirm.Thismaytriggerdeliberateactivitytofreethelegortoinspectthelocalenvironment.Underbreakdownconditionswepaymuchmore attention, investing scarce cognitive resources in care-fulcoordinationofsensoryinputandmotorbehavior;wecannotdependonlearnedautomaticbehaviors,withtheirlimitedassessmentsofrelevanceandprogrammaticcontrolofresponse,todotherightthing.In the terms of Heidegger’s phenomenology (e.g., Heidegger, 1962, 1982;Dreyfus,1991,ch.4),inourordinarymodeofskilfulcoping,weencounterobjectsand theworldasready-to-hand (zuhanden), ineffectasequipmentthroughwhichour“projects”arebeingrealized.However,when there isabreak(Bruch),thatis,abreakdownordisturbanceinthistransparentbeha-vior,thentheabsentorobstructingobjectorconditionbecomesunready-to-hand(unzuhanden)andasaconsequenceconspicuous(auffällig).Thereisashift fromabsorbed coping todeliberate coping.Theobstructingobjectorconditionisthenencounteredaspresent-at-hand(vorhanden),ineffect,asaproblemtobesolvedsotheremaybeareturntotheproject.Itistreatedasathingorobjectivesituationratherthanasequipmentorameansappropriatetoanend.Similarconsiderationsapplytoautonomousrobotswhentheyareoperatingunderexceptionalcircumstancesorlearningnewskills,andsotheyshouldbe
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial403
abletoexertdeliberatecontroloveractivitiesthatareotherwiseautomatic,orthatmaybesooncelearned.Deliberatecontrolinvolvestheintegrationofawiderrangeofinformationthanautomaticcontrol(forthelatterfocusesoninformationwhoserelevancehasbeenestablished)andinvolves theuseoffeedbackfromawidervarietyofsourcestocontrolaction.Informationrepre-sentationislessspecific,moregeneral-purpose(andthereforemoreexpensiveintermsofneuralprocessingresources).Automaticactionmakesuseofmorenarrowlyfocusedinformationrepresentationsandprocessingpathways.One of the ways that consciousness can facilitate deliberately controlledaction is by a process termed conscious awareness, that is, by integratinginformation frommemoryandvarious sensorymodalities (e.g., visual andkinesthetic),andbyusingitformoredetailed,explicitmotorcontrol,whichisanotherfunctionofconsciousness.Normallywewantautomaticallycon-trolledactivitiestotakeplaceinmoreperipheralprocessingsystemsinvolv-ingonlytheinformationresourcesrequiredfortheirskillfulexecution,thusleavingthecentralizedresourcesofconsciousawarenessavailableforhigherlevelprocesses.Humanbeings,andprobablymanyotherspecies,exhibitvisual dominance,that is, informationintegrationisaccomplishedbyrelatingit tovisualrep-resentations.Thus,sounds,odors,tactileperceptions,etc.areboundtopartsofvisualperceptionsandlocalizedwithrespecttovisuallyperceivedspace.Memorymaytriggerthesebindings(e.g.,theappearanceofahostileagenttoitssound)onthebasisofstoredassociations.Thefundamentalreasonforvisualdominance(asopposedtosomeothersensorymodality)canbefoundin theshortnessofopticalwavelengths,whichpermitsdetailed imagingofremoteobjects.Thesameconsiderationsapplytorobots,whichsuggeststhatvisualdominancemaybeagoodbasisforinformationintegrationinartificialconsciousawareness(inthesensedefinedabove).Another functionofconsciousness isself-awareness,which in thiscontextdoesnotrefertotheabilitytocontemplatetheexistentialdilemmasofone’sbeing,butrathertotheawarenessofoneselfasaphysicalobjectintheenvi-ronment.Loweranimals,andespeciallyanimalsthatinteractwiththeirenvi-ronmentsinarelativelylocalizedway(e.g.,tactile,gustatory,andolfactoryinteractions)canoperatefromaprimarilysubjectiveperspective;thatis,theworldisunderstoodfromaperceiver-centeredperspective(theworldisex-periencedascenteredaroundtheanimal,andtheanimal’sactionsareexpe-riencedasreorientingandreorganizingthesurroundingenvironment).Morecomplex animals, especially those that engage in high-speed, complicatedspatialmaneuvers(e.g.,arborealmonkeys:Povinelli&Cant,1995),needtohave representationsof theirbodies’positions,orientations,andconfigura-tionsinspace.Thatis,theyrequireamoreobjectiveperspectiveontheworld,in which they understand their own bodies as objects in an independentlyexistingworld.Theiractionsdonotsomuchaffectasurroundingsubjectiveuniverseasaffecttheirbodyinanobjectiveenvironmentsharedbyotherin-dependentandindependentlyactingobjects.Similarconsiderationsapplytoanimalsthatcoordinatehigh-speed,spatiallydistributedgroupactivitiesinasharedenvironment(e.g.,huntingpacks).Ofcourse,evenfortheseanimals,althoughtheplannedandexperiencedul-timateeffectsofactionareunderstoodinreferencetoanobjectiveenviron-
1
Introductions to evolutionary psychologycanbefoundinmanyrecenttextbooks,such
as Buss (2004) and Gaulin & McBurney(2004).
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial404
ment,thesubject-centeredperspectiveisnotirrelevant(sincetheimmediateeffectofmostactionsistocausesomebodilychange).Therefore,higherani-malsneed tocoordinateseveral referenceframes, includingat leastworld-centered, local-environment-centered, body-centered, and head-centeredframes.This isacomplicatedconstraint satisfactionproblem,whichundernormalconditionsisseamlesslyandunconsciouslysolvedbyneuralinforma-tionprocessing.Autonomousrobotsthatareintendedtooperateundersimilarconditions (high-speedmotion, spatiallydistributedcoordination) similarlyrequirethiskindofself-awarenessinordertocontrol theirmotionthroughashared,objectiveenvironment.Thereforetheytooneedtorepresent theirpositions,orientations,andconfigurationswithrespecttomultiplereferenceframes,andtobeablerapidlymaintainthemutualconsistencyoftheserep-resentations.Anotherfunctionofconsciousness,inhumansatleast,ismetacognition,thatis,awarenessandknowledgeconcerningthefunctioningofone’sownner-voussystem.Forexample,youmaybeawarethatyouarelesscoordinatedwhenyouare tired, thatyouhaveabadmemoryfor faces,or thatyouactrashlywhenangry.Thisis,ofcourse,anotherformofself-objectification,andmaybejustasvaluableinsomeautonomousrobotsasitisinhumans.Anadditionallevelofself-objectificationfacilitatesreasoningaboutthecon-sequencesofone’sactions.Theeffectistostepback,viewoneselfasthoughanother person, and come to an understanding about how one’s own psy-chological processes lead to outcomes that are either desirable or undesir-able(eitherfromone’sownorawiderperspective),usingthesamecognitiveprocessesthatareusedforunderstandingotherpeople’spsychologicalstatesandbehavior (e.g.,neuronal“mirrorcells”:Rizzolatti&Craighero,2004).Forexample,youmayrecognizethatundesirableconsequencesfollowfromhittingpeoplewhenyouareangrywiththem.Inthiswayweacquirealevelofexecutivecontroloverourpsychologicalprocesses(animportantfunctionofego-consciousness,according topsychologists,e.g.,Stevens,2003).Forexamplewecanlearn(externalorinternal)stimulithatshouldtriggermoredeliberate(“voluntary”)controlofbehavior.Similar considerations apply to autonomous robots that implement higher-levellearningandcontrolofbehavior.Sucharobotmayneedtocontroltheoperationof its lower-level behavioral programson thebasis of reasoningabouttheconsequencesofitsownactions(viewedobjectively)initsenviron-ment.2 Such control may be implemented through discursive reasoning aswellasthroughanalogsimulation(e.g.,viamirrorcells).Ishouldremarkthattheaccountofconsciousnesspresentedhereisconsistentwiththatofmanypsychologists(e.g.,Stevens,2003),whoobservethatcon-sciousnessisnotthecentralfacultyofthepsychearoundwhichalltheothersorbit(apointalsostressedbyJung,1969b,§391).Rather,consciousnessisaspecializedmodulethatisdedicatedtohandlingsituationsthatgobeyondthecapabilitiesofothercognitivemodules(sensorimotormodules,automatedbehavioralprograms,etc.).Weexpectconsciousrobots,likeanimals,toper-formmanyoftheiroperationswithminimalengagementoftheirconsciousfaculties.Consciousnessisexpensiveandmustbedeployedselectivelywhereitisneeded.Insummary,wehaveseenfromthisreviewofthefunctionsofconsciousnessinhumansandotheranimalsthatmanyofthesefunctionsmayalsobeusefulinautonomousrobots.Fortunately,applyingtheseideasinroboticsdoesnotraiseanygreat,unsolvedphilosophicalproblems.Thatdoesnotmean that
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial405
theyaresolved,oreasytosolve;onlythatthe“lesshard”–butstilldifficult!–methodsofneuroscienceandneuroethologycanbeappliedtothem.Aswegradually come to understand theneuronalmechanisms implementing thisfunctional consciousness(oraccess consciousness:Block,1995),wemaybe-gintoapplytheminroboticdesignsothatourrobotscanbenefitfromthemaswell(andthusexhibitfunctionalconsciousnessaswell).
Intentionality
Intentionalityisanissuecloselyrelatedtoconsciousness,butnotidenticaltoit,soitwillbeworthwhiletodiscussbrieflyintentionalityinartificialagents,suchasrobots.Intentionalitymaybedefinedasthepropertybywhichsomething(suchasalinguisticexpression)isaboutsomethingelse.3Therefore,itisthroughitsintentionality that something ismeaningful andhascontent.Whenappliedto consciousness, intentionality is the property through which conscious-nesshascontent,forconsciousnessisalwaysconsciousnessofsomething,asBrentano(1925,p.89)stressed.4Ofcourse,mostofthedatainacomputer’smemoryisaboutsomething–forexample,anemployee’spersonnelrecordisaboutthatemployee–butwewouldnotsaythatthedataismeaningfultothecomputerorthatthecomputerunderstandsit.Theintentionalityofthedatainthecomputerisdependentuponitsmeaningfulnesstous.Therefore,phi-losophershavedistinguishedthederived intentionality(ofordinarycomputerdata,books, etc.) from the intrinsic (ororiginal) intentionalityofourcon-sciousstates,memories,communicationacts,etc.(Dennett,1987,pp.288–9,Haugeland,1997,pp.7–8).Robotsstoreandprocessmanykindsofdata.Muchofitwillhaveonlyderivedintentionality, because the robots are collecting and processing the data toservetheneedsofthedesignersorusersoftherobots.However,inthecontextofrobotconsciousness,wearemoreconcernedwithintrinsicintentionality,withtheconditionsunderwhicharobot’sinternalstatesandrepresentationsaremeaningfultotherobotitself(and,hence,wecouldsaythattherobotun-derstands).Eachofuscandeterminebyintrospectionifweareunderstandingsomething(whichisthebasisoftheChineseRoomArgument),butthiswillnothelpustodetermineifarobotisunderstanding,sowemustuseadifferentstrategytoanswerquestionsaboutintrinsicintentionalityinrobots.Theinvestigationofintrinsicintentionalityinnon-humanagentsisacompli-catedproblem,whichcannotbeaddressedindetailhere.5Fortunately,etholo-gistshavehadtodealwiththisprobleminthecontextofanimalcommunica-tionandrelatedphenomena,andsowemaylearnfromthem.Forexample,animalsmayactinmanywaysthatinfluencethebehaviorofotheranimals,butwhichoftheseactionsshouldbeconsideredcommunication?Oneanimal,forinstance,maysharpenitsclawsonatree,andanotheranimal,whenitsees
See, forexample,Blackburn(1994,p.196),Gregory(1987,p.383),Gutenplan(1994,p.379),andSearle(1983,p.1).
4
The philosophical concept of intentionality,in the sense of “aboutness” or meaningful-ness, shouldbecarefullydistinguishedfromtheordinaryideaof“intention”aspurposeorgoal.
B. J. MacLennan, Consciousness: NaturalandArtificial406
themarks,maygoinadifferentdirection.Wasthiscommunication,orwasitanon-communicativeeventinwhichthebehaviorofoneanimalindirectlyinfluencedthatofanother?Wewouldliketobeable todeterminewhetherthepurposeofthefirstanimal’sactionwastoinfluencethebehaviorofotheranimals(e.g.,Grice,1957),orwhetherthatwasmerelyanaccidentalconse-quenceofitsaction(butnotitspurpose).Aswehaveseen,thebestwaytounderstandpurposeinabiologicalcontextistolooktoabehavioraladaptation’sselectiveadvantage,orlackthereof,inaspecies’environmentofevolutionaryadaptedness(EEA).Inthisway,com-municationcanbedefinedasanactionthat, in theEEA,has thestatisticallikelihoodofinfluencingthebehaviorofotheranimalsinsuchawayastoincreasetheinclusivefitnessofthecommunicator(thatis,theselectivead-vantageofthecommunicatororitsgroup)(Burghardt,1970).Inasimilarwaywecanapproachtheintrinsicintentionalityofothermeaning-bearingstatesorrepresentationsinanyagent(animal,robot,etc.).Toafirstapproximationtheirmeaningisgroundedintheirrelevancetothesurvivalorwellbeingofanindividualagent,butitismoreaccuratetogroundmeaningintheagent’sinclusivefitness,whichtakesaccountofitsselectiveadvantagetotheagent’sgroup.Ofcourse,themeaningsofparticularstatesandrepresentationsmaybeonlylooselyanddistantlycorrelatedtoinclusivefitness,whichneverthe-lessprovidestheultimatefoundationofmeaning.Perceptual-behavioral structures and their associated representations thathaveasignificantgeneticcomponentneedtobeinterpretedinreferencetotheEEA.Behaviorsandrepresentationsthathavenoselectiveadvantageinananimal’scurrentenvironment (e.g.huntingbehavior inacaptiveordo-mesticatedanimal)mayhaveameaningthatcanbeunderstoodinthecontextoftheEEA.Thisdoesnotimplythatanagent’sinternalstatesandbehaviorhavenomeaninginotherenvironments,butonlythatthemeaningofinnateperceptual,behavioral,andcognitivestructuresshouldbeinterpretedinthecontextoftheEEA(foritisthatenvironmentthatdefinestheirpurposesandhasgiventhemtheirprimarymeaning).Canartificialagents,suchasrobots,exhibitintrinsicintentionality?Synthetic ethology offers a methodology by which such questions can be addressed(MacLennan,1990,1992,2006;MacLennan&Burghardt,1993).Thegoalofsyntheticethologyistopermitthescientificinvestigationofproblemsre-lating to thephysicalprocessesunderlyingmentalphenomenabystudyingsyntheticagentsin“syntheticworlds”,whicharecompletebutverysimple,andsopermittheconductofcarefullycontrolledexperiments.Forexample,inoneseriesofexperimentsbeginningin1989weusedsynthetic-ethologytechniquestodemonstratetheevolutionofcommunicationinapopulationofsimplemachines(MacLennan,1990,1992).Weshowedthatifthemachinesareabletomodifyandsenseasharedenvironment,andifthereisselectivepressureoncooperativebehavior(whichcouldbefacilitatedbycommunica-tion,butcouldalsooccurwithoutit),thenthemachineswillevolvetheabilitytocommunicate.Thesignalsexchangedby thesemachinesaremeaningfulto them because, in their EEA, these signals are relevant to the machines’continuing“survival”(asorganizedstructures).Asobserverswecanmonitortheirbehaviorandinferthemeaningoftheircommunication,butinthiscaseourunderstandingisderived,whereastheirsisintrinsic.6
B. J. MacLennan, Consciousness: NaturalandArtificial407
bacteria,communicatewitheachotherandhaveinternalstatesthatrepresenttheirenvironment(e.g.,Dretske,1985,p.29;Dunny&Winans,1999);theirinternalstatesandsignalshaveintrinsicintentionality,althoughtheydonotexhibitconsciousnessinthesensethatIhaveusedithitherto.Withthisbackground,wecanaddressthequestionofintrinsicintentionalityinrobotsanditsrelationtoconsciousness.Certainly,trulyautonomousrobotsneedtobeconcernedwiththeirownsurvival:forexample,theyneedtobeabletofindenergysources(e.g.,sunlight,fuel),torepairthemselves(totheextentpossible),toextricatethemselvesfromdangeroussituations(e.g.,stuckinmudorsand), toavoidnaturalthreats(e.g.,weather,unsafeterrain,curi-ousorpredatoryanimals),andperhaps(formilitaryrobots)toevade,escape,orneutralizehostileagents.Functionssuchas these, relevant to the robot’scontinuedexistencequarobot,provideafoundationofintrinsicintentional-ity,whichgroundstherobot’scognitivestates,fortheyaremeaningfulto the robot.Suchfunctionscontributetoanindividualrobot’sfitness,butthereareothercircumstancesinwhichitwouldbeadvantageoustohavearobotsacrificeitsownadvantageforthesakeofotherrobots.Formanypurposesweneedcoop-erativegroupsofrobots,forwhichthecollectivefitnessofthegroupismoreimportantthanthesuccessofitsmembers.Indeed,thesesameconsiderationsapplytorobotsthatdefinetheirgrouptoinclude(certainorall)humanbeingsorothergroupsofanimals,forwhomtheymaysacrificetheirownadvantage.Inallofthese“altruistic”situations,groupfitnessprovidesanexpandedfoun-dationofintrinsicintentionality.Finally,forsomeapplicationsitwillbeusefultohaveself-reproducingro-bots;examplesincludeapplicationsinwhichrobotsmightbedestroyedandneedtohavetheirnumbersreplenished,andsituationsinwhichwewanttohavethenumberofrobotsadapttochangingconditions(e.g.,expandingorcontracting with the magnitude of the task). If the robots reproduce suffi-cientlyrapidly(whichmightbethecase,forexample,withgeneticallyengi-neeredmicroorganisms),thenwemustexpectmicroevolutiontotakeplace(fortheinheritancemechanismisunlikelytobeperfect).Inthesesituations,intrinsicintentionalitywillemergefromtheinclusivefitnessofthemembersoftheevolvingpopulationintheenvironmenttowhichitisadapting,justasitdoesfornaturalpopulations.Thereforewecanseethatunderawidevarietyofcircumstances,theconsciousstatesofrobotswillhaveintrinsicintention-alityandthusgenuinecontent;theirconsciousnesswillbeconsciousnessofsomething, as itmustbe. (Imention inpassing that emotions,whichhavemanyimportantconnectionstoconsciousness,areimportantinalltheseap-plicationsofautonomousrobotics.)
Recentworkoncomputer-based investigati-onsoftheevolutionoflanguageandcommu-nicationcanbefound inCangelosi&Parisi(2001) and Wagner, Reggia, Uriagereka &Wilkinson (2003); unfortunately the latter
incorrectly classify our own experiments asnonsituatedcommunication.SeeMacLennan(2006)foramoredetaileddiscussionoftheseissues.
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial408
to theHardProblem in thecontextofhumanconsciousnessandas itmayariseinthecontextofrobotconsciousness.TheHardProblem,whichaddressestherelationofourordinaryexperienceofsubjectiveawarenesstothescientificworld-view,isarguablytheprincipleproblemofconsciousness(MacLennan,1995),andsoitwillbeworthwhiletosayafewwordsaboutwhatmakesitsohard.7Therootoftheproblemistheuniqueepistemological statusofphenomenal consciousness (Block,1995),forconsciousexperienceistheprivateandpersonalgroundofall observa-tion,whereas traditionallysciencehasbeenbasedonspecific observationsthatarepublicand,inthissense,non-personal.Wearedealingwithseveralinterrelatedepistemologicalissues.First,inorderthatitsconclusionsmaybegenerallyapplicable,sciencestrivestobeapublicenterprise,andsoitisbasedonpubliclyvalidatedobservations,whereastheexperienceofconsciousawarenessisinherentlyprivate.(Verbalaccountsofconsciousawarenesscan,ofcourse,bepublic,butassumingthattheyareveridicalbegsthequestionoftheHardProblem.)Ontheotherhand,itisimportanttorecognizethatallobservationisultimatelyprivate,andthatinconsciousnessstudies,asinthemoredevelopedsciences,abodyofpublicfactscanemergeasaconsensusfromthereportsoftheprivateexperiencesoftrainedobserversofdifferingtheoreticalcommitments.Iwilladdressthesortoftrainingrequiredbelow.Sincethegoalofscienceispublicknowledge(knowledgetrueforallpeople),scienceseekstoseparatetheobserverfromtheobserved,foritwantsitscon-clusionstobefoundedonobservationsthatareindependentoftheobserver.This isnot feasiblewhen theobjectofscientific investigation isconsciousexperience,forconsciousnessconstitutesthestateofobservation,comprisingboththeobserverandtheobserved,thefundamentalrelationofintentional-ity,asdescribedbyBrentano(1995,p.89)andHusserl(1931,p.34):intentio(stretching,direction,attention)towardsanobject.Consciousnessisthevec-torof intentionalityextendingfromtheobserver to theobserved,andso itinvolvesthembothessentially.Further, scienceordinarily strives to separate the individual, subjective as-pectsofanobservation(e.g., feltwarmth) fromtheobjectiveaspects (e.g.,measuredtemperature),aboutwhichitiseasiertoachieveaconsensusamongtrainedobservers.However, in theHardProblemtheindividual,subjectiveaspectsareofcentralconcern.Also, sciencenormally takesa third-personperspectiveonthephenomenaitstudies(it, he, sheis,does,etc.),whereastheexperienceofconsciousawarenessisalwaysfromafirst-personperspective(I feel, perceive, remember, etc.). Indeed, the Hard Problem addresses thequestionofwhy,inafundamentalsense,thereevenisafirst-personperspec-tive.Indeed,itmightseemthatanobjectivescienceofsubjectiveexperienceisim-possible,acontradictioninterms,butthisimpressionresultsfromaconfusionofterminology.HereIuse“subjective”and“objective”torefer,respectively,toprivate,first-personexperienceandtopublic,third-personobservation.Of-ten,however,weunderstand“objective”tomean“unbiasedorfactual”(andthereforegood),and“subjective”tomean“biasedordistorted”(andthereforebad).AsSearle(1992,p.19)suggests,progressonthemind-bodyproblemhasbeenimpededbyapun!Ofcourse,thedescriptiveandevaluativeusagesofthesetermsarenotunrelated,butourgoalhereisobjective(i.e.,unbiased,factual)knowledgeofsubjective(i.e.,first-person,private)phenomena.
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial409
The inherently first-person, subjective character of conscious experiencemakesitresistant totheordinaryreductivepatternsofscience,for it is thethird-person,publiclyobservableaspectsofphenomenathataremostame-nable to reduction to more fundamental physical processes. For example,oncetheprivateexperienceoffeltwarmthhasbeenseparatedfromthepub-licmeasurementoftemperatureandheat,thelattercanbereducedtomorefundamentalphysicalproperties(meankineticenergyofmolecules).Indeed,although third-person objects, properties, and processes can be reduced tootherthird-personobjects,properties,andprocesses,itisacategorymistaketoattempttoreducefirst-personphenomenatothethird-personobjects,prop-erties,orprocesses.Nevertheless,thereisakindofreductionthatisapplica-bletosubjectivephenomena,asexplainedbelow.
Observing Consciousness
Theuniqueepistemologicalstatusofconsciousexperiencemakesitdifficulttoinvestigatebyscientificmeans,butnotimpossible;hereIwillsummarizetheapproachthatIhaveadvocated(MacLennan,1995,1996a,inpress).FirstIwilladdressthequestionofhowwecanobserveconsciousness(i.e.,lookatit),whenallobservationisbymeansofconsciousness(i.e.,looksthroughit).Ananalogywillmaketheapproachclear.Theimageformedbyacameramustpassthroughthecamera’saperture;inthissense,wecantakeapictureofsomeobject(analogoustothecontentofconsciousness),butwecannottakeapictureoftheapertureitself(analogousto observing consciousness). Nevertheless, it is possible to investigate theaperture, because it affects the image in systematic ways (e.g., brightness,diffraction,focus,depthoffield). Inparticular,wecaninvestigatechangesthatoccurwithsystematicvariationof theaperture. In thisway,character-isticsof theaperturemaybe separated from the specificsof the image. Insomecasestheseobservationsarefacilitatedbytheuseofasimpleobject,suchasapointlightsourceoraganzfeld(homogeneousfield),whichrevealssomecharacteristicsoftheaperture,butobscuresothersthatarepeculiartocompleximages.8Therefore,armedwithinsightsgainedfromsimpleimages,itisalsonecessarytoexploretheaffectsoftheapertureoncompleximages.Inanycase,manyofcharacteristicsoftheaperturewillbeunapparenttothenaiveobserver,butwithtrainingtheyareuncovered,andprovidethebasisforabodyofpublicfacts.Trainedinvestigatorswillbeabletoexploretheaffectsofvaryingtheapertureonallimages,andtherebydiscoverobjectiverelationships.This analogy suggests an approach to observing consciousness. Althoughconsciousness cannot be separated from its content, trained observers canseparateaspectsoftheconsciousstatethatdependmoreonitscontentfromthosethatdependonconsciousnessitself.Asinthecameraanalogy,inves-tigationmaybefacilitatedbyconsciouscontent that issimple instructure,
B. J. MacLennan, Consciousness: NaturalandArtificial410
asoccurs incontemplationandmeditation (e.g., emptiness,one-pointcon-centration).Moregenerally,consciousnesscanbeinvestigatedinlaboratorysituationsthatattempttocontrol itscontent.However,asouranalogysug-gests, such approaches reveal only some characteristics of consciousnesswhileobscuringothers.Therefore,itisessentialalsotoinvestigateordinary,everydayconsciousstates,aswellasalteredstatesthataccentuateparticularcharacteristics.Itwillbeapparentthatspecializedtrainingandexperiencearenecessarytoobserve the relevantphenomena,as theyare inall sciences,butespeciallyin the scientific studyof consciousness.Experimentalphenomenologyandphenomenologicalpsychology(e.g.,McCall,1983),since theydirectlyad-dressthestructureofphenomena(consciousexperience),seemtoprovidethebestfoundation.Ihde(1986)showshowsystematicvariationofsimplepheno-menacanhelptorevealthestructureofconsciousness.
Neurophenomenology
Initsliteralsense,aphenomenon(Greek,phainomenon)isanythingthatap-pears(phainetai)inconsciousness;amongthekindsofphenomenaareper-ceptions, thoughts, recollections, plans, intentions, volitions, desires, fears,anticipations,andhallucinations.Butphenomenaarenotindependent;theyexistininterrelationshipsofsequenceandpossibility.Thisnetworkofactualandpotentialphenomenaconstitutesaphenomenal world.Phenomenologyisfundamentallythestudyofthestructureofphenomena,thatis,oftheinvari-antstructureofphenomenalworlds(thestructureindependentofindividualvariation).Sinceanadequatescientifictheoryofconsciousnessmustaccountforqualiaandtheirintegrationintoaphenomenalworld,phenomenologyisfundamentaltotheconsciousnessresearch.By using phenomenological techniques, investigators can avoid an overlysuperficialperspectiveonphenomena,oftenbasedonana priori theoreti-calcommitment.Considerawell-knownexamplefromHusserl’sCartesian Meditations(1960,§§17–19).Supposesomeonerotatesanordinarydiewith-inmyview.Whatwouldbeanaccuratephenomenologicaldescriptionofmyexperience?Onemightsupposethatitmightbeanaccountofneutralvisualdataintermsofchangingconfigurationsofblackellipsesinwhiteparallelo-grams,butthisisnotanaccuratedescription.Infact,Idonotexperienceab-stractellipsesandparallelograms;Iexperiencearotatingcubemarkedblackspots.Indeed,sinceIamacquaintedwithdice,Iwillexperiencearotatingdie.Therecognitionofthisfamiliarthree-dimensionalobjectisanaspectofthephenomenon.Furthermore,thephenomenonisnotconfinedtotheinstantane-ouspresent;bymeansofshort-termmemoryitextendsintotherecentpast(retention,inHusserl’sterms),andbymeansofanticipationitextendsintothenear-termfuture(protention);thisactualandpotentialsequentialstructuregivesthephenomenonitstemporalunity(e.g.,Husserl,1973,§23).Therearealsonon-visualanticipationsandassociations,forweexpectthedietohaveacertainhardnessandweight.Violationsofcertainexpectations(e.g.,discove-ringthatthediehasnoback,orisextremelyheavy)leadtoakindofbreak-down,andachangeinourintentionalrelationtotheobject.Thestructureofthediephenomenonisnotlimitedtoperception,buthaspsychologicalandsocialaspects.Forexample,Iwillassociatethediewithgamesandgambling(andwhateverconnotationthatmayhaveforme),andImayevenexperience
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial411
thedisplayeddieasan invitation tosomesport.All this ispartof therichphenomenologyofsosimpleathingasadie.Accuratephenomenologydependsonawarenessandinvestigationofallas-pectsof thephenomena, a skill that requires significantphenomenologicaltraining.Thetechniqueisfarfromnaiveintrospectionism,whichis,indeed,naive.IfwewanttosolvetheHardProblem,thatis,tounderstandtherelationofconsciousnesstophysicalprocesses,wecannotrelyonphenomenologyalone,butmustintegratephenomenologicalobservationwithneuroscientifictheoryandexperiment.Eachdomainof investigationmaycontribute to theother.Forexample,weknowthatrapidandslowmotionsareprocesseddifferentlyinthebrain(Weiskrantz,1995),whichshouldalertustolookforcorrespond-ingphenomenologicaldifferences.Ontheotherhand,thephenomenologicalsubtletiesofcolor(discussedlater)implycorrespondingneurologicalproc-esses.Therefore, thescientificinvestigationofconsciousnessmustbe, inabroadsense,neurophenomenological.9
Protophenomena
Neurophenomenological Reduction
Thevalueofreductionisthatitallowsustounderstandcomplicatedsystemsbetterbyrelatingthemtosimplersystems.(Reductionismostfruitfulwhenitdoesnotlimititselftounderstandinghowthepartsconstitutethewhole,butalsoconsiderstheroleofthewholeintheconstitutionoftheparts;thisises-peciallythecaseinthebiological,psychological,andsocialsciences.)There-fore,althoughareductionofthesubjectivetotheobjectiveisfundamentallyimpossible,wecanaccomplishareductionofthesubjectivetothesubjective(thatis,areductionofsubjectivephenomenatotheirsubjectiveconstituents)and,further,correlatethissubjectivereductiontoaparallelreduction,intheobjectivedomain,ofneuropsychologicalprocesses to theirconstituentbio-logicalandphysicalprocesses.Reductioninthesubjectivedomaincanbeaccomplishedbyobserverstrainedinphenomenologicalprocedures,whichallowthemtoarriveataconsensusconcerningthestructureofconsciousawarenessasexperiencedbyallpeople.(Thereisalreadyaconsiderablebodyofresults,inthepsychologicallitera-tureaswellasthephenomenologicalliterature.)Aswe’veseen,insightsandresultsfromeachofthesedomains–whichwemaycallthephenomenologi-caland theneurological–cansuggesthypothesesandotherwiseguide theinvestigationsoftheother.Asafirststepwecanattemptaqualitative reduction,essentiallya“separationofvariables”,ofphenomenaonthebasisofsensorymodality;forexamplevisualphenomenaofallsorts(perceptions,memories,etc.)canbeseparatedfromauditoryphenomena.Thustheconsciousstateisdecomposedintophe-nomenaofdifferentkinds.Evenhere,however,wemustbewareofoversim-plification,forneurologicalresearchhasshownthatsomeneuronsisauditory
B. J. MacLennan, Consciousness: NaturalandArtificial412
cortexrespondtovisualstimuli,andconverselyneuronsinvisualcortexcanrespond toauditory stimuli, thus facilitating face-to-facecommunication.10Thissuggeststhatostensiblyvisualphenomenaarenotpurelyvisual,noraresupposedauditoryphenomenapurelyauditory,anditisreasonabletosupposethatthesamemixtureoccursamongothersensorymodalities(assuggestedalsoby thedie example).Therefore, a qualitative reduction canbe at bestapproximate,asweshouldexpectfrombothpurephenomenologyandevolu-tionarypsychology(i.e.,visualdominance).Incontrasttoaqualitativereduction,whichdecomposesphenomenaonthebasisofkind,itispossibletoperformaquantitative reduction,whichdecom-posesphenomenaonthebasisofsize.Thisapproachissuggestedbyphilo-sophicalconsiderations,butalsobyneuroscience.Inparticular,topographic mapsandothercomputational mapsareubiquitousinthebrain(Anderson,1995, ch. 10; Knudsen, du Lac & Esterly, 1987). For example, in sensoryareasthedimensionsofastimulusaresystematicallymappedontocorticalregions.Themostfamiliarexampleisthesomatotopic mapinsomatosensorycortex,inwhichcorticallocationcorrespondstobodilylocation,buttherearesimilarbodymapsinmotorareas.Invisualareasthereareretinotopic maps,inwhichneurallocationcorrespondssystematicallytoretinallocation.Themappeddimensionsofthestimuluscanbemoreabstract.Forexample,inau-ditorycortextherearetonotopic maps,inwhichneurallocationcorrespondstofrequency,andinbatauditorycortex,echolocationisaidedbymapsencod-ingDopplershift(Suga,1985,1989).Although there is much that we do not know about neural representation,theseexamples suggest thatmany representationscanbedecomposed intoelementary units (i.e., individual neurons, or small groups of them),11 thatareessentiallysimilarinfunctionanddistinguishedonlybytheirlocationinsomecomputationalmap.Furthermore,atleastinprimarysensoryareas,ithasbeenpossibletorelateactivityintheseneuronstoelementaryconstituentsofstimuli(e.g.,pressureonaparticularpatchofskin,lightofcertainwave-lengthsonaparticularretinal location), thereceptive fieldsof theneurons.Thisisallintheneurologicaldomain,butwecanperformaparallelreductioninthephenomenologicaldomain,forweareawarethat,forexample,visualphenomenahaveparts,suchasourexperiencesofcoloratdifferentlocationsinthevisualfield(anobservationthatappliestovisualhallucinationsasmuchas to ordinary perception).The elementary components of a phenomenon,then,wouldcorrespondtothesmallestunitsofthecorrespondingneuralrep-resentation(presumably,activityinindividualneurons,butotherpossibilitiesareconsideredbelow).Thus,neurologically-informedphenomenologicalreduction(whichwemaycallneurophenomenological reduction)suggeststhatitmaybefruitfultoun-derstandconsciousexperienceintermsofprotophenomena,whicharetheo-retical entities hypothesized as the elementary constituents of phenomena.Wefurtherhypothesizethateachprotophenomenonhasanintensity(asortoffundamentalquale)representingitspresenceinconsciousness(e.g.,experi-encedpressureonapatchofskin,experiencedbrightnessofapatchofcolorinthevisualfield).Thisintensityisthesubjectiveexperiencecorrespondingtoneuralactivityintheneuralstructuresassociatedwithaprotophenomenon(itsactivity site).(Iwilldiscussactivitysitesinmoredetailbelow.)Thesimplestkindsofprotophenomenaaresimilartoelementarysensedata(suchas“red-here-now”).Forexample,ifweconsidervisualexperience,wecanthinkofitasconstitutedoftinypatchesofcolorandbrightness,muchlike
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial413
pixels,atvariouslocationsinthevisualfield.12However,neurosciencesug-geststhatostensiblyvisualprotophenomenawillalsohaveanauditoryaspect,andviceversa.Furthermore,protophenomenaarenotlimitedtoelementarysensedata,butalsoincludetheelementaryconstituentsofmorecomplexphe-nomena,includingexpectations,moods,feelings,recollections,imaginations,intentions,andinternaldialogues.Inanycase,neurophenomenologysuggeststhatprotophenomenaarevery small comparedwithphenomena,andone’sconsciousstatemightcomprise10to100billionprotophenomena(thenumberofneuralactivitysitesassociatedwithprotophenomena).Protophenomenalinterdependencies are also much more complex than suggested by the no-tionofelementarysensedata(asisdiscussedbelow),sowemustbewareofanoversimplifiedor superficialunderstandingofprotophenomena. Indeed,as neurons both sense their cellular environments (via chemical receptors)andactontheirenvironment(bygeneratingactionpotentials),somostpro-tophenomenahaveanactivecharacter,inthattheirpresenceinconsciousnessconditionsthepresenceorabsenceofotherprotophenomena.Weidentifyone’sphenomenalworldwiththetotalityoftheirprotophenome-na,butthismayseemtoleadtoa“jaggedness”or“grain”problem(Chalmers,1996,pp.306–8)intheabsenceofsomeadditionalfactortounifythepro-tophenomenaintoawhole,butthisisnotthecase.Consideramacroscopicobjectsuchasachair;itisawholebecauseitsconstituentatomsareboundtogether, so that theirmacroscopicmotionsarecoherent.Similarly,aswillbe explained, the intensities of protophenomena are mutually interdepend-ent,andaphenomenon isnomore than thecoherentactivityofmassesofprotophenomena.Soalso,theunityofconsciousnessisaconsequenceoftheunityof thenervoussystems(see“TheUnityofConsciousness”and“TheUnconsciousMind”below).
Ontological Status
Since, in a philosophical context, a phenomenon is anything that appearsinconsciousness,phenomenaare,bydefinition,observable(indeed,fromafirst-personperspective).Paradoxically,protophenomena,whicharetheele-mentaryconstituentsofphenomena,arenot,ingeneral,observable.Thisisbecauseundernormalcircumstancesprotophenomenaareexperiencedonlyaspartsofwholephenomena,whichtypicallycomprisemillionsofprotophe-nomena(aswillbeexplainedbelow), so thatachange inoneprotopheno-menonwouldrarelybenoticed(i.e.,causeonetobehavedifferently).Asananalogy: the changeof onepixel in a high-resolution image is unlikely tohaveanypracticaleffect.Similarly,changingonemoleculeofamacroscopicobject(suchasachair) isunlikely tohaveanoticeableeffect.Conversely,justasboundandcoherentlymovingatomsconstituteamacroscopicobject,soboundandcoherentlyvaryingprotophenomenaconstituteaphenomenonpresentinconsciousness(protophenomenalinterdependenciesarediscussedlater).Wemaysaythattheprotophenomenaconstitutingaphenomenonhave
11
Such as microcolumns, containing perhapselevenneurons(Jones,2000).
12
The primary protophenomena of visual ex-perienceappear,infact,tobemorecomplex
thenpixels;psychophysicalevidencesuggeststheir brightness profiles are more like spa-tiotemporalGaborwavelets(Pribram,1991);seealsoMacLennan(1991)forasurvey.
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial414
essential subjectivity,butarenotthemselvesphenomena.Thatis,protophe-nomenaarenotthesameaslittlephenomena.Theapparentunobservabilityofprotophenomenaraisesquestionsabouttheirexistence.Inourcurrentstateofknowledgeitisperhapsbesttoviewthemastheoretical entities,whichmeanstheyarepostulatedfortheirexplanatoryvalueinthetheoryandarevalidatedbytheirfruitfulnessforscientificinquiry(Hempel,1965,pp.177–9;Maxwell,1980,pp.175–84).Theirontologicalstatusiscomparabletothatofatomsduringthenineteenthandearlytwentiethcenturies,when theycouldnotbeobserveddirectly.Physicistsmighthavediffered(especiallyinthenineteenthcentury)aboutwhetheratomsreallyex-ist,buttheyallagreedonthescientificvalueofatomictheory.(Incontempo-raryphysics,quarksandstringsareunobservedtheoreticalentities.)Thereareotherpossibilities.Forexample,protophenomenamightbeemer-gent properties of sufficiently large or complex brains, but this possibilitydoesnotnecessarilyimplythattheyarenotrealorthatthereissomecriti-calneuralmassbelowwhichtheydonotexist.Again,ananalogywillhelp.Soundisacompressionwaveinamediumsuchasair,andsuchawavecanbeunderstoodbyassigningapressuretoeachpointinavolumeofspace.Weknowthisisamathematicalfiction,sinceairiscomposedofdiscretemole-cules,anditmakeslittlesensetotalkofthepressureofoneortwomoleculesorevenofasmallnumberofthem.Nevertheless,soundandpressuredistribu-tionsareperfectlyobjectivepropertiesofmacroscopicvolumesofair.Soalsowemayfinditismeaningfultotalkofprotophenomenaonlyinthecontextofmacroscopicneuralmass.
Activity Sites and Protophenomenal Intensity
Parallelreductioninthephenomenologicalandneurologicaldomainsleadsto the conclusion that there areactivity sites in thebrain corresponding totheprotophenomena,and that somekindofphysicalprocessatanactivitysitecorrespondstotheintensity(strength)ofthecorrespondingprotopheno-menoninconsciousexperience.Itisimportanttounderstandthataprotophe-nomenonanditsactivitysitearetwomutuallyirreducibleaspectsofasingleunderlyingreality(andthusprotophenomenatheoryisakindofdouble-as-pect monism).13
Unfortunately,Idonotbelievethatwecanidentifytheactivitysitesatthistime.Somereasonablepossibilities includesynapsesandneuralsomata, inwhich cases the intensity of the associated protophenomenon might corre-spond toneurotransmitter flux,boundneurotransmitter receptors, ormem-branepotential.FollowingSherrington,whosaid,“Reflexactionandmindseemalmostmutuallyexclusive–themorereflexthereflex,thelessdoesmindaccompanyit”,Pribramhasarguedthatconsciousnessisassociatedwiththegradedelectricalactivity in thedendritic treesofneurons, rather thanwithall-or-nothingactionpotentialgeneration.14On thisbasiswewouldexpectsynapses to be the activity sites and protophenomenal intensity to be cor-relatedwithneurotransmitterflux,boundreceptors,orpre-orpostsynapticmembranepotential.15Arelatedpossibilityisthatneuralsomataaretheactiv-itysites,andthatintensitycorrespondstosomaticmembranepotential,whichisalsograded;otherpossibilitiesareconsideredbelowin“ConsequencesandIssues”.Inanycase,theseareallscientificquestions,whichcanbeaddressedempirically.
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial415
Aspreviouslydiscussed,aprotophenomenonhasadegreeofpresenceincon-sciousness,whichwecallitsintensity(thinkofthebrightnessofthered-here-nowforaconcreteexample),andwehypothesizethatthisintensityiscorre-latedwithsomephysicalpropertyoftheactivitysite,forexamplemembranepotential,neurotransmitterorionflux,orthenumberofboundreceptors.Thesimplesthypothesisisthatprotophenomenalintensityissimple,nonnegative,scalarquantity(representingdegreeofpresence),butthereareotherpossibili-ties.Forexample,protophenomenaassociatedwithdifferentneurotransmit-ters might have different kinds of intensities, and consequently a differentexperiencedpresenceinconsciousness;thisisanempiricalquestionthatcanbeansweredbyexperimentalphenomenology.
Protophenomenal Dependencies
Animportant issueiswhatdistinguishes,forexample,aprotophenomenonfor “red-here-now” fromone for “middle-C-here-now”, that is,whatgivesprotophenomenatheirqualitativecharacter?Theparallelquestionintheneu-rosciencedomainsuggestsananswer, forneurons invisualcortex, forex-ample,arenotessentiallydifferentfromthoseinauditorycortex.Certainlythesensoryreceptorsaredifferent,buteveninthesenseorgansthereisnoimportantdifferencebetween,forexample,aconerespondingtocertainopti-calwavelengthsatoneplaceontheretinafromthosewiththesameresponseatotherplaces.Rather,thestructureofthesensoryworldisdefinedbytheinterconnectionsamongneurons.Forexample,thespatialstructureofvisionisdefinedbypatternsofconnectionsthatcauseneuronstorespondtoedges,lines,center-surroundpatterns,andotherspatialstructures.Protophenomenaseemtobeorganizedaccordingtosimilarprinciples.Thatis, thetime-varyingintensitiesofprotophenomenaarecorrelatedwitheachotherinaccordwithquantifiableprotophenomenal dependencies;inprinci-plethesecorrelationscanbedescribedbydifferentialequations(MacLennan,1996b,inpress).Thatis,theintensityofeachprotophenomenonisacompli-catedfunctionoftherecentintensitiesofthousands(ortensorhundredsofthousands)ofotherprotophenomena,aswellasofextrinsic variables,thatis,ofvariablesexternaltothephenomenologicaldomain.Asaconsequence,thephenomenalworldisnotcausallyclosed,buttheprotophenomenaldepend-enciesconstrainthepossibilitiesofchangeinconsciousstate,subjecttotheextrinsicvariablesandotherinfluencesdiscussedbelow.Itisreasonabletosaythatprotophenomenahavenoqualitiesoftheirown;theyhaveonlytheirintensities(whicharequantities);protophenomenahavequalitiesonlybyvirtueoftheirinterdependencewithotherprotophenomena.
13
More specifically, protophenomena theoryisanexampleofwhatChalmers(2002)calls type-F monism, which is in the heritage ofRussell (1927). Jung’s phenomenologicalpsychology led him to similar conclusions:“psycheandmatteraretwodifferentaspectsofoneandthesamething”for“thebiologi-cal instinctualpsyche,graduallypassesoverintothephysiologyoftheorganismandthusmerges with its chemical and physical con-ditions” (Jung,1960,§418,420). SeealsoJung&Pauli(1955)andStevens(2003,pp.79–88).
14
See Miller, Galanter & Pribram (1960, pp.23–4) and Pribram (1971, pp. 104–5, 1991,pp.7–8).
15
Thispossibilityisdevelopedmathematicallyin MacLennan (1996b,Appendix); see alsoMacLennan(1999b).
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial416
Therefore,qualiaareemergentpropertiesinaphenomenalworldstructuredbyprotophenomenaldependencies; thatis, thisisessentiallyastructuralisttheoryofqualia.Phenomena are experienced “out there”– in our physical bodies or in thespacearoundthem–andonlyrarelyinsideourheads,wherecorticalneuralactivityoccurs.Weseeobjects (andevenhallucinations)aroundus,not inourvisualcortices,andwefeelpainsinourfingersortoes,notinoursomato-sensorycortices.Whydoweexperienceactivityinonecorticalneuronasapaininafinger,andinanotherasapaininatoe?Topographicmapsinthebrainsuggestananswer,forspatialrelationsinthemapmirrorspatialrela-tionsamongthestimuli.16Butspatialproximityinthecortexisnotinitselftheprimaryfactor(althoughdiffuseelectricalandchemicaleffectsarepos-sible,andthebrain’sEMfieldmayplayarole);rather,thekeyfactoristhatintopographicmapsnearbyneuronsaremorelikelytobeconnectedthanaremoredistantneurons.Interactionsamongnearbyneuronsgenerateatopology(anabstract systemofneighborhood relationships),whichcreates thephe-nomenalspaceintowhichourexperiencesareprojected.Sinceprotophenom-enaldependenciescorrespondtophysicaldependenciesamongtheiractivitysites,protophenomenaldependenciesdefinethetopologyofthephenomenalworld,whichisamajoraspectofitsphenomenology,thatis,ofthepossiblestructureofphenomena(MacLennan,1999b).Recent experiments by Sur (2004) support the dependence of phenomenalqualityonneuralinterconnection.Retinalaxonsinnewbornferretswerein-ducedtoprojectintoauditorycortex(areaA1,viathethalamus)inonehemi-sphere,butprojectedtotheirnormaltargetsinvisualcortex(V1)intheother.Asaconsequence,theauditorycortexthatreceivedretinalinputself-organ-izedintoorientationmapslikethoseinprimaryvisualcortex.Furthermore,andmostsignificantly,neurophysiologicalandbehavioraltestsimpliedthattheferretswereexperiencing visual perceptions intheir“rewired”auditorycortices.
B. J. MacLennan, Consciousness: NaturalandArtificial417
Asmentionedabove,phenomenalworldsarenotcausallyclosed;protophe-nomenaldependenciesdonotcompletelydeterminethedynamicsofaphe-nomenalworld.Theprincipalnon-phenomenologicalcausesaretheextrinsicvariablescorrespondingtosensoryinputs.However,otherphysicalprocessescanalsoaffectthephenomenalworld.Forexample,sicknessandalcoholorothermind-alteringsubstancescan temporarilyaffectprotophenomenalde-pendencies. More permanent changes to one’s phenomenology can resultfromstrokes,braintumorsorinjuries,Alzheimer’sdisease,andthelike.Theincompletenessofphenomenologicalcausalitymightseemtoimplythatthephenomenalworldisultimatelyepiphenomenal,andthatprotophenome-naltheoryisunnecessaryinthepresenceofa(presumably)causallycompletephysicaltheory.However,thisfamiliarperspectiveignorestheHardProblem,sinceitdoesnotaddressphenomenalconsciousnessatall;thatis,asubstan-tialbodyofevidenceremainsunexplained.Incontrast,theprotophenomenalapproachallowsareductionwithinthesubjectivedomain,thecorrelationofelementarysubjectivitywithphysicalprocesses,andtheeventualintegrationofconsciousnessintothescientificworldview.
Consequences and Issues
Inverted Qualia
Theideaofacolorspectruminversiondatesbackat least toLocke’s1690Essay Concerning Human Understanding (e.g.,Hardin,1988;MacLennan,1999a;Nida-Rümelin,1996;Palmer,1999). Is itpossible that Iexperiencephenomenal redness when I perceive short wavelengths (normally experi-encedasviolet),andviceversa?Neurophenomenologicalreductionandtheprotophenomenalapproachprovidemeansforansweringthesequestionsem-pirically.Toillustrate theapproachIwillbeginwithasimplerproblem:anauditoryspectruminversion.ItmightseemconceivablethatIexperienceasapheno-menalhighpitch thesamesoundfrequencies thatyouexperienceasa lowpitch,andviceversa,butthisapparentpossibilityrestsonasuperficialphe-nomenologyofpitch,whichcanbeexposedbysystematicvariationof thephenomena.Ononehand, ifwegradually increase the subjectivepitchofa sound, we will discover a limit beyond which we cannot go (in percep-tionorperceptualimagination).Ontheotherhand,ifwegraduallydecreasesubjectivepitch,wefindthatitcomestobeexperiencedmorearhythmandultimatelyasaperiodicvariationofloudness.(Tobemorespecific,frequen-ciesabove,say,100Hzareexperiencedaspitch,whereasthosebelowabout10Hzareexperiencedasrhythm;intermediatefrequenciesareexperiencedinamixedway.)Thusexperimentalphenomenologydemonstrates thatourexperienceoflowpitchesisdistinguishedfromthatofhighpitchesinthattheformerareinherentlycontinuouswithourexperiencesofrhythmandloud-ness.
16
Additional evidence comes from “referredpain”, a medical condition in which pain inonepartofthebodyistransferredtoanotherpartthatisnotnearbyinthebody,butwhosecorticalmapsareadjacent. Thismayoccur,for example, because of cortical remapping
after lossofabodypart (e.g.,Karl,Birbau-mer,Lutzenberger,Cohen&Flor,2001).
B. J. MacLennan, Consciousness: NaturalandArtificial418
Thisphenomenologicalanalysisisreinforcedbyneuroscience,forhigherfre-quenciesaremappedspatially inauditorycortex in tonotopicmaps,whichlimit the representable frequencies (in perception but also imagination) atboththehighandlowends.However,atlowerfrequencies(about5Hzandbelow),nerveimpulsesbecomesynchronizedwiththesoundwaves(i.e.,thefrequenciesarerepresentedtemporallyratherthanspatially),arepresentationlikethatofarhythm(Adelman,1987,p.91;Suga,1994,pp.299–300;seealsoBendor&Wang,2005).Thereforeamoresystematicneurophenomeno-logicalanalysisofsoundshowsthattheallegedspectralinversionisimpos-sible;abnormalitiesinneuralstructurewouldmanifestinexperience,becausethephenomenonoflowpitchessentiallyincludesaspectsofrhythm,whichhighpitchesdonot.Wecanapplysimilartechniquestoinversionsinvisualqualia.Thesimplestcase is an inversionbetweenphenomenaldark (which I’ll denoteΦ-Dark)andphenomenallight(Φ-Light).Infact,itisimpossiblebecause,asFrancisBacon(Essays,3)remarked,“Allcolorswillagreeinthedark.”Inparticular,theexperienceofΦ-Darkdoesnotadmitofdifferingcolorexperiences.Thepossibilityofacolorinversionissuggestedbytheideaofalinearcolorspectrum,whichisaconsequenceofinaccuratephenomenologycontamina-tedbyknowledgeofthephysicsoflight(thelineardimensionofwavelength)andanalogieswithsound (pitchandwavelength). Indeed,prior toNewtoncolorwaslesslikelytobeunderstoodasalinearspectrum,buthisdiscoveryof the color spectrum established the idea that color is a one-dimensionalphenomenon(Gage,1993).SinceHering’s(1878)developmentofthedou-ble-opponenttheoryofcolorvision,however,ithasbeenapparentthatcolorhasamorecomplextopology,whichisalsosupportedbyneuroscience(e.g.,DeValois&DeValois,1988,1993;Kaiser&Boynton,1996).Phenomenalhuehasacirculartopologystructuredbytwoaxesbetweenop-posingcolors,whichmaybetermed(approximately)theyellow-blueandred-greenaxes(hence,“double-opponent”).Theaxesaredefinedbyfour“uniquehues”(unique-yellow,unique-blue,unique-red,unique-green),whichareex-periencedasbeingunmixedwithanyothercolors.(Forexample,theexperi-enceofunique-greendoesnothaveanymixtureofblueoryellowinit.)Thewavelengthsof light that areperceivedas theseuniquehuesvaries a littlefrompersontoperson,buttheyareanessentialaspectofthephenomenologyofnormalhumancolorvision.(Actually,thereisnosinglewavelengththatproducestheexperienceofunique-red,buttheexperiencecanbecreatedbymixinginblueandyellowwavelengthswithredlight,sothattheycanceleachotherontheyellow-blueaxis;moreonthisbelow.)Therefore,atabasiclevel,humancolorexperienceisdefinedbythreeaxes:yellow-blue,red-green,andlight-dark(YB,RG,andLD,respectively),whichdefineacolor sphere.Thisstructuresuggestsanumberofpossibilitiesforanomalouscolorvision,forwecanentertainexchangesoftheopposedcolors(e.g.,anexchangeofyellowandblue)orexchangesofentireaxes(e.g.,anexchangeofYBwithRG)(cf.Palmer,1999).Aspreviouslydiscussed,Φ-LightandΦ-Darkarephenomenologicallydif-ferentinstructure,andthereforecannotbeexchanged,soIwillfocusonthemoreinterestingcolorexchanges.IndeedthephenomenologicaldifferencesbetweenΦ-LightandΦ-Darkprovideabasisforcolorphenomenology,sinceithasbeenrecognizedsinceancient times(e.g.,Aristotle,De Sensu,442a)thatyellowandbluearethecolorsmostcloselyrelatedtolightanddark(i.e.,whiteandblack);indeed,wemaycallyellowandbluethechromicanalogs
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial419
ofwhiteandblack.Pre-Newtonianlinearcolortheoriesoftenunderstoodthecolorsasintermediariesbetweenwhiteandblack,withyellowandbluebe-ingclosesttotheextremes(Gage,1993).Likewise,moderncolorresearchersrefertothe“yellowanomaly”,whichreferstothefactthatphenomenalyel-low(Φ-Yellow)istheintrinsicallybrightesthue.Thereforetheexperiencesofunique-yellowandunique-bluearephenomenologicallydistinctfromeachotherandfromtheothercolorsbyvirtueoftheirrelationstoΦ-LightandΦ-Dark.Furthermore,ifsomeonehadanexperienceofΦ-Yellowwhenperceiv-ingshort-wavelength(blue)light, theabnormalitywouldbedetectable(fortheywouldreporta“blueanomaly”).ThereforeanundetectableYBinversionisimpossible.Asremarked,thephenomenologicalcharacteristicsofΦ-YellowandΦ-BlueprecludeanundetectableexchangeoftheYBandRGaxes,butaninversionoftheRGaxismightseempossible,sinceonthecolorwheelΦ-RedandΦ-GreenarebothintermediatebetweenΦ-YellowandΦ-Blue(butonoppositesidesofthewheel).AsolutiontothisproblemmaybefoundinthecolortheoryofGoethe(1840),whowasaverycarefulphenomenologist.AlthoughhiscriticismofNewto-nianopticsisoftenviewedas“anembarrassinglapseinthelifeofanother-wisegreatman”,Goethehadamoreaccurateaccountofthephenomenologyofcolor,whichinfactcomplementsNewton’saccount,whichwasbetterforthedevelopmentofphysicaltheory.Indeed,itisnotsurprisingthatGoethethepainterwouldunderstandcolordifferently thanNewton the theoreticalandexperimentalphysicist;whereasNewtonexploredpurewavelengthssplitoutofpurewhite lightbyaprism,Goetheinvestigatednaturallyoccurringcolorinthesky,clouds,plants,andminerals(also,insomecases,bymeansofaprism).Goetheobservedthatbothredandgreenareexperiencedasmeansbetweentheextremesofyellowandblue,butmeansofadifferentkind.Ononehand,greenisasimpleintermediatebetweenyellowandblue,similartoboth,eventhoughunique-greenincludesnoblueoryellow(Goethe,1840,§697).Ontheotherhand,reddoesnothavethisrelationshipbut,accordingtoGoethe,by a process of phenomenological “augmentation” (Steigerung) of yellowandblue(§699–703),onecanproduceaverypurered(Purpur),“likefinecarmineonwhiteporcelain”(§792).(Inthisconnectionit’sworthrecallingthat unique-red is anon-spectral hue, that is, it is an experienceable colorthatcannotbeproducedbymonochromaticlightanddoesnotoccurinthecolorspectrum.)Thuswehaveabasisforthephenomenologicaldistinctionofthefouruniquehues,whichissupportedbytheneuropsychologyofvisualperception.Thisphenomenologicalanalysisissupportedbythecross-culturalstudiesofBerlinandKay(Berlin&Kay,1969;Kay&McDaniel,1978;Saunders&vanBrakel,1997).Ifaculturehastwobasiccolorterms,theyarenominallyequivalenttowhiteandblack,buthavedenotationsclosertowarm-brightandcool-dark,effectivelyΦ-YellowandΦ-Blue.Iftheyhaveathirdbasiccolorterm,itisapproximatelyequivalenttored,andafourthisgreen.Similarly,Goetheclassifiesredasthethirdprimarycolor(afteryellowandblue),andmakesgreenthefirstsecondarycolor.Therefore,colorphenomenology, theneuropsychologyofcolorvision,andstudiesofcross-culturalcolorcategorizationall imply that the fouruniquehuesarephenomenologicallydistinct,thateachhasanindividualcharacter,andthereforethatanomaliesincolorvisionwouldbedetectable.Wecancon-
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial420
clude thatundetectablecolor inversionsare impossible.Furthermore, theseneurophenomenologicalinvestigationsprovideabasisforconstructingato-pologyofcolorexperiencethatismoreaccuratethanalinearscaleorsimpledouble-opponent color wheel (MacLennan, 1999b). In some cases we canpredict the phenomenology of neurologically abnormal vision, and similarapproachesallowustoatleastbegintoconstructtheperceptualexperiencesofnon-humananimals.Wehaveseenthattheplausibilityofspectralinversionsdependsonsuperfi-cialphenomenologicalanalysis,whereasmorecarefulneurophenomenologi-calinvestigationbeginstorevealtheinevitablestructureofperceptualexperi-ence.Therefore,itisworthrecallingthatevenapparentlysimplephenomena,suchascolororpitch,haveconnectionstootheraspectsofourexperience.Infact,itisamistaketoassumethatcolortermsreferprimarilytowavelengthsoflight.Forexample,translatorshavebeenperplexedbytheancientGreekwordchlôros(whichnominallymeansgreen)becauseancienttextsapplyittoblood,dew,tears,andotherthingsthatarenotgreenincolor(Gage,1993,p.272n7;Zajonc,1993,p.15).TheexplanationisthatinancientGreek,asinEnglish,thingsthataremoist,green,orlivingcanbedescribedas“green”;forexample,wecanspeakofagreentwigoragreenriderwithoutmeaningthey are green in color. Similarly, many other color terms were originallymonovalenttermsforminerals,dyes,andothersubstances,andseemtobepolyvalentonlywhensupposedtorefertorangesofwavelengths;forexam-ple,Medieval scarletsmaybegreen,blue,black,orwhite incolor (Gage,1993,pp.34–5).Therefore;wemustexpectthatacomprehensivephenom-enologyofcolor (andotherperceptualqualities)will includeanextensivepenumbraofmaterial,emotional,andotherassociations,bothphylogeneticandontogenetic.
The Unity of Consciousness
Theprotophenomenalapproachcanprovidesomeinsightsintothequestionoftheunityofconsciousness.Forjustasthereisnoreasontopostulatearei-fiedphenomenontointegratethecoherentactivitiesofprotophenomenaintoawhole,sothereisnoreasontopostulateaseparatelyexistingsubjecttoin-tegratethetotalityofphenomenaintoaunifiedconsciousexperience.Rather,theunityofconsciousnessconsistsinthedensenetworkofinterdependenciesamong theprotophenomena,which is the causal nexus of thephenomenalworld.Thisconclusionissupportedbyempiricalevidencefromcerebralcommis-surotomies (split-brain operations), which sever the corpus callosum, thethickbandof800millionnervefibersthatconnectsthecerebralhemispheres(e.g.,Gregory,1987,pp.740–7).Theeffectistoseparateoneconsciousness(onesubject)intotwo,whichisjustwhatwewouldexpect,sincethesurgeryeliminatesconnectionsbetweenactivitysites,andthusremovesdependenciesbetweenprotophenomena.Indeed,wemaysuppose thatas thenerve fibersaresevered theprotophe-nomenaassociatedwiththetwohemispheresareprogressivelydecoupled;sotheonephenomenalworldgraduallydividesintotwo,whichimpliesthattheunityofconsciousnessisamatterofdegree.Idon’tknowiftheexperimentalevidenceisavailable,buttheclaimcertainlyhasempiricalcontent.Onekindofevidenceresultsfromfactthattheseoperationsleavethebrainstem intact.Thus someconnectionsbetween thehemispheres remain,pro-
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial421
ducingtwolooselycoupledphenomenalworlds,whereaswewouldexpecta complete bilateral section of the brain to produce two completely inde-pendentphenomenalworlds.Interestingly,ithasbeenobservedthatthetwohemispheresof thesepatientsmaycommunicatewitheachotherbymeansof“externaltransactions”,suchastwitchingtheskinoftheface,aprocessofwhichbothsubjectsare,apparently,unconscious.Nevertheless,thesetrans-actions establish loose dependencies between the two phenomenal worlds(inonepatient’shead),whichdifferonly indegree from thedependenciesestablished when two people (each their own phenomenal world) interact.Mirrorneurons,whichmimictheactivityofneuronsinanotherperson’sbrain(Rizzolatti&Craighero,2004),alsosuggestthatdifferentindividuals’pheno-menalworldsmaybemorecloselyconnectedthanwehavebeenaccustomedtothink.Theseconsiderationssuggestthattheunityofconsciousnessshouldbeplacedonacontinuumthatincludesprogressivelymorelooselycoupledphenomenalworlds.
“Everythingofwhich Iknow,butofwhich I amnotat themoment thinking;everythingofwhichIwasonceconsciousbuthavenowforgotten;everythingperceivedbymysenses,butnotnotedbymyconsciousmind;everythingwhich,involuntarilyandwithoutpayingattentiontoit,Ifeel,think,remember,want,anddo;allfuturethingsthataretakingshapeinmeandwillsome-timecomeintoconsciousness:allthisisthecontentoftheunconscious.”(Jung,1960,§382)
However,thereareatleastthreewaysinwhichprotophenomenaltheorycanaccommodatetheunconsciousmind.First,recallthatprotophenomenaarenotphenomena;althoughprotopheno-menabearelementarysubjectivity,typicallytheyarenotindividuallysalientinconsciousness.Onlybycoherentactivitydoprotophenomenaemergeasdistinct phenomena in the conscious state. Conversely, incoherently activeprotophenomenaformasortofbackgroundnoiseintheconsciousstate.Wemaycomparethemotionofairmolecules,whichissalientonlyifcoherent(windorabreeze),butisunnoticedifitisrandom.Thebrainstem,midbrain,andrightcerebralhemispherehavebeenmentionedas likelysubstrates for theunconsciousmind (e.g.,Stevens,2003,ch.13),but there is no reason to suppose that neurons in these areas do not haveprotophenomena, whereas those in the (manifestly conscious) left cerebralhemispheredo.Experimentswithsplit-brainpatientssuggestaresolutionofthisparadox,fortheirbrainshousetwoconsciousminds,eachunawareoftheother,thatis,eachanunconsciousmindfromtheperspectiveoftheother.Aspreviouslydiscussed,thehemispheresarecapableoflimitedcommunicationbymeansoftheintactbrainstemand“externaltransactions”,butthesecom-municationsfromonehemisphereareexperiencedbytheotherhemisphericconsciousnessasinexplicable“hunches”,justlikethosefromtheunconscious(Gregory,1987,p.743).Thereforewemayhypothesizethatthenormalbrainhousesseverallooselycommunicatingconsciousnesses,thatis,severallooselycoupledphenomenalworlds,eachaconsciousnessinitself,butexperiencingtheothersasuncon-
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial422
sciousminds.18ThiswouldaccordwithJung’s(1960,§253)observationthatunconsciouscomplexesandarchetypesoftenbehaveasautonomousperson-alities,whointeractwithegoconsciousnessbymeansofhunches,intuitions,compulsions,resistances,moods,dreams,andavarietyofneuroses.Itmight seemunlikely that thesesemi-independentconsciousmindscouldexistunseeninthenormalbrain,butwemustrecallthatinitiallyitwasnotobviousthatthesplit-brainpatientshadtwoconsciousminds;theyappearedperfectly normal until laboratory testing revealed anomalies. Similarly, inthe normal human it may be the mind that includes the verbal and motorprotophenomenathatismostabletomanifestitsexistenceinbehaviorandismosteasilyidentifiedwiththeego.Otherminds,whicharemoreremotefrom the verbal and motor protophenomena (in terms of protophenomenalcontrol),arelessabletomanifesttheirexistenceinobservablebehavior;theynormallyescapenotice.Sothesecondprotophenomenalexplanationoftheunconsciousmindisthatitisnotunconsciousinitself,butonlyfromtheper-spectiveofegoconsciousness.ThethirdexplanationisbasedonthehypothesisofPribramandSherrington,discussedabove, that conscious experience is associatedwithgradedelec-trochemicalprocessesinthedendrites,butnotwiththeall-or-nothinggene-rationofactionpotentialsintheaxons.19Thatis,theactivitysitesresideinthedendrites,butnotintheaxons.ThishypothesisaccordswellwithJung’saccountof“thearchetypesofthecollectiveunconscious”,whichhedescribedascontentlessperceptual-behavioralpatternsgroundedinourbiological(andevenphysical)nature:
Accordingtoprotophenomenaltheory,protophenomenaareassociatedwithactivity sites in the brain, and the structure of a phenomenal world corre-spondstotheinterconnectionsamongtheactivitysites.Feweractivitysitesimplyfewerdegreesoffreedominaphenomenalworld.Therefore,wewouldexpectanimalswithsimplernervoussystemsthanourstohavecorrespond-inglysimplerphenomenalworlds(fewerdegreesoffreedom,simplerstruc-ture).20However,thereareanumberofissuesthatcannotberesolvedwithout
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial423
AccordingtoCook’stheory,whileneuralfiringis thephysicalcorrelateofconsciousness(experience),physicalprocessesinthedendritesarethecor-relatesofcognition(informationprocessing).Others, more controversially, have suggested that consciousness is associ-atedwiththebrain’selectromagnetic(EM)field(John,2002;McFadin,2002,2007;Pockett,2000,2002,2007),andevidencehasbeenadducedthatitcanaffectneuronfiring(McFadden,2002).Morespecifically,McFaddenhypothe-sizes (1) thatneural firing induces anendogenous EM field, that this fieldinfluencesneuralactivity,andthatthisfeedbackthroughtheendogenousEMfield is essential to neural information processing, and (2) reportable con-sciousexperience(i.e.,consciousexperiencethatcanresult inpubliclyob-servablebehavior) isassociatedwithacomponentof this field thataffectsmotorneurons.21
18
This hypothesis does not exclude the firstpossibility, namely that protophenomena insomeareas, such as thebrainstemandmid-brain,arenotsufficientlycoherenttoconsti-tutephenomena.
19
See Miller, Galanter & Pribram (1960, pp.23–4) and Pribram (1971, pp. 104–5, 1991,pp.7–8).
B. J. MacLennan, Consciousness: NaturalandArtificial424
IfMcFadden’shypothesesarecorrect,thenthereareseveralinterestingim-plicationsforthetheoryofprotophenomena.ThefirsthypothesisimpliesthattheEMfieldcanmediateinteractionsamongactivitysites,andthereforethatfieldeffectsmightberelevanttoprotophenomenalinterdependencies,whichcouldbemorediffuseandholisticthanthosecorrespondingtoneuralstruc-tures(seebelowonprotophenomenaldependencies).Thesecondhypothesisraises thepossibility that someactivity sitesmaybe located in theendog-enousEMfield.ThispossibilityisreinforcedbyDennisGabor’s(1946)analysisoftheinfor-mationcarryingcapacityofarbitrarysignals(reviewedinMacLennan,1991).Heapplied theHeisenberg-Weylderivationof theUncertaintyPrinciple toproveaminimumjointlocalizationinanytwoconjugatevariables(e.g.,timeandfrequency),andthereforethatanyfinite,band-limitedsignalhasamaxi-mumnumberofdegreesoffreedomthatmaybeusedtoconveyinformation,its logon content.22Thismaximum is achievedbydecomposing the signalintoasuperpositionofGabor wavelets(Gaussian-modulatedcomplexexpo-nentials,equivalenttothepure statesofquantummechanics),whichareineffectquantaofinformation(calledlogons).Informationisrepresentedinthe(complex-valued)coefficientsofthelogons.AsaconsequencethephysicalactivitysitesarelocalizedbutdistributedpatchesoftheEMfieldofvariousspatialfrequencieswithvariousorientations;theymaybevisualizedasori-entedgratingpatches.Activityisrepresentedintheamplitudeandphaseofeachpatch,whichraisesthequestionofhowtheamplitudeandphaseoftheprotophenomenacoulddifferentlyaffectconsciousexperience.Todeterminethelogoncontentofthebrain’sendogenousEMfield,therele-vant conjugate variables are area and spatial frequency. McFadden (2002)statesthatthespatialresolutionofthefieldissmallerthan1mm.Fromacor-ticalareaof2200cm2wecancalculateanapproximatelogoncontentof2200cm2/(0.1cm)2=220000logons.23Iftheresolutionwereasfineas0.1mm(whichisstillquitecoarseinneuralterms;microcolumnshavediametersanorderofmagnitudesmaller:Jones,2000),thenthefieldcouldsupportapprox-imately22millionlogons.Therefore,ifthelogonsofthebrain’sendogenousEMfieldareactivitysites, theneachofourphenomenalworldscomprisessomehundredsofthousandormillionsofprotophenomena(theintensitiesofwhicharecorrelatedtothecorrespondingGaborcoefficients).OfcoursetheexistenceofactivitysitesintheEMfielddoesnotcontradicttheirexistenceinneuronsaswell.Theseissuescanbeaddressedempirically,butIdonotthinkwehavethetechnologyyet.
Nonbiological Consciousness
What Physical Processes Have Protophenomena?
Ihavediscussedprotophenomenaintermsofhumanconsciousness,butitisnowtimetoconsider themalso in thecontextof robotconsciousness.Thecrucialquestioniswhetherrobotbrainscanbemadesufficientlysimilartohumanbrainsin the relevant ways.Thiscanbeexplainedbyanalogy.Liquid-ityisapropertyofwater,butitdependsonmorefundamentalphysicalprop-ertiesofH2Omolecules,suchastheirfinitevolumeandmutualattractionatclosedistances.Therefore,othersubstancesthathavethesesamefundamentalproperties, but are otherwise dissimilar to water, may be liquid. Similarly,protophenomenaareaconsequenceofcertain(currentlyunknown)physical
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial425
propertiesofactivitysites.Theymightbequitespecifictoneurons,ortheymightoccurinotherphysicalsystemsaswell.Inthelattercase,itwouldbereasonabletosupposethatnonbiologicalsystemswiththesepropertieswouldhaveartificialactivitysitesandcorrespondingprotophenomena.Iftheactiv-itysiteswereappropriatelystructured(alsoverypoorlyunderstood),thentheprotophenomenawould cohere into phenomena and constitute a consciousstate.Obviously,thequestioncannotbeansweredwithoutadequateknowledgeoftheactivitysitesassociatedwith theprotophenomenaofhumanconscious-ness,butIcanoutlinesomeofthepossibilities.Suppose protophenomena are associated with neural somata and that pro-tophenomenalintensitycorrespondstothemembranepotential.Iftherobot’sbrainisnotmadefrombiologicalneurons,thenthequestionbecomeswhetherthebiologicalcharacteroftheneuronisanecessaryconditionforittohaveanassociatedprotophenomenon.If,ontheotherhand,thepresenceofapro-tophenomenondependsonlyoncertainelectrochemicalprocessesoccurringinthecellbody,itmightbepossibletoconstructanartificialdeviceimple-mentingthoseelectrochemicalprocessesandthereforehavinganassociatedprotophenomenon.(Bytheway,itisdifficult,thoughnotimpossible,toan-swerthisquestionempirically,forphenomenologicalobservationcanestab-lishthepresenceorabsenceofcoherentensemblesofprotophenomena,andperhapsinsomecasesofisolatedprotophenomena.)Supposeinsteadthatprotophenomenaareassociatedwithsynapsesandtheirintensitywithneurotransmitterflux.Thisraisesafurtherquestion(whichcanbe answered empirically): are protophenomena associated with all neuro-transmittersandtheirreceptors,oronlywithcertainones?Ifonlywithcertainones,thenwehavethefurtherempiricalquestionofwhycertainneurotrans-mittersshouldbeassociatedwithprotophenomenabutnotothers.Whatistherelevantdifferencebetweentheneurotransmittersorbetweentheirreceptors?When we know the answer to this question, then we can say whether theconstituentsofarobot’sbrainhavetherelevantpropertiestohaveprotophe-nomena.If, on the other hand, as Cook suggests, protophenomenal intensity corre-spondstotheopeningofthecelltoitsenvironmentandionfluxthroughthemembrane,thenwewillneedtodiscoverwhetheranysuchboundaryopeningsufficesforprotophenomenalintensity,oronlyinthecontextofalivingcellmaintainingitsexistenceasanentitydistinctfromitsenvironment.Similarly,ifMcFaddeniscorrectinhisconnectionofthebrain’selectromag-neticfieldwithconsciousexperience,thentoanswerthequestionforrobotswewillneedtounderstandwhataspectsofthemutualcouplingofneuronsandtheirEMfieldarerelevanttoconsciousexperience.Insummary,althoughthesequestionsarecomplexanddifficult,theyarenotunanswerable.Theexperimentsarechallenging,butnotimpossible.AveryinterestingpossibilityisraisedbyChalmers(1996,ch.8).Wehaveseen thatprotophenomenaareessentiallyquality-lessand that theyacquiretheirqualitiesonlythroughtheirmutualinterdependencies;thatis,thesubjec-
22
Gabor’stheorytreatsstructural information,whereasShannon’sbetter-knowntheorytreatsselective information; they are complemen-tary (seeCherry,1978,pp.47–49;MacKay,1969,pp.178–189;MacLennan,1991).
23
The exact value depends onhow spatial re-solutionismeasured(seeMacLennan,1991),buttheorderofmagnitudeiscorrect.
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial426
tivequalityisstructuredbyformalrelationsamongabstractquantities(pro-tophenomenalintensities).(Althoughabstract,theyareexperienced,fortheintensityofaprotophenomenonisthedegreeofitspresenceinconsciousex-perience.)Consistentlywiththis,Chalmerssuggeststhatphysically realized information spacesmightprovidethelinkbetweenthephenomenologicalandphysicaldomains.Whensuchasystemisobservedfromtheoutside,wemaygiveaphysicalaccountofitsbehavior,butwhenitisexperiencedfromtheinside,thatis,whenIamthephysicalinformationsystem,thenImayhaveasubjectiveexperienceoftheinformationprocesses.Inotherwords,physicallyrealizedinformationspacesmaybeexperiencedobjectivelyfromtheoutsideorsubjectivelyfromtheinside.Appliedtoprotophenomena,thistheoryimpliesthatanyphysicallyrealizedinformationspacemightbeanactivitysitewithanassociatedprotopheno-menon.Therefore,iftheconstituentsofarobot’sbrainimplementphysicallyrealized informationspaces,as theysurelymust, then theywouldhaveas-sociatedprotophenomena.Thisdoesnot,initself,implythattherobotwillhaveconsciousexperience,fortheprotophenomenamustbeinterdependentinsuchaswayastocohereintophenomena(i.e.,consciouscontent),butiftherobot’sbrainwerestructuredtoimplementthefunctionsofconsciousnessdis-cussedinSection2,thenconsciousexperiencewouldseemtobeinevitable.IfChalmer’s idea is correct, thenwemust askwhat constitutes somethingasaphysicallyrealizedinformationspace.WehaveShannon’sandGabor’scomplementaryinformationtheories,whichallowustoquantifyinformationandchangesininformationstate.Forexample,wecanquantifytheinforma-tionreceivedwhenanionchannelopensoraligandbindstoareceptoronacellmembrane, information that isused ingoverning latercellularproc-esses(MacLennan,inpress).Itisplausiblethatthesechannelsandreceptorsareactivitysites,andthattheionfluxorreceptoractivationcorrespondstoprotophenomenalintensity.Ifthisistrue,thenprotophenomenaneednotbeconfinedtoneuronsoreventoeukaryoticcells.Ifwetakeafurtherstep,andacceptWheeler’s(1994)ontologicalmaxim,“itfrombit”,whichassertsthatallphysicalprocessesarefundamentallyinfor-mationprocesses,thenwemustentertainthepossibilitythatallfundamentalphysicalprocesses(suchasquantumstatechange,orobjectivewavefunctioncollapse)haveassociatedprotophenomena.24Thisdoesnotimplythatcom-puters,theearth,ortheentireuniverseareconscious,forthatwouldrequirethattheprotophenomenaactinasufficientlycoherentandstructuredmannertoconstitutephenomena.Thiswouldbepanpsychism,amuchstrongerclaimthanpanprotophenomenalism,whichassertsonlythatelementarysubjectivityaccompaniesphysicalprocesses(astrongenoughclaimalready,tobesure!).Panprotophenomenalismdoesnotimplyubiquitousconsciousness.25Interest-ingthoughthesespeculationsmaybe,atthistimeweneedtofocusourinves-tigationsontheonlyprotophenomenathatweknowexist:thoseassociatedwithhumanbrains.
B. J. MacLennan, Consciousness: NaturalandArtificial427
capacity to suffer. Cruel practices, such as vivisection, have been justifiedbytheclaimthat“beasts”(non-humananimals)are“justmachines”,aviewthatbecamewidespreadwiththeascendancyofthemechanicalphilosophyofGassendi andDescartes. (According to this philosophy, humans–or atleastsomehumans!–wereconsideredmorethanmachinesbecausetheyhave“immortalsouls”;incontrast,animalswereconsideredsoulless.)Nowadays,although there is ongoingdebate about the existence and extent of animalrights,wedoacknowledgeanimalsufferingandtrytoavoidit(atleastforsomeanimals:cattle,butchickens?lobsters?oysters?).26SoIthinkitislikelythatwewill face similar issues regarding sophisticatedautonomous robots(especiallythosemadeoutoforganicmaterials).Amore immediate reasonforworryingabout theHardProblemfor robotsisthatitisavaluabletestcaseforourunderstandingofourownconsciousselves.Ifwecannotgiveaprincipledexplanationwhyrobotscanorcannothavesubjectiveexperiences,thenwedonotunderstandourownconscious-nessverywell.Solongaswecannotanswerthequestionforrobots,theex-planatorygapbetweenmindandmatterremains.
4. Conclusions
The“lesshard”problemsofconsciousnessrelatetoitsfunctionsinpercep-tion,cognition,andbehavior,whichinthecaseofanimalscanbedeterminedbyreferencetotheselectiveadvantageofthesefunctionsinthespecies’envi-ronmentofevolutionaryadaptedness.Sincethesefunctionsarealsovaluableforautonomousrobots,Ianticipatethatrobotswillhavetoimplementthesefunctionsaswell,whichwillrequiresolvingthe“lesshard”(butneverthelessverydifficult!)problemsoffunctionalconsciousnessanditsphysicalmecha-nisms.Closelyrelatedtoconsciousnessistheissueofintentionality,the“aboutness”offunctionallyconscious(andother)brainstates.Iarguedthatintrinsicin-tentionalityisgroundedintherelevanceofanagent’srepresentationstothecontinuedexistenceoftheagentoritsgroup,andsointentionalityislargelyindependent of consciousness; indeed, very simple agents (organisms andmachines)canexhibitgenuineintrinsicintentionality.Nevertheless,trulyau-tonomous robotsmust takecare for the survivalof themselvesandothers,andsointrinsicintentionalitywillcharacterizemanyoftheirinternalstates,includingfunctionallyconsciousstates.Finally,IturnedtotheHardProblem–howwecanreconcilephysicalmecha-nismwith the experienceof subjective awareness– and addressed it fromtheperspectiveofneurophenomenologyandthetheoryofprotophenomena.Unfortunately,thepossibilityofa(sufficientlycomplex)robothavingsubjec-tiveexperiencecannotbeansweredwithoutabetterunderstandingoftherela-tionofprotophenomenatotheirphysicalactivitysites.Iconsideredseveralpossibilitiesdiscussedintheliteratureandtheirimplicationsforrobotcon-sciousness.Perhapsthemostintriguingandparsimoniouspossibilityisthat
24
In this connection it is interesting to recallthat Gabor’s quantum of information, thelogon, is mathematically identical to a pure state in quantum mechanics, and obeys thesameUncertaintyPrinciple.
Bekoff,M.(2007)The Emotional Lives of Animals. A Leading Scientist Explores Animal Joy, Sorrow, and Empathy – and Why They Matter,NewYork:NewWorldLibrary.
Berlin,B.,andKay,P.(1969)Basic Color Terms. Their Universality and Evolution,Ber-keley:UniversityofCalifornia.
Blackburn,S. (1994)The Oxford Dictionary of Philosophy,Oxford:OxfordUniversityPress.
Block,N. (1995) “Ona confusionabout a functionof consciousness”,Behavioral and Brain Sciences,vol.18,pp.265–66.
Brentano,F.(1995)Psychology from and Empirical Standpoint,trans.A.C.Rancurello,D.B.Terrell,andL.L.McAlister,London&NewYork:Routledge.
Bridgeman,B. (1982)“Multiplexing in singlecellsof thealertmonkey’svisual cortexduringbrightnessdiscrimination”,Neuropsychologia,vol.20,pp.33–42.
Brooks,R.A.(1987)“AHardwareRetargetableDistributedLayeredArchitectureforMo-bile Robot Control”, in Proceedings IEEE Robotics and Automation, Raleigh, NC, pp.106–110.
Burghardt,G.M.(1970)“Defining‘Communication’”,inJ.W.Johnston,Jr.,D.G.Moul-ton,andA.Turk(eds.),Communication by Chemical Signals,NewYork:Appleton-Cen-tury-Crofts,pp.5–18.
Buss,D.M.(2004)Evolutionary Psychology. The New Science of the Mind,2nded.,Bo-ston:Pearson.
Chalmers,D.J.(2002)“Consciousnessanditsplaceinnature”,inPhilosophy of Mind: Classical and Contemporary Readings,D.Chalmers(ed.),Oxford:Oxford.
Cherry,C.(1978)On Human Communication,Cambridge,MA:MIT.
Cook,N.D. (2000)“Ondefiningawarenessandconsciousness:The importanceof theneuronalmembrane”,inProceeding of the Tokyo-99 Conference on Consciousness,Sin-gapore:WorldScientific.
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial429
Cook,N.D.(2002a)“Bihemisphericlanguage:Howthetwohemispherescollaborateintheprocessingoflanguage”,inThe Speciation of Modern Homo Sapiens,T.Crow(ed.),London:ProceedingsoftheBritishAcademy.
Cook,N.D.(2002b)Tone of Voice and Mind. The Connections Between Intonation, Emo-tion, Cognition and Consciousness,Amsterdam:JohnBenjamins.
Gutenplan,S.(ed.)(1994)A Companion to the Philosophy of the Mind,Oxford:Blackwell.
Hardin, C.L. (1988) Color for Philosophers. Unweaving the Rainbow, Indianapolis &Cambridge:Hackett.
Haugeland,J.(ed.)(1997)Mind Design II. Philosophy, Psychology, Artificial Intelligence,Cambridge,MA:MITPress.
Heidegger,M.(1962)Being and Time,trans.J.MacquarrieandE.Robinson,NewYork:Harper&Row.
Heidegger,M.(1982)The Basic Problems of Phenomenology,trans.A.Hofstadter,Bloo-mington:IndianaUniversityPress.
Hempel,C.G.(1965)Aspects of Scientific Explanation and Other Essays in the Philosop-hy of Science,NewYork:TheFreePress.
Hering,E.(1878)Zur Lehre vom Lichtsinne,Wien:CarlGerold’sSohn.
Husserl,E.(1931)Ideas. General Introduction to Pure Phenomenology,trans.W.R.B.Gibson,NewYork:Macmillan.
Husserl,E.(1960)Cartesian Meditations. An Introduction to Phenomenology, trans.D.Cairns,TheHague:Nijhoff.
Husserl, E. (1973) Experience and Judgment. Investigations in a Genealogy of Logic,trans.J.S.ChurchillandK.Ameriks,Evanston:NorthwesternUniversityPress.
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial430
Ihde,D.(1986)Experimental Phenomenology. An Introduction,Albany:StateUniversityofNewYorkPress.
John,E.R. (2002)“Theneurophysicsofconsciousness”,Brain Research Reviews,vol.39,pp.1–28.
Jones,E.G.(2000)“Microcolumnsinthecerebralcortex”,Proceedings of the National Academy of Sciences USA,vol.97,pp.5019–21.
Jung,C.G.(1960)The Structure and Dynamics of the Psyche(Collected Works,vol.8),2nded.,trans.R.F.C.Hull,Princeton:PrincetonUniversityPress.
Jung,C.G.(1969a)The Archetypes and the Collective Unconscious(Collected Works,vol.9,parti),2nded.,trans.R.F.C.Hull,Princeton:PrincetonUniversityPress.
Jung,C.G.(1969b)Psychology and Religion: West and East(CollectedWorks,vol.11),2nded.,trans.R.F.C.Hull,Princeton:PrincetonUniversityPress.
Jung,C.G.,andPauli,W.(1955)The Interpretation of Nature and the Psyche,trans.R.F.C.HullandP.Silz,London:Routledge&KeganPaul.
Kaiser,P.K.,andBoynton,R.M.(1996)Human Color Vision,2nded.,Washington:Opti-calSocietyofAmerica.
Karl,A.,Birbaumer,N.,Lutzenberger,W.,Cohen,L.G.,andFlor,H.(2001)“Reorganiza-tionofmotorandsomatosensorycortexinupperextremityamputeeswithphantomlimbpain”,The Journal of Neuroscience,vol.21,pp.3609–18.
Knudsen,E.J.,duLac,S.,Esterly,S.D.(1987)“Computationalmapsinthebrain”,Annual Review of Neuroscience,vol.10,pp.41–65.
Laughlin,C.D.,Jr.,McManus,J.,andd’Aquili,E.G.(1990)Brain, Symbol and Experien-ce. Toward a Neurophenomenology of Consciousness,Boston:NewScienceLibrary.
MacKay,D.M.(1969)Information, Mechanism and Meaning,Cambridge,MA:MIT.
MacLennan,B.J.(1990)Evolution of Communication in a Population of Simple Machines(TechnicalReportUT-CS-90–99),Knoxville:UniversityofTennessee,Knoxville,Depar-tmentofComputerScience.
MacLennan,B.J.(1991)Gabor Representations of Spatiotemporal Visual Images(Tech-nicalReportUT-CS-91–144),Knoxville:UniversityofTennessee,Knoxville,DepartmentofComputerScience.
MacLennan,B.J.(1992)“Syntheticethology:Anapproachtothestudyofcommunica-tion”,inArtificial Life II. The Second Workshop on the Synthesis and Simulation of Living Systems,C.G.Langton,C.Taylor,J.D.Farmer,andS.Rasmussen(eds.),RedwoodCity:MITPress,pp.631–658.
MacLennan, B. J. (1995) “The investigation of consciousness through phenomenologyandneuroscience”,inScale in Conscious Experience. Is the Brain Too Important to be Left to Specialists to Study?,J.KingandK.H.Pribram(eds.),Hillsdale:LawrenceErlbaum,pp.25–43.
MacLennan, B. J. (1996a) “The elements of consciousness and their neurodynamicalcorrelates”,Journal of Consciousness Studies,vol.3,pp.409–424.
MacLennan,B.J.(1996b)Protophenomena and their Neurodynamical Correlates(Tech-nicalReportUT-CS-96–311),Knoxville:UniversityofTennessee,Knoxville,DepartmentofComputerScience.
MacLennan,B.J.(1999a)“Neurophenomenologicalconstraintsandpushingbackthesub-jectivitybarrier”,Behavioral and Brain Sciences,vol.22,961–3.
MacLennan,B.J.(1999b)The Protophenomenal Structure of Consciousness with Especial Application to the Experience of Color: Extended Version(TechnicalReportUT-CS-99-418),Knoxville:UniversityofTennessee,Knoxville,DepartmentofComputerScience.
SYNTHESISPHILOSOPHICA44(2/2007)pp.(401–433)
B. J. MacLennan, Consciousness: NaturalandArtificial431
MacLennan,B. J. (inpress) “Protophenomena:Theelementsof consciousness and theirrelation to the brain”, Irreducibly Conscious. Selected Papers on Consciousness, A.Batthyany,A.Elitzur,andD.Constant(eds.),Heidelberg&NewYork:UniversitätsverlagWinter,ch.X(pp.189–214).
Maxwell,G.(1980)“Theontologicalstatusoftheoreticalentities”,inIntroductory Rea-dings in the Philosophy of Science,E.D.Klemke,R.Hollinger,andA.D.Kline(eds.),Buffalo:PrometheusBooks.
McCall,R.J.(1983)Phenomenological Psychology. An Introduction. With a Glossary of Some Key Heideggerian Terms,Madison:UniversityofWisconsinPress.
McFaddin,J.(2002)“Synchronousfiringanditsinfluenceonthebrain’selectromagneticfield:Evidenceforanelectromagnetictheoryofconsciousness”,Journal of Consciousness Studies,vol.9,pp.23–50.
Stevens,A.(2003)Archetype Revisited. An Updated Natural History of the Self,Toronto:InnerCityBooks.
Suga,N.(1984)“Theextenttowhichbiosonarinformationisrepresentedinthebataudi-torycortex”,inG.M.Edelman,W.E.Gall,andW.M.Cowan(eds.),Dynamic Aspects of Neocortical Function,NewYork:Wiley,pp.315–73.
Suga,N.(1989)“Principlesofauditoryinformation-processingderivedfromneuroetholo-gy”,Journal of Experimental Biology,vol.146,pp.277–86.
Sur,M.(2004)“Rewiringcortex:Cross-modalplasticityanditsimplicationsforcorticaldevelopmentandfunction”,inHandbook of Multisensory Processing,B.Stein(ed.),Cam-bridge,MA:MITPress.
Wagner,K.,Reggia,J.A.,Uriagereka,J.,andWilkinson,G.S.(2003)“Progress in thesimulation of emergent communication and language”, Adaptive Behavior, vol. 11 (1),pp.37–69.
Weiskrantz,L.(1995)“Blindsight:Consciousvs.unconsciousaspects”,inScale in Con-scious Experience. Is the Brain Too Important to be Left to Specialists to Study?J.KingandK.H.Pribram(eds.),Hillsdale:LawrenceErlbaum.
Wheeler,J.A.(1994)“Itfrombit”,inAt Home in the Universe,J.A.Wheeler,Woodbury:AmericanInstituteofPhysicsPress.
Zajonc,A.(1993)Catching the Light. The Entwined History of Light and Mind,NewYork:BantamBooks.
Bruce J. MacLennan
Natürliches und künstliches Bewusstsein
ZusammenfassungAusgehend von Erkenntnissen der Evolutionären Psychologie untersucht dieser Beitrag wich-tige Funktionen, die das Bewusstsein autonomer Roboter ausfüllen kann. Gemeint sind will-kürlich kontrolliertes Handeln, bewusstes Wahrnehmen, Eigenwahrnehmung, Metaerkenntnis sowie Bewusstsein des eigenen Selbst. Der Verfasser unterscheidet zwischen intrinsischer In-tentionalität und Bewusstsein, führt jedoch das Argument ins Feld, dass es ebenso wichtig sei, die Erkenntnisweise eines Roboters zu verstehen. Abschließend wird, aus dem Blickwinkel der Theorie von den Protophänomenen, das für Roboter „schwierige Problem” untersucht, d.h. die Frage, ob sie zu subjektiver Wahrnehmung fähig sind.
B. J. MacLennan, Consciousness: NaturalandArtificial433
Bruce J. MacLennan
La conscience, naturelle et artificielle
RésuméEn s’appuyant sur les résultats de la psychologie évolutionniste, nous examinons les différen-tes fonctions importantes que puisse remplir la conscience dans les robots autonomes : action contrôlée, prise de conscience, conscience de soi, métacognition, conscience du moi. Nous dis-tinguons l’intentionnalité intrinsèque de la conscience, mais soutenons également l’importance de la compréhension de la cognition robotique. Enfin, nous étudions le « Hard Problem » con-cernant les robots, c’est-à-dire la question de savoir s’ils peuvent connaître une prise de consci-ence subjective, dans une perspective de la théorie du protophénomène.