Vous êtes sur la page 1sur 24

08/09/2016

BadNews:NoamChomsky

AReviewofB.F.Skinner'sVerbalBehavior
byNoamChomsky
"AReviewofB.F.Skinner'sVerbalBehavior"inLanguage,35,No.1(1959),2658.
Preface
Prefacetothe1967reprintof"AReviewofSkinner'sVerbalBehavior"
AppearedinReadingsinthePsychologyofLanguage,ed.LeonA.JakobovitsandMurrayS.Miron
(PrenticeHall,Inc.,1967),pp.142143
byNoamChomsky
Rereadingthisreviewaftereightyears,IfindlittleofsubstancethatIwouldchangeifIweretowriteit
today.Iamnotawareofanytheoreticalorexperimentalworkthatchallengesitsconclusionsnor,sofarasI
know,hastherebeenanyattempttomeetthecriticismsthatareraisedinthereviewortoshowthattheyare
erroneousorillfounded.
IhadintendedthisreviewnotspecificallyasacriticismofSkinner'sspeculationsregardinglanguage,but
ratherasamoregeneralcritiqueofbehaviorist(Iwouldnowprefertosay"empiricist")speculationastothe
natureofhighermentalprocesses.MyreasonfordiscussingSkinner'sbookinsuchdetailwasthatitwasthe
mostcarefulandthoroughgoingpresentationofsuchspeculations,anevaluationthatIfeelisstillaccurate.
Therefore,iftheconclusionsIattemptedtosubstantiateinthereviewarecorrect,asIbelievetheyare,then
Skinner'sworkcanberegardedas,ineffect,areductioadabsurdumofbehavioristassumptions.My
personalviewisthatitisadefinitemerit,notadefect,ofSkinner'sworkthatitcanbeusedforthispurpose,
anditwasforthisreasonthatItriedtodealwithitfairlyexhaustively.Idonotseehowhisproposalscanbe
improvedupon,asidefromoccasionaldetailsandoversights,withintheframeworkofthegeneral
assumptionsthatheaccepts.Idonot,inotherwords,seeanywayinwhichhisproposalscanbesubstantially
improvedwithinthegeneralframeworkofbehavioristorneobehaviorist,or,moregenerally,empiricistideas
thathasdominatedmuchofmodernlinguistics,psychology,andphilosophy.TheconclusionthatIhopedto
establishinthereview,bydiscussingthesespeculationsintheirmostexplicitanddetailedform,wasthatthe
generalpointofviewwaslargelymythology,andthatitswidespreadacceptanceisnottheresultofempirical
support,persuasivereasoning,ortheabsenceofaplausiblealternative.
IfIwerewritingtodayonthesametopic,IwouldtrytomakeitmoreclearthanIdidthatIwasdiscussing
Skinner'sproposalsasaparadigmexampleofafutiletendencyinmodernspeculationaboutlanguageand
mind.Iwouldalsobesomewhatlessapologeticandhesitantaboutproposingthealternativeviewsketchedin
Sections5and11andalsolessahistoricalinproposingthisalternative,sinceinfactitembodies
assumptionsthatarenotonlyplausibleandrelativelywellconfirmed,soitappearstome,butalsodeeply
rootedinarichandlargelyforgottentraditionofrationalistpsychologyandlinguistics.Ihavetriedtocorrect
thisimbalanceinlaterpublications(Chomsky,1962,1964,1966seealsoMilleretal.,1960Katzand
Postal,1964Fodor,1965Lenneberg,1966).
Ithinkitwouldalsohavebeenvaluabletotrytosketchsomeofthereasonsandthereweremanythat
havemadetheviewIwascriticizingseemplausibleoveralongperiod,andalsotodiscussthereasonsfor
thedeclineofthealternativerationalistconceptionwhich,Iwassuggesting,shouldberehabilitated.Sucha
discussionwould,perhaps,havehelpedtoplacethespecificcritiqueofSkinnerinamoremeaningful
context.
ReferencesinthePreface
Chomsky,N.,"ExplanatoryModelsinLinguistics,"inLogic,MethodologyandPhilosophyofScience,ed.E.
Nagel,P.Suppes,andA.Tarski.StanfordCalif.:StanfordUniversityPress,1962.
,CurrentIssuesinLinguisticTheory.TheHague:MoutonandCo.,1964.
http://cogprints.org/1148/1/chomsky.htm

1/24

08/09/2016

BadNews:NoamChomsky

,CartesianLinguistics.NewYork:HarperandRow,Publishers,1966.
Fodor,J.,"CouldMeaningBean'rm',"JournalofVerbalLearningandVerbalBehavior,4(1965),7381.
Katz,J.andP.Postal,AnIntegratedTheoryofLinguisticDescription.Cambridge,Mass:M.I.T.Press,1964.
Lenneberg,E.,BiologicalBasesofLanguage.(Inpress.)
Miller,G.A.,E.Galanter,andK.H.Pribram,PlansandtheStructureofBehavior.NewYork:Holt,
Rhinehart,andWinston,Inc.,1960.
1959review:
I
Agreatmanylinguistsandphilosophersconcernedwithlanguagehaveexpressedthehopethattheirstudies
mightultimatelybeembeddedinaframeworkprovidedbybehavioristpsychology,andthatrefractoryareas
ofinvestigation,particularlythoseinwhichmeaningisinvolved,willinthiswaybeopeneduptofruitful
exploration.Sincethisvolumeisthefirstlargescaleattempttoincorporatethemajoraspectsoflinguistic
behaviorwithinabehavioristframework,itmeritsandwillundoubtedlyreceivecarefulattention.Skinneris
notedforhiscontributionstothestudyofanimalbehavior.Thebookunderreviewistheproductofstudyof
linguisticbehaviorextendingovermorethantwentyyears.Earlierversionsofithavebeenfairlywidely
circulated,andtherearequiteafewreferencesinthepsychologicalliteraturetoitsmajorideas.
Theproblemtowhichthisbookisaddressedisthatofgivinga"functionalanalysis"ofverbalbehavior.By
functionalanalysis,Skinnermeansidentificationofthevariablesthatcontrolthisbehaviorandspecification
ofhowtheyinteracttodetermineaparticularverbalresponse.Furthermore,thecontrollingvariablesareto
bedescribedcompletelyintermsofsuchnotionsasstimulus,reinforcement,deprivation,whichhavebeen
givenareasonablyclearmeaninginanimalexperimentation.Inotherwords,thegoalofthebookisto
provideawaytopredictandcontrolverbalbehaviorbyobservingandmanipulatingthephysical
environmentofthespeaker.
Skinnerfeelsthatrecentadvancesinthelaboratorystudyofanimalbehaviorpermitustoapproachthis
problemwithacertainoptimism,since"thebasicprocessesandrelationswhichgiveverbalbehaviorits
specialcharacteristicsarenowfairlywellunderstood...theresults[ofthisexperimentalwork]havebeen
surprisinglyfreeofspeciesrestrictions.Recentworkhasshownthatthemethodscanbeextendedtohuman
behaviorwithoutseriousmodification"(3).1
ItisimportanttoseeclearlyjustwhatitisinSkinner'sprogramandclaimsthatmakesthemappearsobold
andremarkable,Itisnotprimarilythefactthathehassetfunctionalanalysisashisproblem,orthathelimits
himselftostudyofobservables,i.e.,inputoutputrelations.Whatissosurprisingistheparticularlimitations
hehasimposedonthewayinwhichtheobservablesofbehavioraretobestudied,and,aboveall,the
particularlysimplenatureofthefunctionwhich,heclaims,describesthecausationofbehavior.Onewould
naturallyexpectthatpredictionofthebehaviorofacomplexorganism(ormachine)wouldrequire,in
additiontoinformationaboutexternalstimulation,knowledgeoftheinternalstructureoftheorganism,the
waysinwhichitprocessesinputinformationandorganizesitsownbehavior.Thesecharacteristicsofthe
organismareingeneralacomplicatedproductofinbornstructure,thegeneticallydeterminedcourseof
maturation,andpastexperience.Insofarasindependentneurophysiologicalevidenceisnotavailable,itis
obviousthatinferencesconcerningthestructureoftheorganismarebasedonobservationofbehaviorand
outsideevents.Nevertheless,one'sestimateoftherelativeimportanceofexternalfactorsandinternal
structureinthedeterminationofbehaviorwillhaveanimportanteffectonthedirectionofresearchon
linguistic(oranyother)behavior,andonthekindsofanalogiesfromanimalbehaviorstudiesthatwillbe
consideredrelevantorsuggestive.
Puttingitdifferently,anyonewhosetshimselftheproblemofanalyzingthecausationofbehaviorwill(inthe
absenceofindependentneurophysiologicalevidence)concernhimselfwiththeonlydataavailable,namely
therecordofinputstotheorganismandtheorganism'spresentresponse,andwilltrytodescribethefunction
specifyingtheresponseintermsofthehistoryofinputs.Thisisnothingmorethanthedefinitionofhis
http://cogprints.org/1148/1/chomsky.htm

2/24

08/09/2016

BadNews:NoamChomsky

problem.Therearenopossiblegroundsforargumenthere,ifoneacceptstheproblemaslegitimate,though
Skinnerhasoftenadvancedanddefendedthisdefinitionofaproblemasifitwereathesiswhichother
investigatorsreject.Thedifferencesthatarisebetweenthosewhoaffirmandthosewhodenytheimportance
ofthespecific"contributionoftheorganism"tolearningandperformanceconcerntheparticularcharacter
andcomplexityofthisfunction,andthekindsofobservationsandresearchnecessaryforarrivingataprecise
specificationofit.Ifthecontributionoftheorganismiscomplex,theonlyhopeofpredictingbehavioreven
inagrosswaywillbethroughaveryindirectprogramofresearchthatbeginsbystudyingthedetailed
characterofthebehavioritselfandtheparticularcapacitiesoftheorganisminvolved.
Skinner'sthesisisthatexternalfactorsconsistingofpresentstimulationandthehistoryofreinforcement(in
particular,thefrequency,arrangement,andwithholdingofreinforcingstimuli)areofoverwhelming
importance,andthatthegeneralprinciplesrevealedinlaboratorystudiesofthesephenomenaprovidethe
basisforunderstandingthecomplexitiesofverbalbehavior.Heconfidentlyandrepeatedlyvoiceshisclaim
tohavedemonstratedthatthecontributionofthespeakerisquitetrivialandelementary,andthatprecise
predictionofverbalbehaviorinvolvesonlyspecificationofthefewexternalfactorsthathehasisolated
experimentallywithlowerorganisms.
Carefulstudyofthisbook(andoftheresearchonwhichitdraws)reveals,however,thattheseastonishing
claimsarefarfromjustified.Itindicates,furthermore,thattheinsightsthathavebeenachievedinthe
laboratoriesofthereinforcementtheorist,thoughquitegenuine,canbeappliedtocomplexhumanbehavior
onlyinthemostgrossandsuperficialway,andthatspeculativeattemptstodiscusslinguisticbehaviorin
thesetermsaloneomitfromconsiderationfactorsoffundamentalimportancethatare,nodoubt,amenableto
scientificstudy,althoughtheirspecificcharactercannotatpresentbepreciselyformulated.SinceSkinner's
workisthemostextensiveattempttoaccommodatehumanbehaviorinvolvinghighermentalfacultieswithin
astrictbehavioristschemaofthetypethathasattractedmanylinguistsandphilosophers,aswellas
psychologists,adetaileddocumentationisofindependentinterest.Themagnitudeofthefailureofthis
attempttoaccountforverbalbehaviorservesasakindofmeasureoftheimportanceofthefactorsomitted
fromconsideration,andanindicationofhowlittleisreallyknownaboutthisremarkablycomplex
phenomenon.
TheforceofSkinner'sargumentliesintheenormouswealthandrangeofexamplesforwhichheproposesa
functionalanalysis.Theonlywaytoevaluatethesuccessofhisprogramandthecorrectnessofhisbasic
assumptionsaboutverbalbehavioristoreviewtheseexamplesindetailandtodeterminetheprecise
characteroftheconceptsintermsofwhichthefunctionalanalysisispresented.Section2ofthisreview
describestheexperimentalcontextwithrespecttowhichtheseconceptsareoriginallydefined.Sections3
and4dealwiththebasicconceptsstimulus,response,andreinforcement,Sections6to10withthenew
descriptivemachinerydevelopedspecificallyforthedescriptionofverbalbehavior.InSection5weconsider
thestatusofthefundamentalclaim,drawnfromthelaboratory,whichservesasthebasisfortheanalogic
guessesabouthumanbehaviorthathavebeenproposedbymanypsychologists.Thefinalsection(Section
11)willconsidersomewaysinwhichfurtherlinguisticworkmayplayapartinclarifyingsomeofthese
problems.
II
Althoughthisbookmakesnodirectreferencetoexperimentalwork,itcanbeunderstoodonlyintermsofthe
generalframeworkthatSkinnerhasdevelopedforthedescriptionofbehavior.Skinnerdividestheresponses
oftheanimalintotwomaincategories.Respondentsarepurelyreflexresponseselicitedbyparticularstimuli.
Operantsareemittedresponses,forwhichnoobviousstimuluscanbediscovered.Skinnerhasbeen
concernedprimarilywithoperantbehavior.Theexperimentalarrangementthatheintroducedconsists
basicallyofaboxwithabarattachedtoonewallinsuchawaythatwhenthebarispressed,afoodpelletis
droppedintoatray(andthebarpressisrecorded).Aratplacedintheboxwillsoonpressthebar,releasinga
pelletintothetray.Thisstateofaffairs,resultingfromthebarpress,increasesthestrengthofthebarpressing
operant.Thefoodpelletiscalledareinforcertheevent,areinforcingevent.Thestrengthofanoperantis
definedbySkinnerintermsoftherateofresponseduringextinction(i.e,afterthelastreinforcementand
beforereturntothepreconditioningrate).
Supposethatreleaseofthepelletisconditionalontheflashingofalight.Thentheratwillcometopressthe
baronlywhenthelightflashes.Thisiscalledstimulusdiscrimination.Theresponseiscalledadiscriminated
http://cogprints.org/1148/1/chomsky.htm

3/24

08/09/2016

BadNews:NoamChomsky

operantandthelightiscalledtheoccasionforitsemission:thisistobedistinguishedfromelicitationofa
responsebyastimulusinthecaseoftherespondent.2Supposethattheapparatusissoarrangedthatbar
pressingofonlyacertaincharacter(e.g.,duration)willreleasethepellet.Theratwillthencometopressthe
barintherequiredway.Thisprocessiscalledresponsedifferentiation.Bysuccessiveslightchangesinthe
conditionsunderwhichtheresponsewillbereinforced,itispossibletoshapetheresponseofaratora
pigeoninverysurprisingwaysinaveryshorttime,sothatrathercomplexbehaviorcanbeproducedbya
processofsuccessiveapproximation.
Astimuluscanbecomereinforcingbyrepeatedassociationwithanalreadyreinforcingstimulus.Sucha
stimulusiscalledasecondaryreinforcer.Likemanycontemporarybehaviorists,Skinnerconsidersmoney,
approval,andtheliketobesecondaryreinforcerswhichhavebecomereinforcingbecauseoftheir
associationwithfood,etc.3Secondaryreinforcerscanbegeneralizedbyassociatingthemwithavarietyof
differentprimaryreinforcers.
Anothervariablethatcanaffecttherateofthebarpressingoperantisdrive,whichSkinnerdefines
operationallyintermsofhoursofdeprivation.Hismajorscientificbook,BehaviorofOrganisms,isastudy
oftheeffectsoffooddeprivationandconditioningonthestrengthofthebarpressingresponseofhealthy
maturerats.ProbablySkinner'smostoriginalcontributiontoanimalbehaviorstudieshasbeenhis
investigationoftheeffectsofintermittentreinforcement,arrangedinvariousdifferentways,presentedin
BehaviorofOrganismsandextended(withpeckingofpigeonsastheoperantunderinvestigation)inthe
recentSchedulesofReinforcementbyFersterandSkinner(1957).ItisapparentlythesestudiesthatSkinner
hasinmindwhenhereferstotherecentadvancesinthestudyofanimalbehavior.4
Thenotionsstimulus,response,reinforcementarerelativelywelldefinedwithrespecttothebarpressing
experimentsandotherssimilarlyrestricted.Beforewecanextendthemtoreallifebehavior,however,certain
difficultiesmustbefaced.Wemustdecide,firstofall,whetheranyphysicaleventtowhichtheorganismis
capableofreactingistobecalledastimulusonagivenoccasion,oronlyonetowhichtheorganisminfact
reactsandcorrespondingly,wemustdecidewhetheranypartofbehavioristobecalledaresponse,oronly
oneconnectedwithstimuliinlawfulways.Questionsofthissortposesomethingofadilemmaforthe
experimentalpsychologist.Ifheacceptsthebroaddefinitions,characterizinganyphysicaleventimpinging
ontheorganismasastimulusandanypartoftheorganism'sbehaviorasaresponse,hemustconcludethat
behaviorhasnotbeendemonstratedtobelawful.Inthepresentstateofourknowledge,wemustattributean
overwhelminginfluenceonactualbehaviortoilldefinedfactorsofattention,set,volition,andcaprice.Ifwe
acceptthenarrowerdefinitions,thenbehaviorislawfulbydefinition(ifitconsistsofresponses)butthisfact
isoflimitedsignificance,sincemostofwhattheanimaldoeswillsimplynotbeconsideredbehavior.Hence,
thepsychologisteithermustadmitthatbehaviorisnotlawful(orthathecannotatpresentshowthatitis
notatalladamagingadmissionforadevelopingscience),ormustrestricthisattentiontothosehighly
limitedareasinwhichitislawful(e.g.,withadequatecontrols,barpressinginratslawfulnessofthe
observedbehaviorprovides,forSkinner,animplicitdefinitionofagoodexperiment).
Skinnerdoesnotconsistentlyadopteithercourse.Heutilizestheexperimentalresultsasevidenceforthe
scientificcharacterofhissystemofbehavior,andanalogicguesses(formulatedintermsofametaphoric
extensionofthetechnicalvocabularyofthelaboratory)asevidenceforitsscope.Thiscreatestheillusionof
arigorousscientifictheorywithaverybroadscope,althoughinfactthetermsusedinthedescriptionofreal
lifeandoflaboratorybehaviormaybemerehomonyms,withatmostavaguesimilarityofmeaning.To
substantiatethisevaluation,acriticalaccountofhisbookmustshowthatwithaliteralreading(wherethe
termsofthedescriptivesystemhavesomethinglikethetechnicalmeaningsgiveninSkinner'sdefinitions)
thebookcoversalmostnoaspectoflinguisticbehavior,andthatwithametaphoricreading,itisnomore
scientificthanthetraditionalapproachestothissubjectmatter,andrarelyasclearandcareful.5
III
ConsiderfirstSkinner'suseofthenotionsstimulusandresponse.InBehaviorofOrganisms(9)hecommits
himselftothenarrowdefinitionsfortheseterms.Apartoftheenvironmentandapartofbehaviorarecalled
stimulus(eliciting,discriminated,orreinforcing)andresponse,respectively,onlyiftheyarelawfullyrelated
thatis,ifthedynamiclawsrelatingthemshowsmoothandreproduciblecurves.Evidently,stimuliand
responses,sodefined,havenotbeenshowntofigureverywidelyinordinaryhumanbehavior.6Wecan,in
thefaceofpresentlyavailableevidence,continuetomaintainthelawfulnessoftherelationbetweenstimulus
http://cogprints.org/1148/1/chomsky.htm

4/24

08/09/2016

BadNews:NoamChomsky

andresponseonlybydeprivingthemoftheirobjectivecharacter.Atypicalexampleofstimuluscontrolfor
SkinnerwouldbetheresponsetoapieceofmusicwiththeutteranceMozartortoapaintingwiththe
responseDutch.Theseresponsesareassertedtobe"underthecontrolofextremelysubtleproperties"ofthe
physicalobjectorevent(108).SupposeinsteadofsayingDutchwehadsaidClasheswiththewallpaper,I
thoughtyoulikedabstractwork,Neversawitbefore,Tilted,Hangingtoolow,Beautiful,Hideous,Remember
ourcampingtriplastsummer?,orwhateverelsemightcomeintoourmindswhenlookingatapicture(in
Skinneriantranslation,whateverotherresponsesexistinsufficientstrength).Skinnercouldonlysaythat
eachoftheseresponsesisunderthecontrolofsomeotherstimuluspropertyofthephysicalobject.Ifwelook
ataredchairandsayred,theresponseisunderthecontrolofthestimulusrednessifwesaychair,itis
underthecontrolofthecollectionofproperties(forSkinner,theobject)chairness(110),andsimilarlyfor
anyotherresponse.Thisdeviceisassimpleasitisempty.Sincepropertiesarefreefortheasking(wehave
asmanyofthemaswehavenonsynonymousdescriptiveexpressionsinourlanguage,whateverthismeans
exactly),wecanaccountforawideclassofresponsesintermsofSkinnerianfunctionalanalysisby
identifyingthecontrollingstimuli.Butthewordstimulushaslostallobjectivityinthisusage.Stimuliareno
longerpartoftheoutsidephysicalworldtheyaredrivenbackintotheorganism.Weidentifythestimulus
whenweheartheresponse.Itisclearfromsuchexamples,whichabound,thatthetalkofstimuluscontrol
simplydisguisesacompleteretreattomentalisticpsychology.Wecannotpredictverbalbehaviorintermsof
thestimuliinthespeaker'senvironment,sincewedonotknowwhatthecurrentstimuliareuntilheresponds.
Furthermore,sincewecannotcontrolthepropertyofaphysicalobjecttowhichanindividualwillrespond,
exceptinhighlyartificialcases,Skinner'sclaimthathissystem,asopposedtothetraditionalone,permitsthe
practicalcontrolofverbalbehavior7isquitefalse.
Otherexamplesofstimuluscontrolmerelyaddtothegeneralmystification.Thus,apropernounisheldtobe
aresponse"underthecontrolofaspecificpersonorthing"(ascontrollingstimulus,113).Ihaveoftenused
thewordsEisenhowerandMoscow,whichIpresumearepropernounsifanythingis,buthaveneverbeen
stimulatedbythecorrespondingobjects.Howcanthisfactbemadecompatiblewiththisdefinition?Suppose
thatIusethenameofafriendwhoisnotpresent.Isthisaninstanceofapropernoununderthecontrolofthe
friendasstimulus?Elsewhereitisassertedthatastimuluscontrolsaresponseinthesensethatpresenceof
thestimulusincreasestheprobabilityoftheresponse.Butitisobviouslyuntruethattheprobabilitythata
speakerwillproduceafullnameisincreasedwhenitsbearerfacesthespeaker.Furthermore,howcanone's
ownnamebeapropernouninthissense?
Amultitudeofsimilarquestionsariseimmediately.Itappearsthatthewordcontrolhereismerelya
misleadingparaphraseforthetraditionaldenoteorrefer.Theassertion(115)thatsofarasthespeakeris
concerned,therelationofreferenceis"simplytheprobabilitythatthespeakerwillemitaresponseofagiven
forminthepresenceofastimulushavingspecifiedproperties"issurelyincorrectifwetakethewords
presence,stimulus,andprobabilityintheirliteralsense.Thattheyarenotintendedtobetakenliterallyis
indicatedbymanyexamples,aswhenaresponseissaidtobe"controlled"byasituationorstateofaffairsas
"stimulus."Thus,theexpressionaneedleinahaystack"maybecontrolledasaunitbyaparticulartypeof
situation"(116)thewordsinasinglepartofspeech,e.g.,alladjectives,areunderthecontrolofasingleset
ofsubtlepropertiesofstimuli(121)"thesentenceTheboyrunsastoreisunderthecontrolofanextremely
complexstimulussituation"(335)"Heisnotatallwellmayfunctionasastandardresponseunderthe
controlofastateofaffairswhichmightalsocontrolHeisailing"(325)whenanenvoyobserveseventsina
foreigncountryandreportsuponhisreturn,hisreportisunder"remotestimuluscontrol"(416)thestatement
Thisiswarmaybearesponsetoa"confusinginternationalsituation"(441)thesuffixediscontrolledby
that"subtlepropertyofstimuliwhichwespeakofasactioninthepast"(121)justasthesinTheboyrunsis
underthecontrolofsuchspecificfeaturesofthesituationasits"currency"(332).Nocharacterizationofthe
notionstimuluscontrolthatisremotelyrelatedtothebarpressingexperiment(orthatpreservesthefaintest
objectivity)canbemadetocoverasetofexampleslikethese,inwhich,forexample,thecontrollingstimulus
neednotevenimpingeontherespondingorganism.
ConsidernowSkinner'suseofthenotionresponse.Theproblemofidentifyingunitsinverbalbehaviorhas
ofcoursebeenaprimaryconcernoflinguists,anditseemsverylikelythatexperimentalpsychologists
shouldbeabletoprovidemuchneededassistanceinclearingupthemanyremainingdifficultiesin
systematicidentification.Skinnerrecognizes(20)thefundamentalcharacteroftheproblemofidentification
ofaunitofverbalbehavior,butissatisfiedwithananswersovagueandsubjectivethatitdoesnotreally
contributetoitssolution.Theunitofverbalbehaviortheverbaloperantisdefinedasaclassofresponses
ofidentifiableformfunctionallyrelatedtooneormorecontrollingvariables.Nomethodissuggestedfor
http://cogprints.org/1148/1/chomsky.htm

5/24

08/09/2016

BadNews:NoamChomsky

determininginaparticularinstancewhatarethecontrollingvariables,howmanysuchunitshaveoccurred,
orwheretheirboundariesareinthetotalresponse.Norisanyattemptmadetospecifyhowmuchorwhat
kindofsimilarityinformorcontrolisrequiredfortwophysicaleventstobeconsideredinstancesofthe
sameoperant.Inshort,noanswersaresuggestedforthemostelementaryquestionsthatmustbeaskedof
anyoneproposingamethodfordescriptionofbehavior.Skinneriscontentwithwhathecallsan
extrapolationoftheconceptofoperantdevelopedinthelaboratorytotheverbalfield.Inthetypical
Skinnerianexperiment,theproblemofidentifyingtheunitofbehaviorisnottoocrucial.Itisdefined,byfiat,
asarecordedpeckorbarpress,andsystematicvariationsintherateofthisoperantanditsresistanceto
extinctionarestudiedasafunctionofdeprivationandschedulingofreinforcement(pellets).Theoperantis
thusdefinedwithrespecttoaparticularexperimentalprocedure.Thisisperfectlyreasonableandhasledto
manyinterestingresults.Itis,however,completelymeaninglesstospeakofextrapolatingthisconceptof
operanttoordinaryverbalbehavior.Such"extrapolation"leavesuswithnowayofjustifyingoneoranother
decisionabouttheunitsinthe"verbalrepertoire."
Skinnerspecifies"responsestrength"asthebasicdatum,thebasicdependentvariableinhisfunctional
analysis.Inthebarpressingexperiment,responsestrengthisdefinedintermsofrateofemissionduring
extinction.Skinnerhasargued8thatthisis"theonlydatumthatvariessignificantlyandintheexpected
directionunderconditionswhicharerelevanttothe'learningprocess.'"Inthebookunderreview,response
strengthisdefinedas"probabilityofemission"(22).Thisdefinitionprovidesacomfortingimpressionof
objectivity,which,however,isquicklydispelledwhenwelookintothemattermoreclosely.Theterm
probabilityhassomeratherobscuremeaningforSkinnerinthisbook.9Wearetold,ontheonehand,that
"ourevidenceforthecontributionofeachvariable[toresponsestrength]isbasedonobservationof
frequenciesalone"(28).Atthesametime,itappearsthatfrequencyisaverymisleadingmeasureofstrength,
since,forexample,thefrequencyofaresponsemaybe"primarilyattributabletothefrequencyofoccurrence
ofcontrollingvariables"(27).Itisnotclearhowthefrequencyofaresponsecanbeattributabletoanything
BUTthefrequencyofoccurrenceofitscontrollingvariablesifweacceptSkinner'sviewthatthebehavior
occurringinagivensituationis"fullydetermined"bytherelevantcontrollingvariables(175,228).
Furthermore,althoughtheevidenceforthecontributionofeachvariabletoresponsestrengthisbasedon
observationoffrequenciesalone,itturnsoutthat"webasethenotionofstrengthuponseveralkindsof
evidence"(22),inparticular(2228):emissionoftheresponse(particularlyinunusualcircumstances),
energylevel(stress),pitchlevel,speedanddelayofemission,sizeoflettersetc.inwriting,immediate
repetition,andafinalfactor,relevantbutmisleadingoverallfrequency.
Ofcourse,Skinnerrecognizesthatthesemeasuresdonotcovary,because(amongotherreasons)pitch,
stress,quantity,andreduplicationmayhaveinternallinguisticfunctions.10However,hedoesnotholdthese
conflictstobeveryimportant,sincetheproposedfactorsindicativeofstrengthare"fullyunderstoodby
everyone"intheculture(27).Forexample,"ifweareshownaprizedworkofartandexclaimBeautiful!,the
speedandenergyoftheresponsewillnotbelostontheowner."Itdoesnotappeartotallyobviousthatinthis
casethewaytoimpresstheowneristoshriekBeautifulinaloud,highpitchedvoice,repeatedly,andwithno
delay(highresponsestrength).Itmaybeequallyeffectivetolookatthepicturesilently(longdelay)andthen
tomurmurBeautifulinasoft,lowpitchedvoice(bydefinition,verylowresponsestrength).
Itisnotunfair,Ibelieve,toconcludefromSkinner'sdiscussionofresponsestrength,thebasicdatumin
functionalanalysis,thathisextrapolationofthenotionofprobabilitycanbestbeinterpretedas,ineffect,
nothingmorethanadecisiontousethewordprobability,withitsfavorableconnotationsofobjectivity,asa
covertermtoparaphrasesuchlowstatuswordsasinterest,intention,belief,andthelike.Thisinterpretation
isfullyjustifiedbythewayinwhichSkinnerusesthetermsprobabilityandstrength.Tocitejustone
example,Skinnerdefinestheprocessofconfirminganassertioninscienceasoneof"generatingadditional
variablestoincreaseitsprobability"(425),andmoregenerally,itsstrength(42529).Ifwetakethis
suggestionquiteliterally,thedegreeofconfirmationofascientificassertioncanbemeasuredasasimple
functionoftheloudness,pitch,andfrequencywithwhichitisproclaimed,andageneralprocedurefor
increasingitsdegreeofconfirmationwouldbe,forinstance,totrainmachinegunsonlargecrowdsofpeople
whohavebeeninstructedtoshoutit.AbetterindicationofwhatSkinnerprobablyhasinmindhereisgiven
byhisdescriptionofhowthetheoryofevolution,asanexample,isconfirmed.This"singlesetofverbal
responses...ismademoreplausibleisstrengthenedbyseveraltypesofconstructionbaseduponverbal
responsesingeology,paleontology,genetics,andsoon"(427).Wearenodoubttointerprettheterms
strengthandprobabilityinthiscontextasparaphrasesofmorefamiliarlocutionssuchas"justifiedbelief"or
"warrantedassertability,"orsomethingofthesort.Similarlatitudeofinterpretationispresumablyexpected
http://cogprints.org/1148/1/chomsky.htm

6/24

08/09/2016

BadNews:NoamChomsky

whenwereadthat"frequencyofeffectiveactionaccountsinturnforwhatwemaycallthelistener's'belief'"
(88)orthat"ourbeliefinwhatsomeonetellsusissimilarlyafunctionof,oridenticalwith,ourtendencyto
actupontheverbalstimuliwhichheprovides"(160).11
Ithinkitisevident,then,thatSkinner'suseofthetermsstimulus,control,response,andstrengthjustifythe
generalconclusionstatedinthelastparagraphofSection2.Thewayinwhichthesetermsarebroughttobear
ontheactualdataindicatesthatwemustinterpretthemasmereparaphrasesforthepopularvocabulary
commonlyusedtodescribebehaviorandashavingnoparticularconnectionwiththehomonymous
expressionsusedinthedescriptionoflaboratoryexperiments.Naturally,thisterminologicalrevisionaddsno
objectivitytothefamiliarmentalisticmodeofdescription.
IV
Theotherfundamentalnotionborrowedfromthedescriptionofbarpressingexperimentsisreinforcement.It
raisesproblemswhicharesimilar,andevenmoreserious.InBehaviorofOrganisms,"theoperationof
reinforcementisdefinedasthepresentationofacertainkindofstimulusinatemporalrelationwitheithera
stimulusorresponse.Areinforcingstimulusisdefinedassuchbyitspowertoproducetheresultingchange
[instrength].Thereisnocircularityaboutthis:somestimuliarefoundtoproducethechange,othersnot,and
theyareclassifiedasreinforcingandnonreinforcingaccordingly"(62).Thisisaperfectlyappropriate
definition12forthestudyofschedulesofreinforcement.Itisperfectlyuseless,however,inthediscussionof
reallifebehavior,unlesswecansomehowcharacterizethestimuliwhicharereinforcing(andthesituations
andconditionsunderwhichtheyarereinforcing).Considerfirstofallthestatusofthebasicprinciplethat
Skinnercallsthe"lawofconditioning"(lawofeffect).Itreads:"iftheoccurrenceofanoperantisfollowed
bypresenceofareinforcingstimulus,thestrengthisincreased"(BehaviorofOrganisms,21).As
reinforcementwasdefined,thislawbecomesatautology.13ForSkinner,learningisjustchangeinresponse
strength.14Althoughthestatementthatpresenceofreinforcementisasufficientconditionforlearningand
maintenanceofbehaviorisvacuous,theclaimthatitisanecessaryconditionmayhavesomecontent,
dependingonhowtheclassofreinforcers(andappropriatesituations)ischaracterized.Skinnerdoesmakeit
veryclearthatinhisviewreinforcementisanecessaryconditionforlanguagelearningandforthecontinued
availabilityoflinguisticresponsesintheadult.15However,theloosenessofthetermreinforcementas
Skinnerusesitinthebookunderreviewmakesitentirelypointlesstoinquireintothetruthorfalsityofthis
claim.ExaminingtheinstancesofwhatSkinnercallsreinforcement,wefindthatnoteventherequirement
thatareinforcerbeanidentifiablestimulusistakenseriously.Infact,thetermisusedinsuchawaythatthe
assertionthatreinforcementisnecessaryforlearningandcontinuedavailabilityofbehaviorislikewise
empty.
Toshowthis,weconsidersomeexamplesofreinforcement.Firstofall,wefindaheavyappealtoautomatic
selfreinforcement,Thus,"amantalkstohimself...becauseofthereinforcementhereceives"(163)"the
childisreinforcedautomaticallywhenheduplicatesthesoundsofairplanes,streetcars..."(164)"theyoung
childaloneinthenurserymayautomaticallyreinforcehisownexploratoryverbalbehaviorwhenhe
producessoundswhichhehasheardinthespeechofothers"(58)"thespeakerwhoisalsoanaccomplished
listener'knowswhenhehascorrectlyechoedaresponse'andisreinforcedthereby"(68)thinkingis
"behavingwhichautomaticallyaffectsthebehaverandisreinforcingbecauseitdoesso"(438cuttingone's
fingershouldthusbereinforcing,andanexampleofthinking)"theverbalfantasy,whetherovertorcovert,
isautomaticallyreinforcingtothespeakeraslistener.Justasthemusicianplaysorcomposeswhatheis
reinforcedbyhearing,orastheartistpaintswhatreinforceshimvisually,sothespeakerengagedinverbal
fantasysayswhatheisreinforcedbyhearingorwriteswhatheisreinforcedbyreading"(439)similarly,
careinproblemsolving,andrationalization,areautomaticallyselfreinforcing(44243).Wecanalso
reinforcesomeonebyemittingverbalbehaviorassuch(sincethisrulesoutaclassofaversivestimulations,
167),bynotemittingverbalbehavior(keepingsilentandpayingattention,199),orbyactingappropriately
onsomefutureoccasion(152:"thestrengthof[thespeaker's]behaviorisdeterminedmainlybythebehavior
whichthelistenerwillexhibitwithrespecttoagivenstateofaffairs"thisSkinnerconsidersthegeneralcase
of"communication"or"lettingthelistenerknow").Inmostsuchcases,ofcourse,thespeakerisnotpresent
atthetimewhenthereinforcementtakesplace,aswhen"theartist...isreinforcedbytheeffectshisworks
haveupon...others"(224),orwhenthewriterisreinforcedbythefactthathis"verbalbehaviormayreach
overcenturiesortothousandsoflistenersorreadersatthesametime.Thewritermaynotbereinforcedoften
orimmediately,buthisnetreinforcementmaybegreat"(206thisaccountsforthegreat"strength"ofhis
behavior).Anindividualmayalsofinditreinforcingtoinjuresomeonebycriticismorbybringingbadnews,
http://cogprints.org/1148/1/chomsky.htm

7/24

08/09/2016

BadNews:NoamChomsky

ortopublishanexperimentalresultwhichupsetsthetheoryofarival(154),todescribecircumstanceswhich
wouldbereinforcingiftheyweretooccur(165),toavoidrepetition(222),to"hear"hisownnamethoughin
factitwasnotmentionedortohearnonexistentwordsinhischild'sbabbling(259),toclarifyorotherwise
intensifytheeffectofastimuluswhichservesanimportantdiscriminativefunction(416),andsoon.
Fromthissample,itcanbeseenthatthenotionofreinforcementhastotallylostwhateverobjectivemeaning
itmayeverhavehad.Runningthroughtheseexamples,weseethatapersoncanbereinforcedthoughhe
emitsnoresponseatall,andthatthereinforcingstimulusneednotimpingeonthereinforcedpersonorneed
notevenexist(itissufficientthatitbeimaginedorhopedfor).Whenwereadthatapersonplayswhatmusic
helikes(165),sayswhathelikes(165),thinkswhathelikes(43839),readswhatbookshelikes(163),etc.,
BECAUSEhefindsitreinforcingtodoso,orthatwewritebooksorinformothersoffactsBECAUSEwe
arereinforcedbywhatwehopewillbetheultimatebehaviorofreaderorlistener,wecanonlyconcludethat
thetermreinforcementhasapurelyritualfunction.Thephrase"XisreinforcedbyY(stimulus,stateof
affairs,event,etc.)"isbeingusedasacovertermfor"XwantsY,""XlikesY,""XwishesthatYwerethe
case,"etc.Invokingthetermreinforcementhasnoexplanatoryforce,andanyideathatthisparaphrase
introducesanynewclarityorobjectivityintothedescriptionofwishing,liking,etc.,isaseriousdelusion.
Theonlyeffectistoobscuretheimportantdifferencesamongthenotionsbeingparaphrased.Oncewe
recognizethelatitudewithwhichthetermreinforcementisbeingused,manyratherstartlingcommentslose
theirinitialeffectforinstance,thatthebehaviorofthecreativeartistis"controlledentirelybythe
contingenciesofreinforcement"(150).Whathasbeenhopedforfromthepsychologistissomeindication
howthecasualandinformaldescriptionofeverydaybehaviorinthepopularvocabularycanbeexplainedor
clarifiedintermsofthenotionsdevelopedincarefulexperimentandobservation,orperhapsreplacedin
termsofabetterscheme.Amereterminologicalrevision,inwhichatermborrowedfromthelaboratoryis
usedwiththefullvaguenessoftheordinaryvocabulary,isofnoconceivableinterest.
ItseemsthatSkinner'sclaimthatallverbalbehaviorisacquiredandmaintainedin"strength"through
reinforcementisquiteempty,becausehisnotionofreinforcementhasnoclearcontent,functioningonlyasa
covertermforanyfactor,detectableornot,relatedtoacquisitionormaintenanceofverbalbehavior.16
Skinner'suseofthetermconditioningsuffersfromasimilardifficulty.Pavlovianandoperantconditioning
areprocessesaboutwhichpsychologistshavedevelopedrealunderstanding.Instructionofhumanbeingsis
not.Theclaimthatinstructionandimpartingofinformationaresimplymattersofconditioning(35766)is
pointless.Theclaimistrue,ifweextendthetermconditioningtocovertheseprocesses,butweknowno
moreaboutthemafterhavingrevisedthisterminsuchawayastodepriveitofitsrelativelyclearand
objectivecharacter.Itis,asfarasweknow,quitefalse,ifweuseconditioninginitsliteralsense.Similarly,
whenwesaythat"itisthefunctionofpredicationtofacilitatethetransferofresponsefromonetermto
anotherorfromoneobjecttoanother"(361),wehavesaidnothingofanysignificance.Inwhatsenseisthis
trueofthepredicationWhalesaremammals?Or,totakeSkinner'sexample,whatpointisthereinsayingthat
theeffectofThetelephoneisoutoforderonthelisteneristobringbehaviorformerlycontrolledbythe
stimulusoutoforderundercontrolofthestimulustelephone(orthetelephoneitself)byaprocessofsimple
conditioning(362)?Whatlawsofconditioningholdinthiscase?Furthermore,whatbehavioriscontrolled
bythestimulusoutoforder,intheabstract?Dependingontheobjectofwhichthisispredicated,thepresent
stateofmotivationofthelistener,etc.,thebehaviormayvaryfromragetopleasure,fromfixingtheobjectto
throwingitout,fromsimplynotusingittotryingtouseitinthenormalway(e.g.,toseeifitisreallyoutof
order),andsoon.Tospeakof"conditioning"or"bringingpreviouslyavailablebehaviorundercontrolofa
newstimulus"insuchacaseisjustakindofplayactingatscience(cf.also43n).
V
Theclaimthatcarefularrangementofcontingenciesofreinforcementbytheverbalcommunityisanecessary
conditionforlanguagelearninghasappeared,inoneformoranother,inmanyplaces.17Sinceitisbasednot
onactualobservation,butonanalogiestolaboratorystudyoflowerorganisms,itisimportanttodetermine
thestatusoftheunderlyingassertionwithinexperimentalpsychologyproper.Themostcommon
characterizationofreinforcement(onewhichSkinnerexplicitlyrejects,incidentally)isintermsofdrive
reduction.Thischaracterizationcanbegivensubstancebydefiningdrivesinsomewayindependentlyof
whatinfactislearned.Ifadriveispostulatedonthebasisofthefactthatlearningtakesplace,theclaimthat
reinforcementisnecessaryforlearningwillagainbecomeasemptyasitisintheSkinnerianframework.
Thereisanextensiveliteratureonthequestionofwhethertherecanbelearningwithoutdrivereduction
(latentlearning).The"classical"experimentofBlodgettindicatedthatratswhohadexploredamazewithout
http://cogprints.org/1148/1/chomsky.htm

8/24

08/09/2016

BadNews:NoamChomsky

rewardshowedamarkeddropinnumberoferrors(ascomparedtoacontrolgroupwhichhadnotexplored
themaze)uponintroductionofafoodreward,indicatingthattherathadlearnedthestructureofthemaze
withoutreductionofthehungerdrive.Drivereductiontheoristscounteredwithanexploratorydrivewhich
wasreducedduringtheprerewardlearning,andclaimedthataslightdecrementinerrorscouldbenoted
beforefoodreward.Awidevarietyofexperiments,withsomewhatconflictingresults,havebeencarriedout
withasimilardesign.18Fewinvestigatorsstilldoubttheexistenceofthephenomenon,E.R.Hilgard,inhis
generalreviewoflearningtheory,19concludesthat"thereisnolongeranydoubtbutthat,underappropriate
circumstances,latentlearningisdemonstrable."
Morerecentworkhasshownthatnoveltyandvarietyofstimulusaresufficienttoarousecuriosityintherat
andtomotivateittoexplore(visually),andinfact,tolearn(sinceonapresentationoftwostimuli,one
novel,onerepeated,theratwillattendtothenovelone),20thatratswilllearntochoosethearmofasingle
choicemazethatleadstoacomplexmaze,runningthroughthisbeingtheironly"reward"21thatmonkeys
canlearnobjectdiscriminationsandmaintaintheirperformanceatahighlevelofefficiencywithvisual
exploration(lookingoutofawindowfor30seconds)astheonlyreward22and,perhapsmoststrikinglyof
all,thatmonkeysandapeswillsolverathercomplexmanipulationproblemsthataresimplyplacedintheir
cages,andwillsolvediscriminationproblemswithonlyexplorationandmanipulationasincentives.23In
thesecases,solvingtheproblemisapparentlyitsown"reward."Resultsofthiskindcanbehandledby
reinforcementtheoristsonlyiftheyarewillingtosetupcuriosity,exploration,andmanipulationdrives,orto
speculatesomehowaboutacquireddrives24forwhichthereisnoevidenceoutsideofthefactthatlearning
takesplaceinthesecases.
Thereisavarietyofotherkindsofevidencethathasbeenofferedtochallengetheviewthatdrivereduction
isnecessaryforlearning.Resultsonsensorysensoryconditioninghavebeeninterpretedasdemonstrating
learningwithoutdrivereduction.25Oldshasreportedreinforcementbydirectstimulationofthebrain,from
whichheconcludesthatrewardneednotsatisfyaphysiologicalneedorwithdrawadrivestimulus.26The
phenomenonofimprinting,longobservedbyzoologists,isofparticularinterestinthisconnection.Someof
themostcomplexpatternsofbehaviorofbirds,inparticular,aredirectedtowardsobjectsandanimalsofthe
typetowhichtheyhavebeenexposedatcertaincriticalearlyperiodsoflife.27Imprintingisthemost
strikingevidencefortheinnatedispositionoftheanimaltolearninacertaindirectionandtoreact
appropriatelytopatternsandobjectsofcertainrestrictedtypes,oftenonlylongaftertheoriginallearninghas
takenplace.Itis,consequently,unrewardedlearning,thoughtheresultingpatternsofbehaviormaybe
refinedthroughreinforcement.Acquisitionofthetypicalsongsofsongbirdsis,insomecases,atypeof
imprinting.Thorpereportsstudiesthatshow"thatsomecharacteristicsofthenormalsonghavebeenlearned
intheearliestyouth,beforethebirditselfisabletoproduceanykindoffullsong."28Thephenomenonof
imprintinghasrecentlybeeninvestigatedunderlaboratoryconditionsandcontrolswithpositiveresults.29
Phenomenaofthisgeneraltypearecertainlyfamiliarfromeverydayexperience.Werecognizepeopleand
placestowhichwehavegivennoparticularattention.Wecanlookupsomethinginabookandlearnit
perfectlywellwithnoothermotivethantoconfutereinforcementtheory,oroutofboredom,oridlecuriosity.
Everyoneengagedinresearchmusthavehadtheexperienceofworkingwithfeverishandprolonged
intensitytowriteapaperwhichnooneelsewillreadortosolveaproblemwhichnooneelsethinks
importantandwhichwillbringnoconceivablerewardwhichmayonlyconfirmageneralopinionthatthe
researcheriswastinghistimeonirrelevancies.Thefactthatratsandmonkeysdolikewiseisinterestingand
importanttoshowincarefulexperiment.Infact,studiesofbehaviorofthetypementionedabovehavean
independentandpositivesignificancethatfaroutweighstheirincidentalimportanceinbringingintoquestion
theclaimthatlearningisimpossiblewithoutdrivereduction.Itisnotatallunlikelythatinsightsarisingfrom
animalbehaviorstudieswiththisbroadenedscopemayhavethekindofrelevancetosuchcomplexactivities
asverbalbehaviorthatreinforcementtheoryhas,sofar,failedtoexhibit.Inanyevent,inthelightof
presentlyavailableevidence,itisdifficulttoseehowanyonecanbewillingtoclaimthatreinforcementis
necessaryforlearning,ifreinforcementistakenseriouslyassomethingidentifiableindependentlyofthe
resultingchangeinbehavior.
Similarly,itseemsquitebeyondquestionthatchildrenacquireagooddealoftheirverbalandnonverbal
behaviorbycasualobservationandimitationofadultsandotherchildren.30Itissimplynottruethatchildren
canlearnlanguageonlythrough"meticulouscare"onthepartofadultswhoshapetheirverbalrepertoire
throughcarefuldifferentialreinforcement,thoughitmaybethatsuchcareisoftenthecustominacademic
families.Itisacommonobservationthatayoungchildofimmigrantparentsmaylearnasecondlanguagein
http://cogprints.org/1148/1/chomsky.htm

9/24

08/09/2016

BadNews:NoamChomsky

thestreets,fromotherchildren,withamazingrapidity,andthathisspeechmaybecompletelyfluentand
correcttothelastallophone,whilethesubtletiesthatbecomesecondnaturetothechildmayeludehis
parentsdespitehighmotivationandcontinuedpractice.Achildmaypickupalargepartofhisvocabulary
and"feel"forsentencestructurefromtelevision,fromreading,fromlisteningtoadults,etc.Evenavery
youngchildwhohasnotyetacquiredaminimalrepertoirefromwhichtoformnewutterancesmayimitatea
wordquitewellonanearlytry,withnoattemptonthepartofhisparentstoteachittohim.Itisalso
perfectlyobviousthat,atalaterstage,achildwillbeabletoconstructandunderstandutteranceswhichare
quitenew,andare,atthesametime,acceptablesentencesinhislanguage.Everytimeanadultreadsa
newspaper,heundoubtedlycomesuponcountlessnewsentenceswhicharenotatallsimilar,inasimple,
physicalsense,toanythathehasheardbefore,andwhichhewillrecognizeassentencesandunderstandhe
willalsobeabletodetectslightdistortionsormisprints.Talkof"stimulusgeneralization"insuchacase
simplyperpetuatesthemysteryunderanewtitle.Theseabilitiesindicatethattheremustbefundamental
processesatworkquiteindependentlyof"feedback"fromtheenvironment.Ihavebeenabletofindno
supportwhatsoeverforthedoctrineofSkinnerandothersthatslowandcarefulshapingofverbalbehavior
throughdifferentialreinforcementisanabsolutenecessity.Ifreinforcementtheoryreallyrequiresthe
assumptionthattherebesuchmeticulouscare,itseemsbesttoregardthissimplyasareductioadabsurdum
argumentagainstthisapproach.Itisalsonoteasytofindanybasis(or,forthatmatter,toattachverymuch
content)totheclaimthatreinforcingcontingenciessetupbytheverbalcommunityarethesinglefactor
responsibleformaintainingthestrengthofverbalbehavior.Thesourcesofthe"strength"ofthisbehaviorare
almostatotalmysteryatpresent.Reinforcementundoubtedlyplaysasignificantrole,butsodoavarietyof
motivationalfactorsaboutwhichnothingseriousisknowninthecaseofhumanbeings.
Asfarasacquisitionoflanguageisconcerned,itseemsclearthatreinforcement,casualobservation,and
naturalinquisitiveness(coupledwithastrongtendencytoimitate)areimportantfactors,asistheremarkable
capacityofthechildtogeneralize,hypothesize,and"processinformation"inavarietyofveryspecialand
apparentlyhighlycomplexwayswhichwecannotyetdescribeorbegintounderstand,andwhichmaybe
largelyinnate,ormaydevelopthroughsomesortoflearningorthroughmaturationofthenervoussystem.
Themannerinwhichsuchfactorsoperateandinteractinlanguageacquisitioniscompletelyunknown.Itis
clearthatwhatisnecessaryinsuchacaseisresearch,notdogmaticandperfectlyarbitraryclaims,basedon
analogiestothatsmallpartoftheexperimentalliteratureinwhichonehappenstobeinterested.
Thepointlessnessoftheseclaimsbecomesclearwhenweconsiderthewellknowndifficultiesin
determiningtowhatextentinbornstructure,maturation,andlearningareresponsiblefortheparticularform
ofaskilledorcomplexperformance.31Totakejustoneexample,32thegapingresponseofanestlingthrush
isatfirstreleasedbyjarringofthenest,and,atalaterstage,byamovingobjectofspecificsize,shape,and
positionrelativetothenestling.Atthislaterstagetheresponseisdirectedtowardthepartofthestimulus
objectcorrespondingtotheparent'shead,andcharacterizedbyacomplexconfigurationofstimulithatcanbe
preciselydescribed.Knowingjustthis,itwouldbepossibletoconstructaspeculative,learningtheoretic
accountofhowthissequenceofbehaviorpatternsmighthavedevelopedthroughaprocessofdifferential
reinforcement,anditwouldnodoubtbepossibletotrainratstodosomethingsimilar.However,there
appearstobegoodevidencethattheseresponsestofairlycomplex"signstimuli"aregeneticallydetermined
andmaturewithoutlearning.Clearly,thepossibilitycannotbediscounted.Considernowthecomparable
caseofachildimitatingnewwords.Atanearlystagewemayfindrathergrosscorrespondences.Atalater
stage,wefindthatrepetitionisofcoursefarfromexact(i.e.,itisnotmimicry,afactwhichitselfis
interesting),butthatitreproducesthehighlycomplexconfigurationofsoundfeaturesthatconstitutethe
phonologicalstructureofthelanguageinquestion.Again,wecanproposeaspeculativeaccountofhowthis
resultmighthavebeenobtainedthroughelaboratearrangementofreinforcingcontingencies.Heretoo,
however,itispossiblethatabilitytoselectoutofthecomplexauditoryinputthosefeaturesthatare
phonologicallyrelevantmaydeveloplargelyindependentlyofreinforcement,throughgeneticallydetermined
maturation.Totheextentthatthisistrue,anaccountofthedevelopmentandcausationofbehaviorthatfails
toconsiderthestructureoftheorganismwillprovidenounderstandingoftherealprocessesinvolved.
Itisoftenarguedthatexperience,ratherthaninnatecapacitytohandleinformationincertainspecificways,
mustbethefactorofoverwhelmingdominanceindeterminingthespecificcharacteroflanguageacquisition,
sinceachildspeaksthelanguageofthegroupinwhichhelives.Butthisisasuperficialargument.Aslong
aswearespeculating,wemayconsiderthepossibilitythatthebrainhasevolvedtothepointwhere,givenan
inputofobservedChinesesentences,itproduces(byaninductionofapparentlyfantasticcomplexityand
suddenness)therulesofChinesegrammar,andgivenaninputofobservedEnglishsentences,itproduces
http://cogprints.org/1148/1/chomsky.htm

10/24

08/09/2016

BadNews:NoamChomsky

(by,perhaps,exactlythesameprocessofinduction)therulesofEnglishgrammarorthatgivenanobserved
applicationofatermtocertaininstances,itautomaticallypredictstheextensiontoaclassofcomplexly
relatedinstances.Ifclearlyrecognizedassuch,thisspeculationisneitherunreasonablenorfantasticnor,for
thatmatter,isitbeyondtheboundsofpossiblestudy.Thereisofcoursenoknownneuralstructurecapableof
performingthistaskinthespecificwaysthatobservationoftheresultingbehaviormightleadustopostulate
butforthatmatter,thestructurescapableofaccountingforeventhesimplestkindsoflearninghavesimilarly
defieddetection.33Summarizingthisbriefdiscussion,itseemsthatthereisneitherempiricalevidencenor
anyknownargumenttosupportanyspecificclaimabouttherelativeimportanceof"feedback"fromthe
environmentandthe"independentcontributionoftheorganism"intheprocessoflanguageacquisition.
VI
WenowturntothesystemthatSkinnerdevelopsspecificallyforthedescriptionofverbalbehavior.Since
thissystemisbasedonthenotionsstimulus,response,andreinforcement,wecanconcludefromthe
precedingsectionsthatitwillbevagueandarbitrary.ForreasonsnotedinSection1,however,Ithinkitis
importanttoseeindetailhowfarfromthemarkanyanalysisphrasedsolelyinthesetermsmustbeandhow
completelythissystemfailstoaccountforthefactsofverbalbehavior.Considerfirstthetermverbal
behavioritself.Thisisdefinedas"behaviorreinforcedthroughthemediationofotherpersons"(2).The
definitionisclearlymuchtoobroad.Itwouldincludeasverbalbehavior,forexample,aratpressingthebar
inaSkinnerbox,achildbrushinghisteeth,aboxerretreatingbeforeanopponent,andamechanicrepairing
anautomobile.Exactlyhowmuchofordinarylinguisticbehaviorisverbalinthissense,however,is
somethingofaquestion:perhaps,asIhavepointedoutabove,afairlysmallfractionofit,ifanysubstantive
meaningisassignedtothetermreinforced.Thisdefinitionissubsequentlyrefinedbytheadditionalprovision
thatthemediatingresponseofthereinforcingperson(thelistener)mustitself"havebeenconditioned
preciselyinordertoreinforcethebehaviorofthespeaker"(225,italicshis).Thisstillcoverstheexamples
givenabove,ifwecanassumethatthereinforcingbehaviorofthepsychologist,theparent,theopposing
boxer,andthepayingcustomeraretheresultofappropriatetraining,whichisperhapsnotunreasonable.A
significantpartofthefragmentoflinguisticbehaviorcoveredbytheearlierdefinitionwillnodoubtbe
excludedbytherefinement,however.Suppose,forexample,thatwhilecrossingthestreetIhearsomeone
shoutWatchoutforthecarandjumpoutoftheway.Itcanhardlybeproposedthatmyjumping(the
mediating,reinforcingresponseinSkinner'susage)wasconditioned(thatis,Iwastrainedtojump)precisely
inordertoreinforcethebehaviorofthespeakerandsimilarly,forawideclassofcases.Skinner'sassertion
thatwiththisrefineddefinition"wenarrowoursubjecttowhatistraditionallyrecognizedastheverbalfield"
(225)appearstobegrosslyinerror.
VII
VerbaloperantsareclassifiedbySkinnerintermsoftheir"functional"relationtodiscriminatedstimulus,
reinforcement,andotherverbalresponses.Amandisdefinedas"averbaloperantinwhichtheresponseis
reinforcedbyacharacteristicconsequenceandisthereforeunderthefunctionalcontrolofrelevantconditions
ofdeprivationoraversivestimulation"(35).Thisismeanttoincludequestions,commands,etc.Eachofthe
termsinthisdefinitionraisesahostofproblems.AmandsuchasPassthesaltisaclassofresponses.We
cannottellbyobservingtheformofaresponsewhetheritbelongstothisclass(Skinnerisveryclearabout
this),butonlybyidentifyingthecontrollingvariables.Thisisgenerallyimpossible.Deprivationisdefinedin
thebarpressingexperimentintermsoflengthoftimethattheanimalhasnotbeenfedorpermittedtodrink.
Inthepresentcontext,however,itisquiteamysteriousnotion.Noattemptismadeheretodescribeamethod
fordetermining"relevantconditionsofdeprivation"independentlyofthe"controlled"response.Itisofno
helpatalltobetold(32)thatitcanbecharacterizedintermsoftheoperationsoftheexperimenter.Ifwe
definedeprivationintermsofelapsedtime,thenatanymomentapersonisincountlessstatesof
deprivation.34Itappearsthatwemustdecidethattherelevantconditionofdeprivationwas(say)salt
deprivation,onthebasisofthefactthatthespeakeraskedforsalt(thereinforcingcommunitywhich"sets
up"themandisinasimilarpredicament).Inthiscase,theassertionthatamandisunderthecontrolof
relevantdeprivationisempty,andweare(contrarytoSkinner'sintention)identifyingtheresponseasamand
completelyintermsofform.Thewordrelevantinthedefinitionaboveconcealssomeratherserious
complications.
InthecaseofthemandPassthesalt,theworddeprivationisnotoutofplace,thoughitappearstobeoflittle
useforfunctionalanalysis.SupposehoweverthatthespeakersaysGivemethebook,Takemeforaride,or
http://cogprints.org/1148/1/chomsky.htm

11/24

08/09/2016

BadNews:NoamChomsky

Letmefixit.Whatkindsofdeprivationcanbeassociatedwiththesemands?Howdowedetermineor
measuretherelevantdeprivation?Ithinkwemustconcludeinthiscase,asbefore,eitherthatthenotion
deprivationisrelevantatmosttoaminutefragmentofverbalbehavior,orelsethatthestatement"Xisunder
Ydeprivation"isjustanoddparaphrasefor"XwantsY,"bearingamisleadingandunjustifiableconnotation
ofobjectivity.
Thenotionaversivecontrolisjustasconfused.Thisisintendedtocoverthreats,beating,andthelike(33).
Themannerinwhichaversivestimulationfunctionsissimplydescribed.Ifaspeakerhashadahistoryof
appropriatereinforcement(e.g.,ifacertainresponsewasfollowedby"cessationofthethreatofsuchinjury
ofeventswhichhavepreviouslybeenfollowedbysuchinjuryandwhicharethereforeconditionedaversive
stimuli"),thenhewilltendtogivetheproperresponsewhenthethreatwhichhadpreviouslybeenfollowed
bytheinjuryispresented.Itwouldappeartofollowfromthisdescriptionthataspeakerwillnotrespond
properlytothemandYourmoneyoryourlife(38)unlesshehasapasthistoryofbeingkilled.Butevenifthe
difficultiesindescribingthemechanismofaversivecontrolaresomehowremovedbyamorecareful
analysis,itwillbeoflittleuseforidentifyingoperantsforreasonssimilartothosementionedinthecaseof
deprivation.
Itseems,then,thatinSkinner'stermsthereisinmostcasesnowaytodecidewhetheragivenresponseisan
instanceofaparticularmand.Henceitismeaningless,withinthetermsofhissystem,tospeakofthe
characteristicconsequencesofamand,asinthedefinitionabove.Furthermore,evenifweextendthesystem
sothatmandscansomehowbeidentified,wewillhavetofacetheobviousfactthatmostofusarenot
fortunateenoughtohaveourrequests,commands,advice,andsooncharacteristicallyreinforced(theymay
neverthelessexistinconsiderablestrength).Theseresponsescouldthereforenotbeconsideredmandsby
Skinner.Infact,Skinnersetsupacategoryof"magicalmands"(4849)tocoverthecaseof"mandswhich
cannotbeaccountedforbyshowingthattheyhaveeverhadtheeffectspecifiedoranysimilareffectupon
similaroccasions"(thewordeverinthisstatementshouldbereplacedbycharacteristically).Inthese
pseudomands,"thespeakersimplydescribesthereinforcementappropriatetoagivenstateofdeprivationor
aversivestimulation."Inotherwords,giventhemeaningthatwehavebeenledtoassigntoreinforcement
anddeprivation,thespeakerasksforwhathewants.Theremarkthat"aspeakerappearstocreatenewmands
ontheanalogyofoldones"isalsonotveryhelpful.
Skinner'sclaimthathisnewdescriptivesystemissuperiortothetraditionalone"becauseitstermscanbe
definedwithrespecttoexperimentaloperations"(45)is,weseeonceagain,anillusion.Thestatement"X
wantsY"isnotclarifiedbypointingoutarelationbetweenrateofbarpressingandhoursoffood
deprivationreplacing"XwantsY"by"XisdeprivedofY"addsnonewobjectivitytothedescriptionof
behavior.Hisfurtherclaimforthesuperiorityofthenewanalysisofmandsisthatitprovidesanobjective
basisforthetraditionalclassificationintorequests,commands,etc.(3841).Thetraditionalclassificationis
intermsoftheintentionofthespeaker.Butintention,Skinnerholds,canbereducedtocontingenciesof
reinforcement,and,correspondingly,wecanexplainthetraditionalclassificationintermsofthereinforcing
behaviorofthelistener.Thus,aquestionisamandwhich"specifiesverbalaction,andthebehaviorofthe
listenerpermitsustoclassifyitasarequest,acommand,oraprayer"(39).Itisarequestif"thelisteneris
independentlymotivatedtoreinforcethespeaker"acommandif"thelistener'sbehavioris...reinforcedby
reducingathreat,aprayerifthemand"promotesreinforcementbygeneratinganemotionaldisposition."The
mandisadviceifthelistenerispositivelyreinforcedbytheconsequencesofmediatingthereinforcementof
thespeakeritisawarningif"bycarryingoutthebehaviorspecifiedbythespeaker,thelistenerescapesfrom
aversivestimulation"andsoon.AllthisisobviouslywrongifSkinnerisusingthewordsrequest,command,
etc.,inanythinglikethesenseofthecorrespondingEnglishwords.Thewordquestiondoesnotcover
commands.Pleasepassthesaltisarequest(butnotaquestion),whetherornotthelistenerhappenstobe
motivatedtofulfillitnoteveryonetowhomarequestisaddressedisfavorablydisposed.Aresponsedoes
notceasetobeacommandifitisnotfollowednordoesaquestionbecomeacommandifthespeaker
answersitbecauseofanimpliedorimaginedthreat.Notalladviceisgoodadvice,andaresponsedoesnot
ceasetobeadviceifitisnotfollowed.Similarly,awarningmaybemisguidedheedingitmaycause
aversivestimulation,andignoringitmightbepositivelyreinforcing.Inshort,theentireclassificationis
besidethepoint.Amoment'sthoughtissufficienttodemonstratetheimpossibilityofdistinguishingbetween
requests,commands,advice,etc.,onthebasisofthebehaviorordispositionoftheparticularlistener.Norcan
wedothisonthebasisofthetypicalbehaviorofalllisteners.Someadviceisnevertaken,isalwaysbad,etc.,
andsimilarly,withotherkindsofmands.Skinner'sevidentsatisfactionwiththisanalysisofthetraditional
classificationisextremelypuzzling.
http://cogprints.org/1148/1/chomsky.htm

12/24

08/09/2016

BadNews:NoamChomsky

VIII
Mandsareoperantswithnospecifiedrelationtoapriorstimulus.Atact,ontheotherhand,isdefinedas"a
verbaloperantinwhicharesponseofgivenformisevoked(oratleaststrengthened)byaparticularobjector
eventorpropertyofanobjectorevent"(81).Theexamplesquotedinthediscussionofstimuluscontrol
(Section3)arealltacts.Theobscurityofthenotionstimuluscontrolmakestheconceptofthetactrather
mystical.Since,however,thetactis"themostimportantofverbaloperants,"itisimportanttoinvestigatethe
developmentofthisconceptinmoredetail.
Wefirstaskwhytheverbalcommunity"setsup"tactsinthechildthatis,howtheparentisreinforcedby
settingupthetact.Thebasicexplanationforthisbehavioroftheparent(8586)isthereinforcementhe
obtainsbythefactthathiscontactwiththeenvironmentisextendedtouseSkinner'sexample,thechildmay
laterbeabletocallhimtothetelephone.(Itisdifficulttosee,then,howfirstchildrenacquiretacts,sincethe
parentdoesnothavetheappropriatehistoryofreinforcement.)Reasoninginthesameway,wemayconclude
thattheparentinducesthechildtowalksothathecanmakesomemoneydeliveringnewspapers.Similarly,
theparentsetsupan"echoicrepertoire"(e.g.,aphonemicsystem)inthechildbecausethismakesiteasierto
teachhimnewvocabulary,andextendingthechild'svocabularyisultimatelyusefultotheparent."Inall
thesecasesweexplainthebehaviorofthereinforcinglistenerbypointingtoanimprovementinthe
possibilityofcontrollingthespeakerwhomhereinforces"(56).Perhapsthisprovidestheexplanationforthe
behavioroftheparentininducingthechildtowalk:theparentisreinforcedbytheimprovementinhis
controlofthechildwhenthechild'smobilityincreases.Underlyingthesemodesofexplanationisacurious
viewthatitissomehowmorescientifictoattributetoaparentadesiretocontrolthechildorenhancehis
ownpossibilitiesforactionthanadesiretoseethechilddevelopandextendhiscapacities.Needlesstosay,
noevidenceisofferedtosupportthiscontention.
Considernowtheproblemofexplainingtheresponseofthelistenertoatact.Suppose,forexample,thatB
hearsAsayfoxandreactsappropriatelylooksaround,runsaway,aimshisrifle,etc.Howcanweexplain
B'sbehavior?SkinnerrightlyrejectsanalysesofthisofferedbyJ.B.WatsonandBertrandRussell.Hisown
equallyinadequateanalysisproceedsasfollows(8788).Weassume(l)"thatinthehistoryof[B]the
stimulusfoxhasbeenanoccasionuponwhichlookingaroundhasbeenfollowedbyseeingafox"and(2)
"thatthelistenerhassomecurrent'interestinseeingfoxes'thatbehaviorwhichdependsuponaseenfox
foritsexecutionisstrong,andthatthestimulussuppliedbyafoxisthereforereinforcing."Bcarriesoutthe
appropriatebehavior,then,because"theheardstimulusfoxistheoccasionuponwhichturningandlooking
aboutisfrequentlyfollowedbythereinforcementofseeingafox,"i.e,hisbehaviorisadiscriminated
operant.Thisexplanationisunconvincing.Bmayneverhaveseenafoxandmayhavenocurrentinterestin
seeingone,andyetmayreactappropriatelytothestimulusfox.35Sinceexactlythesamebehaviormaytake
placewhenneitheroftheassumptionsisfulfilled,someothermechanismmustbeoperativehere.
Skinnerremarksseveraltimesthathisanalysisofthetactintermsofstimuluscontrolisanimprovement
overthetraditionalformulationsintermsofreferenceandmeaning.Thisissimplynottrue.Hisanalysisis
fundamentallythesameasthetraditionalone,thoughmuchlesscarefullyphrased.Inparticular,itdiffers
onlybyindiscriminateparaphraseofsuchnotionsasdenotation(reference)andconnotation(meaning),
whichhavebeenkeptclearlyapartintraditionalformulations,intermsofthevagueconceptstimulus
control.Inonetraditionalformulationadescriptivetermissaidtodenoteasetofentitiesandtoconnoteor
designateacertainpropertyorconditionthatanentitymustpossessorfulfilifthetermistoapplytoit.36
Thus,thetermvertebraterefersto(denotes,istrueof)vertebratesandconnotesthepropertyhavingaspine
orsomethingofthesort.Thisconnoteddefiningpropertyiscalledthemeaningoftheterm.Twotermsmay
havethesamereferencebutdifferentmeanings.Thus,itisapparentlytruethatthecreatureswithheartsare
allandonlythevertebrates.Ifso,thenthetermcreaturewithaheartreferstovertebratesanddesignatesthe
propertyhavingaheart.Thisispresumablyadifferentproperty(adifferentgeneralcondition)fromhavinga
spinehencethetermsvertebrateandcreaturewithaheartaresaidtohavedifferentmeanings.Thisanalysis
isnotincorrect(foratleastonesenseofmeaning),butitsmanylimitationshavefrequentlybeenpointed
out.37Themajorproblemisthatthereisnogoodwaytodecidewhethertwodescriptivetermsdesignatethe
sameproperty.38Aswehavejustseen,itisnotsufficientthattheyrefertothesameobjects.Vertebrateand
creaturewithaspinewouldbesaidtodesignatethesameproperty(distinctfromthatdesignatedbycreature
withaheart).Ifweaskwhythisisso,theonlyanswerappearstobethatthetermsaresynonymous.The
notionpropertythusseemssomehowlanguagebound,andappealto"definingproperties"shedslittlelight
onquestionsofmeaningandsynonymy.
http://cogprints.org/1148/1/chomsky.htm

13/24

08/09/2016

BadNews:NoamChomsky

Skinneracceptsthetraditionalaccountintoto,ascanbeseenfromhisdefinitionofatactasaresponseunder
controlofaproperty(stimulus)ofsomephysicalobjectorevent.Wehavefoundthatthenotioncontrolhas
norealsubstanceandisperhapsbestunderstoodasaparaphraseofdenoteorconnoteor,ambiguously,both.
Theonlyconsequenceofadoptingthenewtermstimuluscontrolisthattheimportantdifferencesbetween
referenceandmeaningareobscured.Itprovidesnonewobjectivity.Thestimuluscontrollingtheresponseis
determinedbytheresponseitselfthereisnoindependentandobjectivemethodofidentification(seeSection
3).Consequently,whenSkinnerdefinessynonymyasthecaseinwhich"thesamestimulusleadstoquite
differentresponses"(118),wecanhavenoobjection.Theresponseschairandredmadealternativelytothe
sameobjectarenotsynonymous,becausethestimuliarecalleddifferent.Theresponsesvertebrateand
creaturewithaspinewouldbeconsideredsynonymousbecausetheyarecontrolledbythesamepropertyof
theobjectunderinvestigationinmoretraditionalandnolessscientificterms,theyevokethesameconcept.
Similarly,whenmetaphoricalextensionisexplainedasdueto"thecontrolexercisedbypropertiesofthe
stimuluswhich,thoughpresentatreinforcement,donotenterintothecontingencyrespectedbytheverbal
community"(92traditionally,accidentalproperties),noobjectioncanberaisedwhichhasnotalreadybeen
leveledagainstthetraditionalaccount.Justaswecould"explain"theresponseMozarttoapieceofmusicin
termsofsubtlepropertiesofthecontrollingstimuli,wecan,withequalfacility,explaintheappearanceofthe
responsesunwhennosunispresent,asinJulietis[like]thesun."WedosobynotingthatJulietandthesun
havecommonproperties,atleastintheireffectonthespeaker"(93).Sinceanytwoobjectshaveindefinitely
manypropertiesincommon,wecanbecertainthatwewillneverbeatalosstoexplainaresponseofthe
formAislikeB,forarbitraryAandB.Itisclear,however,thatSkinner'srecurrentclaimthathisformulation
issimplerandmorescientificthanthetraditionalaccounthasnobasisinfact.
Tactsunderthecontrolofprivatestimuli(Bloomfield's"displacedspeech")formalargeandimportantclass
(13046),includingnotonlysuchresponsesasfamiliarandbeautiful,butalsoverbalresponsesreferringto
past,potential,orfutureeventsorbehavior.Forexample,theresponseTherewasanelephantatthezoo
"mustbeunderstoodasaresponsetocurrentstimuli,includingeventswithinthespeakerhimself"(143).39If
wenowaskourselveswhatproportionofthetactsinactuallifeareresponsesto(descriptionsof)actual
currentoutsidestimulation,wecanseejusthowlargearolemustbeattributedtoprivatestimuli.Aminute
amountofverbalbehavior,outsidethenursery,consistsofsuchremarksasThisisredandThereisaman.
Thefactthatfunctionalanalysismustmakesuchaheavyappealtoobscureinternalstimuliisagaina
measureofitsactualadvanceovertraditionalformulations.
IX
Responsesunderthecontrolofpriorverbalstimuliareconsideredunderadifferentheadingfromthetact.An
echoicoperantisaresponsewhich"generatesasoundpatternsimilartothatofthestimulus"(55).Itcovers
onlycasesofimmediateimitation.40Noattemptismadetodefinethesenseinwhichachild'sechoic
responseis"similar"tothestimulusspokeninthefather'sbassvoiceitseems,thoughtherearenoclear
statementsaboutthis,thatSkinnerwouldnotaccepttheaccountofthephonologistinthisrespect,but
nothingelseisoffered.Thedevelopmentofanechoicrepertoireisattributedcompletelytodifferential
reinforcement.Sincethespeakerwilldonomore,accordingtoSkinner,thanwhatisdemandedofhimbythe
verbalcommunity,thedegreeofaccuracyinsistedonbythiscommunitywilldeterminetheelementsofthe
repertoire,whateverthesemaybe(notnecessarilyphonemes)."Inaverbalcommunitywhichdoesnotinsist
onaprecisecorrespondence,anechoicrepertoiremayremainslackandwillbelesssuccessfullyappliedto
novelpatterns."Thereisnodiscussionofsuchfamiliarphenomenaastheaccuracywithwhichachildwill
pickupasecondlanguageoralocaldialectinthecourseofplayingwithotherchildren,whichseemsharply
inconflictwiththeseassertions.Noanthropologicalevidenceiscitedtosupporttheclaimthataneffective
phonemicsystemdoesnotdevelop(thisisthesubstanceofthequotedremark)incommunitiesthatdonot
insistonprecisecorrespondence.
Averbalresponsetoawrittenstimulus(reading)iscalledtextualbehavior.
Otherverbalresponsestoverbalstimuliarecalledintraverbaloperants.Paradigminstancesaretheresponse
fourtothestimulustwoplustwoortheresponseParistothestimuluscapitalofFrance.Simpleconditioning
maybesufficienttoaccountfortheresponsefourtotwoplustwo,41butthenotionofintraverbalresponse
losesallmeaningwhenwefinditextendedtocovermostofthefactsofhistoryandmanyofthefactsof
science(72,129)allwordassociationand"flightofideas"(7376)alltranslationsandparaphrase(77)
reportsofthingsseen,heard,orremembered(315)and,ingeneral,largesegmentsofscientific,
http://cogprints.org/1148/1/chomsky.htm

14/24

08/09/2016

BadNews:NoamChomsky

mathematical,andliterarydiscourse.Obviously,thekindofexplanationthatmightbeproposedfora
student'sabilitytorespondwithParistocapitalofFrance,aftersuitablepractice,canhardlybeseriously
offeredtoaccountforhisabilitytomakeajudiciousguessinansweringthequestions(tohimnew):Whatis
theseatoftheFrenchgovernment?,...thesourceoftheliterarydialect?,..thechieftargetoftheGerman
blitzkrieg?,etc.,orhisabilitytoproveanewtheorem,translateanewpassage,orparaphrasearemarkforthe
firsttimeorinanewway.
Theprocessof"gettingsomeonetoseeapoint,"toseesomethingyourway,ortounderstandacomplexstate
ofaffairs(e.g.,adifficultpoliticalsituationoramathematicalproof)is,forSkinner,simplyamatterof
increasingthestrengthofthelistener'salreadyavailablebehavior.42Since"theprocessisoftenexemplified
byrelativelyintellectualscientificorphilosophicaldiscourse,"Skinnerconsidersit"allthemoresurprising
thatitmaybereducedtoechoic,textual,orintraverbalsupplementation"(269).Again,itisonlythe
vaguenessandlatitudewithwhichthenotionsstrengthandintraverbalresponseareusedthatsavethisfrom
absurdity.Ifweusethesetermsintheirliteralsense,itisclearthatunderstandingastatementcannotbe
equatedtoshoutingitfrequentlyinahighpitchedvoice(highresponsestrength),andacleverand
convincingargumentcannotbeaccountedforonthebasisofahistoryofpairingsofverbalresponses.43
X
Afinalclassofoperants,calledautoclitics,includesthosethatareinvolvedinassertion,negation,
quantification,qualificationofresponses,constructionofsentences,andthe"highlycomplexmanipulations
ofverbalthinking."Alltheseactsaretobeexplained"intermsofbehaviorwhichisevokedbyoractsupon
otherbehaviorofthespeaker"(313).Autocliticsare,then,responsestoalreadygivenresponses,orrather,as
wefindinreadingthroughthissection,theyareresponsestocovertorincipientorpotentialverbalbehavior.
AmongtheautocliticsarelistedsuchexpressionsasIrecall,Iimagine,forexample,assume,letXequal...,
thetermsofnegation,theisofpredicationandassertion,all,some,if,then,and,ingeneral,allmorphemes
otherthannouns,verbs,andadjectives,aswellasgrammaticalprocessesoforderingandarrangement.
Hardlyaremarkinthissectioncanbeacceptedwithoutseriousqualification.Totakejustoneexample,
considerSkinner'saccountoftheautocliticallinAllswansarewhite(329).Obviouslywecannotassume
thatthisisatacttoallswansasstimulus.Itissuggested,therefore,thatwetakealltobeanautoclitic
modifyingthewholesentenceSwansarewhite.Allcanthenbetakenasequivalenttoalways,oralwaysitis
possibletosay.Notice,however,thatthemodifiedsentenceSwansarewhiteisjustasgeneralasAllswans
arewhite.Furthermore,theproposedtranslationofallisincorrectiftakenliterally.Itisjustaspossibleto
saySwansaregreenastosaySwansarewhite.Itisnotalwayspossibletosayeither(e.g.,whileyouare
sayingsomethingelseorsleeping).ProbablywhatSkinnermeansisthatthesentencecanbeparaphrased"X
iswhiteistrue,foreachswanX."Butthisparaphrasecannotbegivenwithinhissystem,whichhasnoplace
fortrue.
Skinner'saccountofgrammarandsyntaxasautocliticprocesses(Chap.13)differsfromafamiliartraditional
accountmainlyintheuseofthepseudoscientifictermscontrolorevokeinplaceofthetraditionalrefer.
Thus,inTheboyruns,thefinalsofrunsisatactundercontrolofsuch"subtlepropertiesofasituation"as
"thenatureofrunningasanactivityratherthananobjectorpropertyofanobject."44(Presumably,then,in
Theattemptfails,Thedifficultyremains,Hisanxietyincreases,etc.,wemustalsosaythatthesindicatesthat
theobjectdescribedastheattemptiscarryingouttheactivityoffailing,etc.)Intheboy'sgun,however,thes
denotespossession(as,presumably,intheboy'sarrival,...story,...age,etc.)andisunderthecontrolofthis
"relationalaspectofthesituation"(336).The"relationalautocliticoforder"(whateveritmaymeantocall
theorderofasetofresponsesaresponsetothem)inTheboyrunsthestoreisunderthecontrolofan
"extremelycomplexstimulussituation,"namely,thattheboyisrunningthestore(335).Andinthehatand
theshoeisunderthecontroloftheproperty"pair."Throughinthedogwentthroughthehedgeisunderthe
controlofthe"relationbetweenthegoingdogandthehedge"(342).Ingeneral,nounsareevokedbyobjects,
verbsbyactions,andsoon.Skinnerconsidersasentencetobeasetofkeyresponses(nouns,verbs,
adjectives)onaskeletalframe(346).IfweareconcernedwiththefactthatSamrentedaleakyboat,theraw
responsestothesituationarerent,boat,leak,andSam.Autoclitics(includingorder)whichqualifythese
responses,expressrelationsbetweenthem,andthelike,arethenaddedbyaprocesscalledcompositionand
theresultisagrammaticalsentence,oneofmanyalternativesamongwhichselectionisratherarbitrary.The
ideathatsentencesconsistoflexicalitemsplacedinagrammaticalframeisofcourseatraditionalone,
withinbothphilosophyandlinguistics.Skinneraddstoitonlytheveryimplausiblespeculationthatinthe
http://cogprints.org/1148/1/chomsky.htm

15/24

08/09/2016

BadNews:NoamChomsky

internalprocessofcomposition,thenouns,verbs,andadjectivesarechosenfirstandthenarearranged,
qualified,etc.,byautocliticresponsestotheseinternalactivities.45
Thisviewofsentencestructure,whetherphrasedintermsofautoclitics,syncategorematicexpressions,or
grammaticalandlexicalmorphemes,isinadequate.Sheepprovidewoolhasno(physical)frameatall,butno
otherarrangementofthesewordsisanEnglishsentence.Thesequencesfuriouslysleepideasgreencolorless
andfriendlyyoungdogsseemharmlesshavethesameframes,butonlyoneisasentenceofEnglish
(similarly,onlyoneofthesequencesformedbyreadingthesefrombacktofront).Strugglingartistscanbea
nuisancehasthesameframeasmarkingpaperscanbeanuisance,butisquitedifferentinsentence
structure,ascanbeseenbyreplacingcanbebyisorareinbothcases.Therearemanyothersimilarand
equallysimpleexamples.Itisevidentthatmoreisinvolvedinsentencestructurethaninsertionoflexical
itemsingrammaticalframesnoapproachtolanguagethatfailstotakethesedeeperprocessesintoaccount
canpossiblyachievemuchsuccessinaccountingforactuallinguisticbehavior.
XI
TheprecedingdiscussioncoversallthemajornotionsthatSkinnerintroducesinhisdescriptivesystem.My
purposeindiscussingtheconceptsonebyonewastoshowthatineachcase,ifwetakehistermsintheir
literalmeaning,thedescriptioncoversalmostnoaspectofverbalbehavior,andifwetakethem
metaphorically,thedescriptionoffersnoimprovementovervarioustraditionalformulations.Theterms
borrowedfromexperimentalpsychologysimplylosetheirobjectivemeaningwiththisextension,andtake
overthefullvaguenessofordinarylanguage.SinceSkinnerlimitshimselftosuchasmallsetoftermsfor
paraphrase,manyimportantdistinctionsareobscured.Ithinkthatthisanalysissupportstheviewexpressed
inSectionI,thateliminationoftheindependentcontributionofthespeakerandlearner(aresultwhich
Skinnerconsidersofgreatimportance,cf.31112)canbeachievedonlyatthecostofeliminatingall
significancefromthedescriptivesystem,whichthenoperatesatalevelsogrossandcrudethatnoanswers
aresuggestedtothemostelementaryquestions.46ThequestionstowhichSkinnerhasaddressedhis
speculationsarehopelesslypremature.Itisfutiletoinquireintothecausationofverbalbehavioruntilmuch
moreisknownaboutthespecificcharacterofthisbehaviorandthereislittlepointinspeculatingaboutthe
processofacquisitionwithoutmuchbetterunderstandingofwhatisacquired.
Anyonewhoseriouslyapproachesthestudyoflinguisticbehavior,whetherlinguist,psychologist,or
philosopher,mustquicklybecomeawareoftheenormousdifficultyofstatingaproblemwhichwilldefine
theareaofhisinvestigations,andwhichwillnotbeeithercompletelytrivialorhopelesslybeyondtherange
ofpresentdayunderstandingandtechnique.Inselectingfunctionalanalysisashisproblem,Skinnerhasset
himselfataskofthelattertype.Inanextremelyinterestingandinsightfulpaper,47K.S.Lashleyhas
implicitlydelimitedaclassofproblemswhichcanbeapproachedinafruitfulwaybythelinguistand
psychologist,andwhichareclearlypreliminarytothosewithwhichSkinnerisconcerned.Lashley
recognizes,asanyonemustwhoseriouslyconsidersthedata,thatthecompositionandproductionofan
utteranceisnotsimplyamatterofstringingtogetherasequenceofresponsesunderthecontrolofoutside
stimulationandintraverbalassociation,andthatthesyntacticorganizationofanutteranceisnotsomething
directlyrepresentedinanysimplewayinthephysicalstructureoftheutteranceitself.Avarietyof
observationsleadhimtoconcludethatsyntacticstructureis"ageneralizedpatternimposedonthespecific
actsastheyoccur"(512),andthat"aconsiderationofthestructureofthesentenceandothermotorsequences
willshow...thatthereare,behindtheovertlyexpressedsequences,amultiplicityofintegrativeprocesses
whichcanonlybeinferredfromthefinalresultsoftheiractivity"(509).Healsocommentsonthegreat
difficultyofdeterminingthe"selectivemechanisms"usedintheactualconstructionofaparticularutterance
(522).
Althoughpresentdaylinguisticscannotprovideapreciseaccountoftheseintegrativeprocesses,imposed
patterns,andselectivemechanisms,itcanatleastsetitselftheproblemofcharacterizingthesecompletely.It
isreasonabletoregardthegrammarofalanguageLideallyasamechanismthatprovidesanenumerationof
thesentencesofLinsomethinglikethewayinwhichadeductivetheorygivesanenumerationofasetof
theorems.(Grammar,inthissenseoftheword,includesphonology.)Furthermore,thetheoryoflanguage
canberegardedasastudyoftheformalpropertiesofsuchgrammars,and,withapreciseenough
formulation,thisgeneraltheorycanprovideauniformmethodfordetermining,fromtheprocessof
generationofagivensentence,astructuraldescriptionwhichcangiveagooddealofinsightintohowthis
sentenceisusedandunderstood.Inshort,itshouldbepossibletoderivefromaproperlyformulated
http://cogprints.org/1148/1/chomsky.htm

16/24

08/09/2016

BadNews:NoamChomsky

grammarastatementoftheintegrativeprocessesandgeneralizedpatternsimposedonthespecificactsthat
constituteanutterance.Therulesofagrammaroftheappropriateformcanbesubdividedintothetwotypes,
optionalandobligatoryonlythelattermustbeappliedingeneratinganutterance.Theoptionalrulesofthe
grammarcanbeviewed,then,astheselectivemechanismsinvolvedintheproductionofaparticular
utterance.Theproblemofspecifyingtheseintegrativeprocessesandselectivemechanismsisnontrivialand
notbeyondtherangeofpossibleinvestigation.Theresultsofsuchastudymight,asLashleysuggests,beof
independentinterestforpsychologyandneurology(andconversely).Althoughsuchastudy,evenif
successful,wouldbynomeansanswerthemajorproblemsinvolvedintheinvestigationofmeaningandthe
causationofbehavior,itsurelywillnotbeunrelatedtothese.Itisatleastpossible,furthermore,thatsucha
notionassemanticgeneralization,towhichsuchheavyappealismadeinallapproachestolanguageinuse,
concealscomplexitiesandspecificstructureofinferencenotfardifferentfromthosethatcanbestudiedand
exhibitedinthecaseofsyntax,andthatconsequentlythegeneralcharacteroftheresultsofsyntactic
investigationsmaybeacorrectivetooversimplifiedapproachestothetheoryofmeaning.
Thebehaviorofthespeaker,listener,andlearneroflanguageconstitutes,ofcourse,theactualdataforany
studyoflanguage.Theconstructionofagrammarwhichenumeratessentencesinsuchawaythata
meaningfulstructuraldescriptioncanbedeterminedforeachsentencedoesnotinitselfprovideanaccountof
thisactualbehavior.Itmerelycharacterizesabstractlytheabilityofonewhohasmasteredthelanguageto
distinguishsentencesfromnonsentences,tounderstandnewsentences(inpart),tonotecertainambiguities,
etc.Theseareveryremarkableabilities.Weconstantlyreadandhearnewsequencesofwords,recognize
themassentences,andunderstandthem.Itiseasytoshowthattheneweventsthatweacceptandunderstand
assentencesarenotrelatedtothosewithwhichwearefamiliarbyanysimplenotionofformal(orsemantic
orstatistical)similarityoridentityofgrammaticalframe.Talkofgeneralizationinthiscaseisentirely
pointlessandempty.Itappearsthatwerecognizeanewitemasasentencenotbecauseitmatchessome
familiariteminanysimpleway,butbecauseitisgeneratedbythegrammarthateachindividualhas
somehowandinsomeforminternalized.Andweunderstandanewsentence,inpart,becauseweare
somehowcapableofdeterminingtheprocessbywhichthissentenceisderivedinthisgrammar.
Supposethatwemanagetoconstructgrammarshavingthepropertiesoutlinedabove.Wecanthenattemptto
describeandstudytheachievementofthespeaker,listener,andlearner.Thespeakerandthelistener,we
mustassume,havealreadyacquiredthecapacitiescharacterizedabstractlybythegrammar.Thespeaker's
taskistoselectaparticularcompatiblesetofoptionalrules.Ifweknow,fromgrammaticalstudy,what
choicesareavailabletohimandwhatconditionsofcompatibilitythechoicesmustmeet,wecanproceed
meaningfullytoinvestigatethefactorsthatleadhimtomakeoneoranotherchoice.Thelistener(orreader)
mustdetermine,fromanexhibitedutterance,whatoptionalruleswerechosenintheconstructionofthe
utterance.Itmustbeadmittedthattheabilityofahumanbeingtodothisfarsurpassesourpresent
understanding.Thechildwholearnsalanguagehasinsomesenseconstructedthegrammarforhimselfon
thebasisofhisobservationofsentencesandnonsentences(i.e.,correctionsbytheverbalcommunity).Study
oftheactualobservedabilityofaspeakertodistinguishsentencesfromnonsentences,detectambiguities,
etc.,apparentlyforcesustotheconclusionthatthisgrammarisofanextremelycomplexandabstract
character,andthattheyoungchildhassucceededincarryingoutwhatfromtheformalpointofview,atleast,
seemstobearemarkabletypeoftheoryconstruction.Furthermore,thistaskisaccomplishedinan
astonishinglyshorttime,toalargeextentindependentlyofintelligence,andinacomparablewaybyall
children.Anytheoryoflearningmustcopewiththesefacts.
Itisnoteasytoaccepttheviewthatachildiscapableofconstructinganextremelycomplexmechanismfor
generatingasetofsentences,someofwhichhehasheard,orthatanadultcaninstantaneouslydetermine
whether(andifso,how)aparticularitemisgeneratedbythismechanism,whichhasmanyoftheproperties
ofanabstractdeductivetheory.Yetthisappearstobeafairdescriptionoftheperformanceofthespeaker,
listener,andlearner.Ifthisiscorrect,wecanpredictthatadirectattempttoaccountfortheactualbehavior
ofspeaker,listener,andlearner,notbasedonapriorunderstandingofthestructureofgrammars,willachieve
verylimitedsuccess.Thegrammarmustberegardedasacomponentinthebehaviorofthespeakerand
listenerwhichcanonlybeinferred,asLashleyhasputit,fromtheresultingphysicalacts.Thefactthatall
normalchildrenacquireessentiallycomparablegrammarsofgreatcomplexitywithremarkablerapidity
suggeststhathumanbeingsaresomehowspeciallydesignedtodothis,withdatahandlingor"hypothesis
formulating"abilityofunknowncharacterandcomplexity.48Thestudyoflinguisticstructuremayultimately
leadtosomesignificantinsightsintothismatter.Atthemomentthequestioncannotbeseriouslyposed,but
inprincipleitmaybepossibletostudytheproblemofdeterminingwhatthebuiltinstructureofan
http://cogprints.org/1148/1/chomsky.htm

17/24

08/09/2016

BadNews:NoamChomsky

informationprocessing(hypothesisforming)systemmustbetoenableittoarriveatthegrammarofa
languagefromtheavailabledataintheavailabletime.Atanyrate,justastheattempttoeliminatethe
contributionofthespeakerleadstoa"mentalistic"descriptivesystemthatsucceedsonlyinblurring
importanttraditionaldistinctions,arefusaltostudythecontributionofthechildtolanguagelearningpermits
onlyasuperficialaccountoflanguageacquisition,withavastandunanalyzedcontributionattributedtoa
stepcalledgeneralizationwhichinfactincludesjustabouteverythingofinterestinthisprocess.Ifthestudy
oflanguageislimitedintheseways,itseemsinevitablethatmajoraspectsofverbalbehaviorwillremaina
mystery.
Notes
1Skinner'sconfidenceinrecentachievementsinthestudyofanimalbehaviorandtheirapplicabilityto
complexhumanbehaviordoesnotappeartobewidelyshared.Inmanyrecentpublicationsofconfirmed
behavioriststhereisaprevailingnoteofskepticismwithregardtothescopeoftheseachievements.For
representativecomments,seethecontributionstoModernLearningTheory(byW.K.Estesetal.NewYork:
AppletonCenturyCrofts,Inc.,1954)B.R.Bugelski,PsychologyofLearning(NewYork:Holt,Rinehart&
Winston,Inc.,1956)S.Koch,inNebraskaSymposiumonMotivation,58(Lincoln,1956)W.S.Verplanck,
"LearnedandInnateBehavior,"Psych.Rev.,52,(1955),139.PerhapsthestrongestviewisthatofH.Harlow,
whohasasserted("Mice,Monkeys,Men,andMotives,"Psych.Rev.,60,[1953]2332)that"astrongcase
canbemadeforthepropositionthattheimportanceofthepsychologicalproblemsstudiedduringthelast15
yearshasdecreasedasanegativelyacceleratedfunctionapproachinganasymptoteofcompleteindifference."
N.Tinbergen,aleadingrepresentativeofadifferentapproachtoanimalbehaviorstudies(comparative
ethology),concludesadiscussionoffunctionalanalysiswiththecommentthat"wemaynowdrawthe
conclusionthatthecausationofbehaviorisimmenselymorecomplexthanwasassumedinthe
generalizationsofthepast.Anumberofinternalandexternalfactorsactuponcomplexcentralnervous
structures.Second,itwillbeobviousthatthefactsatourdisposalareveryfragmentaryindeed"TheStudy
ofInstinct(Toronto:OxfordUniv.Press,1951),p.74.
2InBehaviorofOrganisms(NewYork:AppletonCenturyCrofts,Inc.,1938),Skinnerremarksthat
"althoughaconditionedoperantistheresultofthecorrelationoftheresponsewithaparticular
reinforcement,arelationbetweenitandadiscriminativestimulusactingpriortotheresponseisthealmost
universalrule"(17879).Evenemittedbehaviorisheldtobeproducedbysomesortof"originatingforce"
(51)which,inthecaseofoperantbehaviorisnotunderexperimentalcontrol.Thedistinctionbetween
elicitingstimuli,discriminatedstimuli,and"originatingforces"hasneverbeenadequatelyclarifiedand
becomesevenmoreconfusingwhenprivateinternaleventsareconsideredtobediscriminatedstimuli(see
below).
3Inafamousexperiment,chimpanzeesweretaughttoperformcomplextaskstoreceivetokenswhichhad
becomesecondaryreinforcersbecauseofassociationwithfood.Theideathatmoney,approval,prestige,etc.
actuallyacquiretheirmotivatingeffectsonhumanbehavioraccordingtothisparadigmisunproved,andnot
particularlyplausible.Manypsychologistswithinthebehavioristmovementarequiteskepticalaboutthis(cf.
23n).Asinthecaseofmostaspectsofhumanbehavior,theevidenceaboutsecondaryreinforcementisso
fragmentary,conflicting,andcomplexthatalmostanyviewcanfindsomesupport.
4Skinner'sremarkquotedaboveaboutthegeneralityofhisbasicresultsmustbeunderstoodinthelightof
theexperimentallimitationshehasimposed.Ifitweretrueinanydeepsensethatthebasicprocessesin
languagearewellunderstoodandfreeofspeciesrestriction,itwouldbeextremelyoddthatlanguageis
limitedtoman.Withtheexceptionofafewscatteredobservations(cf.hisarticle,"ACaseHistoryin
ScientificMethod,"TheAmericanPsychologist,11[1956]22133),Skinnerisapparentlybasingthisclaim
onthefactthatqualitativelysimilarresultsareobtainedwithbarpressingofratsandpeckingofpigeons
underspecialconditionsofdeprivationandvariousschedulesofreinforcement.Oneimmediatelyquestions
howmuchcanbebasedonthesefacts,whichareinpartatleastanartifacttraceabletoexperimentaldesign
andthedefinitionofstimulusandresponseintermsofsmoothdynamiccurves(seebelow).Thedangers
inherentinanyattempttoextrapolatetocomplexbehaviorfromthestudyofsuchsimpleresponsesasbar
pressingshouldbeobviousandhaveoftenbeencommentedon(cf.,e.g.,Harlow,op.cit.).Thegeneralityof
eventhesimplestresultsisopentoseriousquestion.Cf.inthisconnectionM.E.Bitterman,J.Wodinsky,and
D.K.Candland,"SomeComparativePsychology,"Am.Jour.ofPsych.,71(1958),94110,whereitisshown
http://cogprints.org/1148/1/chomsky.htm

18/24

08/09/2016

BadNews:NoamChomsky

thatthereareimportantqualitativedifferencesinsolutionofcomparableelementaryproblemsbyratsand
fish.
5Ananalogousargument,inconnectionwithadifferentaspectofSkinner'sthinking,isgivenbyM.Scriven
in"AStudyofRadicalBehaviorism,"Univ.ofMinn.StudiesinPhilosophyofScience,I.Cf.Verplanck's
contributiontoModernLearningTheory,op.cit.pp.28388,formoregeneraldiscussionofthedifficulties
informulatinganadequatedefinitionofstimulusandresponse.Heconcludes,quitecorrectly,thatin
Skinner'ssenseoftheword,stimuliarenotobjectivelyidentifiableindependentlyoftheresultingbehavior,
noraretheymanipulable.VerplanckpresentsacleardiscussionofmanyotheraspectsofSkinner'ssystem,
commentingontheuntestabilityofmanyofthesocalled"lawsofbehavior"andthelimitedscopeofmany
oftheothers,andthearbitraryandobscurecharacterofSkinner'snotionoflawfulrelationand,atthesame
time,notingtheimportanceoftheexperimentaldatathatSkinnerhasaccumulated.
6InBehaviorofOrganisms,Skinnerapparentlywaswillingtoacceptthisconsequence.Heinsists(4142)
thatthetermsofcasualdescriptioninthepopularvocabularyarenotvalidlydescriptiveuntilthedefining
propertiesofstimulusandresponsearespecified,thecorrelationisdemonstratedexperimentally,andthe
dynamicchangesinitareshowntobelawful.Thus,indescribingachildashidingfromadog,"itwillnotbe
enoughtodignifythepopularvocabularybyappealingtoessentialpropertiesofdognessorhidingnessandto
supposethemintuitivelyknown."ButthisisexactlywhatSkinnerdoesinthebookunderreview,aswewill
seedirectly.
7253f.andelsewhere,repeatedly.Asanexampleofhowwellwecancontrolbehaviorusingthenotions
developedinthisbook,Skinnershowsherehowhewouldgoaboutevokingtheresponsepencil.Themost
effectiveway,hesuggests,istosaytothesubject,"Pleasesaypencil"(ourchanceswould,presumably,be
evenfurtherimprovedbyuseof"aversivestimulation,"e.g.,holdingaguntohishead).Wecanalso"make
surethatnopencilorwritinginstrumentisavailable,thenhandoursubjectapadofpaperappropriateto
pencilsketching,andofferhimahandsomerewardforarecognizablepictureofacat."Itwouldalsobe
usefultohavevoicessayingpencilorpenand...inthebackgroundsignsreadingpencilorpenand...orto
placea"largeandunusualpencilinanunusualplaceclearlyinsight.""Undersuchcircumstances,itis
highlyprobablethatoursubjectwillsaypencil.""Theavailabletechniquesareallillustratedinthissample."
Thesecontributionsofbehaviortheorytothepracticalcontrolofhumanbehaviorareamplyillustrated
elsewhereinthebook,aswhenSkinnershows(11314)howwecanevoketheresponsered(thedevice
suggestedistoholdaredobjectbeforethesubjectandsay,"Tellmewhatcolorthisis").
Infairness,itmustbementionedthattherearecertainnontrivialapplicationsofoperantconditioningtothe
controlofhumanbehavior.Awidevarietyofexperimentshaveshownthatthenumberofpluralnouns(for
example)producedbyasubjectwillincreaseiftheexperimentersays"right"or"good"whenoneis
produced(similarly,positiveattitudesonacertainissue,storieswithparticularcontent,etc.cf.L.Krasner,
"StudiesoftheConditioningofVerbalBehavior,"Psych.Bull.,55[1958],forasurveyofseveraldozen
experimentsofthiskind,mostlywithpositiveresults).Itisofsomeinterestthatthesubjectisusually
unawareoftheprocess.Justwhatinsightthisgivesintonormalverbalbehaviorisnotobvious.Nevertheless,
itisanexampleofpositiveandnottotallyexpectedresultsusingtheSkinnerianparadigm.
8"AreTheoriesofLearningNecessary?",Psych.Rev.,57(1950),193216.
9Andelsewhere.Inhispaper"AreTheoriesofLearningNecessary?"Skinnerconsiderstheproblemhowto
extendhisanalysisofbehaviortoexperimentalsituationsinwhichitisimpossibletoobservefrequencies,
rateofresponsebeingtheonlyvaliddatum.Hisansweristhat"thenotionofprobabilityisusually
extrapolatedtocasesinwhichafrequencyanalysiscannotbecarriedout.Inthefieldofbehaviorwearrange
asituationinwhichfrequenciesareavailableasdata,butweusethenotionofprobabilityinanalyzingor
formulatinginstancesofeventypesofbehaviorwhicharenotsusceptibletothisanalysis"(199).Thereare,
ofcourse,conceptionsofprobabilitynotbaseddirectlyonfrequency,butIdonotseehowanyoftheseapply
tothecasesthatSkinnerhasinmind.Iseenowayofinterpretingthequotedpassageotherthanassignifying
anintentiontousethewordprobabilityindescribingbehaviorquiteindependentlyofwhetherthenotionof
probabilityisatallrelevant.
10Fortunately,"InEnglishthispresentsnogreatdifficulty"since,forexample,"relativepitchlevels...are
not...important"(25).Noreferenceismadetothenumerousstudiesofthefunctionofrelativepitchlevels
http://cogprints.org/1148/1/chomsky.htm

19/24

08/09/2016

BadNews:NoamChomsky

andotherintonationalfeaturesinEnglish.
11Thevaguenessofthewordtendency,asopposedtofrequency,savesthelatterquotationfromtheobvious
incorrectnessoftheformer.Nevertheless,agooddealofstretchingisnecessary.Iftendencyhasanythinglike
itsordinarymeaning,theremarkisclearlyfalse.OnemaybelievestronglytheassertionthatJupiterhasfour
moons,thatmanyofSophocles'playshavebeenirretrievablylost,thattheearthwillburntoacrispinten
millionyears,andsoon,withoutexperiencingtheslightesttendencytoactupontheseverbalstimuli.We
may,ofcourse,turnSkinner'sassertionintoaveryunilluminatingtruthbydefining"tendencytoact"to
includetendenciestoanswerquestionsincertainways,undermotivationtosaywhatonebelievesistrue.
12Oneshouldadd,however,thatitisingeneralnotthestimulusassuchthatisreinforcing,butthestimulus
inaparticularsituationalcontext.Dependingonexperimentalarrangement,aparticularphysicaleventor
objectmaybereinforcing,punishing,orunnoticed.BecauseSkinnerlimitshimselftoaparticular,very
simpleexperimentalarrangement,itisnotnecessaryforhimtoaddthisqualification,whichwouldnotbeat
alleasytoformulateprecisely.Butitisofcoursenecessaryifheexpectstoextendhisdescriptivesystemto
behavioringeneral.
13Thishasbeenfrequentlynoted.
14See,forexample,"AreTheoriesofLearningNecessary?",op.cit.,p.199.Elsewhere,hesuggeststhatthe
termlearningberestrictedtocomplexsituations,butthesearenotcharacterized.
15"Achildacquiresverbalbehaviorwhenrelativelyunpatternedvocalizations,selectivelyreinforced,
graduallyassumeformswhichproduceappropriateconsequencesinagivenverbalcommunity"(31).
"Differentialreinforcementshapesupallverbalforms,andwhenapriorstimulusentersintothecontingency,
reinforcementisresponsibleforitsresultingcontrol....Theavailabilityofbehavior,itsprobabilityor
strength,dependsonwhetherreinforcementscontinueineffectandaccordingtowhatschedules"(2034)
elsewhere,frequently.
16Talkofschedulesofreinforcementhereisentirelypointless.Howarewetodecide,forexample,
accordingtowhatschedulescovertreinforcementisarranged,asinthinkingorverbalfantasy,orwhatthe
schedulingisofsuchfactorsassilence,speech,andappropriatefuturereactionstocommunicated
information?
17See,forexample,N.E.MillerandJ.Dollard,SocialLearningandImitation(NewYork,1941),pp.82
83,foradiscussionofthe"meticuloustraining"thattheyseemtoconsidernecessaryforachildtolearnthe
meaningsofwordsandsyntacticpatterns.ThesamenotionisimplicitinO.H.Mowrer'sspeculativeaccount
ofhowlanguagemightbeacquired,inLearningTheoryandPersonalityDynamics,(NewYork:TheRonald
Press,Inc.,1950),Chap.23.Actually,theviewappearstobequitegeneral.
18Forageneralreviewandanalysisofthisliterature,seeD.L.Thistlethwaite,"ACriticalReviewofLatent
LearningandRelatedExperiments,"Psych.Bull.,48(1951),97129.K.MacCorquodaleandP.E.Meehl,in
theircontributiontoModernLearningTheoryop.cit.,carryoutaseriousandconsideredattempttohandle
thelatentlearningmaterialfromthestandpointofdrivereductiontheory,with(astheypointout)notentirely
satisfactoryresults.W.H.Thorpereviewstheliteraturefromthestandpointoftheethologist,addingalso
materialonhomingandtopographicalorientation(LearningandInstinctinAnimals[Cambridge,1956]).
19TheoriesofLearning,214(1956).
20O.E.Berlyne,"NoveltyandCuriosityasDeterminantsofExploratoryBehavior,"Brit.Jour.ofPsych.,41
(1950),6880id.,"PerceptualCuriosityintheRat,"Jour.ofComp.Physiol.Psych.,48(1955),23846W.
R.ThompsonandL.M.Solomon,"SpontaneousPatternDiscriminationintheRat,"ibid.,47(1954),1047.
21K.C.Montgomery,"TheRoleoftheExploratoryDriveinLearning,"ibid.pp.6063.Manyotherpapers
inthesamejournalaredesignedtoshowthatexploratorybehaviorisarelativelyindependentprimary
"drive"arousedbynovelexternalstimulation.
22R.A.Butler,"DiscriminationLearningbyRhesusMonkeystoVisualExplorationMotivation,"ibid.,46
(1953),9598.Laterexperimentsshowedthatthis"drive"ishighlypersistent,asopposedtoderiveddrives
http://cogprints.org/1148/1/chomsky.htm

20/24

08/09/2016

BadNews:NoamChomsky

whichrapidlyextinguish.
23H.F.Harlow,M.K.Harlow,andD.R.Meyer,"LearningMotivatedbyaManipulationDrive,"Jour.Exp.
Psych.,40(1950),22834,andlaterinvestigationsinitiatedbyHarlow.Harlowhasbeenparticularlyinsistent
onmaintainingtheinadequacyofphysiologicallybaseddrivesandhomeostaticneedstatesforexplainingthe
persistenceofmotivationandrapidityoflearninginprimates.Hepointsout,inmanypapers,thatcuriosity,
play,exploration,andmanipulationare,forprimates,oftenmorepotentdrivesthanhungerandthelike,and
thattheyshownoneofthecharacteristicsofacquireddrives.Hebbalsopresentsbehavioralandsupporting
neurologicalevidenceinsupportoftheviewthatinhigheranimalsthereisapositiveattractioninwork,risk,
puzzle,intellectualactivity,mildfearandfrustration,andsoon."DrivesandtheCNS,"Psych.Rev.,62
[1955],24354.)Heconcludesthat"weneednotworkouttortuousandimprobablewaystoexplainwhymen
workformoney,whychildrenlearnwithoutpain,whypeopledislikedoingnothing."Inabriefnote"Early
RecognitionoftheManipulativeDriveinMonkeys,"BritishJournalofAnimalBehaviour,3[1955],7172),
W.Denniscallsattentiontothefactthatearlyinvestigators(G.J.Romanes,1882E.L.Thorndike,1901),
whose"perceptionwasrelativelyunaffectedbylearningtheory,didnotetheintrinsicallymotivatedbehavior
ofmonkeys,"although,heasserts,nosimilarobservationsonmonkeyshavebeenmadeuntilHarlow's
experiments.HequotesRomanes(AnimalIntelligence[1882])assayingthat"muchthemoststrikingfeature
inthepsychologyofthisanimal,andtheonewhichisleastlikeanythingmetwithinotheranimals,wasthe
tirelessspiritofinvestigation."Analogousdevelopments,inwhichgenuinediscoverieshaveblinded
systematicinvestigatorstotheimportantinsightsofearlierwork,areeasilyfoundwithinrecentstructural
linguisticsaswell.
24Thus,J.S.Brown,incommentingonapaperofHarlow'sinCurrentTheoryandResearchinMotivation
(Lincoln:Univ.ofNebraskaPress,1953),arguesthat"inprobablyeveryinstance[oftheexperimentscited
byHarlow]aningeniousdrivereductiontheoristcouldfindsomefragmentoffear,insecurity,frustration,or
whatever,thathecouldinsistwasreducedandhencewasreinforcing"(53).Thesamesortofthingcouldbe
saidfortheingeniousphlogistonorethertheorist.
25Cf.H.G.BirchandM.E.Bitterman,"ReinforcementandLearning:TheprocessofSensoryIntegration,"
Psych.Rev.,56(1949),292308.
26See,forexample,hispaper"APhysiologicalStudyofReward"inD.C.McClelland,ed.,Studiesin
Motivation(NewYork:AppletonCenturyCrafts,Inc.,1955),pp.13443.
27SeeThorpe,op.cit.,particularlypp.11518and33776,foranexcellentdiscussionofthisphenomenon,
whichhasbeenbroughttoprominenceparticularlybytheworkofK.Lorenz(cf."DerKumpaninder
UmweltdesVogels,"partsofwhicharereprintedinEnglishtranslationinC.M.Schiller,ed.,Instinctive
Behavior[NewYork:InternationalUniversitiesPress,1957],pp.83128).
28Op.cit.,p.372.
29See,e.g.,J.Jaynes,"Imprinting:InteractionofLearnedandInnateBehavior,"Jour.ofComp.Physiol.
Psych.,49(1956),2016,wheretheconclusionisreachedthat"theexperimentsprovethatwithoutany
observablereward,youngbirdsofthisspeciesfollowamovingstimulusobjectandveryrapidlycometo
preferthatobjecttoothers."
30Ofcourse,itisperfectlypossibletoincorporatethisfactwithintheSkinnerianframework.If,for
example,achildwatchesanadultusingacombandthen,withnoinstruction,triestocombhisownhair,we
canexplainthisactbysayingthatheperformsitbecausehefindsitreinforcingtodoso,orbecauseofthe
reinforcementprovidedbybehavinglikeapersonwhois"reinforcing"(cf.164).Similarly,anautomatic
explanationisavailableforanyotherbehavior.ItseemsstrangeatfirstthatSkinnerpayssolittleattentionto
theliteratureonlatentlearningandrelatedtopics,consideringthetremendousreliancethatheplacesonthe
notionofreinforcementIhaveseennoreferencetoitinhiswritings.Similarly,F.S.KellerandW.N.
Schoenfeld,inwhatappearstobetheonlytextwrittenunderpredominantlySkinnerianinfluence,Principles
ofPsychology(NewYork:AppletonCenturyCrofts,Inc.,1950),dismissthelatentlearningliteratureinone
sentenceas"besidethepoint,"servingonly"toobscure,ratherthanclarify,afundamentalprinciple"(thelaw
ofeffect,41).However,thisneglectisperfectlyappropriateinSkinner'scase.Tothedrivereductionist,or
anyoneelseforwhomthenotionreinforcementhassomesubstantivemeaning,theseexperimentsand
http://cogprints.org/1148/1/chomsky.htm

21/24

08/09/2016

BadNews:NoamChomsky

observationsareimportant(andoftenembarrassing).ButintheSkinneriansenseoftheword,neitherthese
resultsnoranyconceivableotherscancastanydoubtontheclaimthatreinforcementisessentialforthe
acquisitionandmaintenanceofbehavior.Behaviorcertainlyhassomeconcomitantcircumstances,and
whatevertheyare,wecancallthemreinforcement.
31Tinbergen,op.cit.,Chap.VI,reviewssomeaspectsofthisproblem,discussingtheprimaryroleof
maturationinthedevelopmentofmanycomplexmotorpatterns(e.g.,flying,swimming)inlowerorganisms,
andtheeffectofan"innatedispositiontolearn"incertainspecificwaysandatcertainspecifictimes.Cf.also
P.Schiller,"InnateMotorActionasaBasisforLearning,"inC.H.Schiller,ed.,InstinctiveBehavior(New
York:InternationalUniversitiesPress,1957),pp.26588,foradiscussionoftheroleofmaturingmotor
patternsinapparentlyinsightfulbehaviorinthechimpanzee.
Lenneberg("TheCapacityforLanguageAcquisition",inJ.A.Fodor,ed.,TheStructureofLanguage
[PrenticeHall,Inc.,1964])presentsaveryinterestingdiscussionofthepartthatbiologicalstructuremay
playintheacquisitionoflanguage,andthedangersinneglectingthispossibility.
32FromamongmanycitedbyTinbergen,op.cit.,p.85.
33Cf.K.S.Lashley,"InSearchoftheEngram,"SymposiumoftheSocietyforExperimentalBiology,4
(1950),45482.R.Sperry,"OntheNeuralBasisoftheConditionedResponse,"BritishJournalofAnimal
Behavior,3(1955),4144,arguesthattoaccountfortheexperimentalresultsofLashleyandothers,andfor
otherfactsthathecites,itisnecessarytoassumethathighlevelcerebralactivityofthetypeofinsight,
expectancy,andsoonisinvolvedeveninsimpleconditioning.Hestatesthat"westilllacktodaya
satisfactorypictureoftheunderlyingneuralmechanism"oftheconditionedresponse.
34Furthermore,themotivationofthespeakerdoesnot,exceptinthesimplestcases,correspondinintensity
tothedurationofdeprivation.AnobviouscounterexampleiswhatHebbhascalledthe"saltednut
phenomenon"(OrganizationofBehavior[NewYork,1949],p.199).Thedifficultyisofcourseevenmore
seriouswhenweconsiderdeprivationsnotrelatedtophysiologicaldrives.
35Justashemayhavetheappropriatereaction,bothemotionalandbehavioral,tosuchutterancesasthe
volcanoiseruptingorthere'sahomicidalmaniacinthenextroomwithoutanypreviouspairingoftheverbal
andthephysicalstimulus.Skinner'sdiscussionofPavlovianconditioninginlanguage(154)issimilarly
unconvincing.
36J.S.Mill,ASystemofLogic(1843).R.Carnapgivesarecentreformulationin"MeaningandSynonymy
inNaturalLanguages,"Phil.Studies,6(1955),3347,definingthemeaning(intension)ofapredicateQfora
speakerXas"thegeneralconditionwhichanobjectymustfulfilinorderforXtobewillingtoascribethe
predicateQtoy."Theconnotationofanexpressionisoftensaidtoconstituteits"cognitivemeaning"as
opposedtoits"emotivemeaning,"whichis,essentially,theemotionalreactiontotheexpression.
Whetherornotthisisthebestwaytoapproachmeaning,itisclearthatdenotation,cognitivemeaning,and
emotivemeaningarequitedifferentthings.Thedifferencesareoftenobscuredinempiricalstudiesof
meaning,withmuchconsequentconfusion.Thus,Osgoodhassethimselfthetaskofaccountingforthefact
thatastimuluscomestobeasignforanotherstimulus(abuzzerbecomesasignforfood,awordforathing,
etc.).Thisisclearly(forlinguisticsigns)aproblemofdenotation.Themethodthatheactuallydevelopsfor
quantifyingandmeasuringmeaning(cf.C.E.Osgood,G.Suci,P.Tannenbaum,TheMeasurementof
Meaning[Urbana:Univ.ofIllinoisPress,1957])applies,however,onlytoemotivemeaning.Suppose,for
example,thatAhatesbothHitlerandscienceintensely,andconsidersbothhighlypotentand"active,"while
B,agreeingwithAaboutHitler,likesscienceverymuch,althoughheconsidersitratherineffectiveandnot
tooimportant.Then,Amayassignto"Hitler"and"science"thesamepositiononthesemanticdifferential,
whileBwillassign"Hitler"thesamepositionasAdid,but"science"atotallydifferentposition.Yet,Adoes
notthinkthat"Hitler"and"science"aresynonymousorthattheyhavethesamereference,andAandBmay
agreepreciselyonthecognitivemeaningof"science."Clearly,itistheattitudetowardthethings(the
emotivemeaningofthewords)thatisbeingmeasuredhere.ThereisagradualshiftinOsgood'saccount
fromdenotationtocognitivemeaningtoemotivemeaning.Theconfusioniscaused,nodoubt,bythefact
thatthetermmeaningisusedinallthreesenses(andothers).[SeeJ.Carroll'sreviewofthebookbyOsgood,
Suci,andTannenbauminLanguage,35,No.1(1959).]
http://cogprints.org/1148/1/chomsky.htm

22/24

08/09/2016

BadNews:NoamChomsky

37MostclearlybyQuine.SeeFromaLogicalPointofView(Cambridge,1953),especiallyChaps.2,3,and
7.
38AmethodforcharacterizingsynonymyintermsofreferenceissuggestedbyGoodman,"OnLikenessof
Meaning,"Analysis,10(1949),17.DifficultiesarediscussedbyGoodman,"OnSomeDifferencesabout
Meaning,"ibid.,13(1953)9096.Carnap,op.cit.,presentsaverysimilaridea(Section6),butsomewhat
misleadinglyphrased,sincehedoesnotbringoutthefactthatonlyextensional(referential)notionsarebeing
used.
39Ingeneral,theexamplesdiscussedherearebadlyhandled,andthesuccessoftheproposedanalysesis
overstated.Ineachcase,itiseasytoseethattheproposedanalysis,whichusuallyhasanairofobjectivity,is
notequivalenttotheanalyzedexpression.Totakejustoneexample,theresponseIamlookingformyglasses
iscertainlynotequivalenttotheproposedparaphrases:"WhenIhavebehavedinthiswayinthepast,Ihave
foundmyglassesandhavethenstoppedbehavinginthisway,"or"CircumstanceshaveariseninwhichIam
inclinedtoemitanybehaviorwhichinthepasthasledtothediscoveryofmyglassessuchbehaviorincludes
thebehavioroflookinginwhichIamnowengaged."Onemaylookforone'sglassesforthefirsttimeorone
mayemitthesamebehaviorinlookingforone'sglassesasinlookingforone'swatch,inwhichcaseIam
lookingformyglassesandIamlookingformywatchareequivalent,undertheSkinnerianparaphrase.The
difficultquestionsofpurposivenesscannotbehandledinthissuperficialmanner.
40Skinnertakesgreatpains,however,todenytheexistenceinhumanbeings(orparrots)ofanyinnate
facultyortendencytoimitate.Hisonlyargumentisthatnoonewouldsuggestaninnatetendencytoread,yet
readingandechoicbehaviorhavesimilar"dynamicproperties."Thissimilarity,however,simplyindicates
thegrossnessofhisdescriptivecategories.Inthecaseofparrots,Skinnerclaimsthattheyhavenoinstinctive
capacitytoimitate,butonlytobereinforcedbysuccessfulimitation(59).GivenSkinner'suseoftheword
reinforcement,itisdifficulttoperceiveanydistinctionhere,sinceexactlythesamethingcouldbesaidofany
otherinstinctivebehavior.Forexample,whereanotherscientistwouldsaythatacertainbirdinstinctively
buildsanestinacertainway,wecouldsayinSkinner'sterminology(equivalently)thatthebirdis
instinctivelyreinforcedbybuildingthenestinthisway.Oneisthereforeinclinedtodismissthisclaimas
anotherritualintroductionofthewordreinforce.Thoughtheremay,undersomesuitableclarification,be
sometruthinit,itisdifficulttoseehowmanyofthecasesreportedbycompetentobserverscanbehandledif
reinforcementisgivensomesubstantivemeaning.Cf.Thorpe,op.cit.p.353f.K.Lorenz,KingSolomon's
Ring(NewYork,1952),pp.8588evenMowrer,whotriestoshowhowimitationmightdevelopthrough
secondaryreinforcement,citesacase,op.cit.,p.694,whichheapparentlybelieves,butwherethiscould
hardlybetrue.Inyoungchildren,itseemsmostimplausibletoexplainimitationintermsofsecondary
reinforcement.
41Althougheventhispossibilityislimited.Ifweweretotaketheseparadigminstancesseriously,itshould
followthatachildwhoknowshowtocountfromoneto100couldlearnanarbitrary10x10matrixwith
thesenumbersasentriesasreadilyasthemultiplicationtable.
42Similarly,"theuniversalityofaliteraryworkreferstothenumberofpotentialreadersinclinedtosaythe
samething"(275i.e.,themost"universal"workisadictionaryofclichsandgreetings)aspeakeris
"stimulating"ifhesayswhatweareabouttosayourselves(272)etc.
43Similarly,considerSkinner'scontention(36265)thatcommunicationofknowledgeorfactsisjustthe
processofmakinganewresponseavailabletothespeaker.Heretheanalogytoanimalexperimentsis
particularlyweak.Whenwetrainarattocarryoutsomepeculiaract,itmakessensetoconsiderthisamatter
ofaddingaresponsetohisrepertoire.Inthecaseofhumancommunication,however,itisverydifficultto
attachanymeaningtothisterminology.IfAimpartstoBtheinformation(newtoB)thattherailroadsface
collapse,inwhatsensecantheresponseTherailroadsfacecollapsebesaidtobenow,butnotpreviously,
availabletoB?SurelyBcouldhavesaiditbefore(notknowingwhetheritwastrue),andknownthatitwasa
sentence(asopposedtoCollapsefacerailroadsthe).Noristhereanyreasontoassumethattheresponsehas
increasedinstrength,whateverthismeansexactly(e.g.,Bmayhavenointerestinthefact,orhemaywantit
suppressed).Itisnotclearhowwecancharacterizethisnotionof"makingaresponseavailable"without
reducingSkinner'saccountof"impartingknowledge"toatriviality.

http://cogprints.org/1148/1/chomsky.htm

23/24

08/09/2016

BadNews:NoamChomsky

44(332).Onthenextpage,however,thesinthesameexampleindicatesthat"theobjectdescribedasthe
boypossessesthepropertyofrunning."Thedifficultyofevenmaintainingconsistencywithaconceptual
schemelikethisiseasytoappreciate.
45Onemightjustaswellarguethatexactlytheoppositeistrue.Thestudyofhesitationpauseshasshown
thatthesetendtooccurbeforethelargecategoriesnoun,verb,adjectivethisfindingisusuallydescribed
bythestatementthatthepausesoccurwherethereismaximumuncertaintyorinformation.Insofaras
hesitationindicatesongoingcomposition(ifitdoesatall),itwouldappearthatthe"keyresponses"are
chosenonlyafterthe"grammaticalframe."Cf.C.E.Osgood,unpublishedpaperF.GoldmanEisler,
"SpeechAnalysisandMentalProcesses,"LanguageandSpeech,1(1958),67.
46E.g.,whatareinfacttheactualunitsofverbalbehavior?Underwhatconditionswillaphysicalevent
capturetheattention(beastimulus)orbeareinforcer?Howdowedecidewhatstimuliarein"control"ina
specificcase?Whenarestimuli"similar"?Andsoon.(Itisnotinterestingtobetold,e.g.,thatwesayStopto
anautomobileorbilliardballbecausetheyaresufficientlysimilartoreinforcingpeople[46].)Theuseof
unanalyzednotionslikesimilarandgeneralizationisparticularlydisturbing,sinceitindicatesanapparent
lackofinterestineverysignificantaspectofthelearningortheuseoflanguageinnewsituations.Noonehas
everdoubtedthatinsomesense,languageislearnedbygeneralization,orthatnovelutterancesandsituations
areinsomewaysimilartofamiliarones.Theonlymatterofseriousinterestisthespecific"similarity."
Skinnerhas,apparently,nointerestinthis.KellerandSchoenfeld,op.cit.,proceedtoincorporatethese
notions(whichtheyidentify)intotheirSkinnerian"modernobjectivepsychology"bydefiningtwostimulito
besimilarwhen"wemakethesamesortofresponsetothem"(124butwhenareresponsesofthe"same
sort"?).Theydonotseemtonoticethatthisdefinitionconvertstheir"principleofgeneralization"(116),
underanyreasonableinterpretationofthis,intoatautology.Itisobviousthatsuchadefinitionwillnotbeof
muchhelpinthestudyoflanguagelearningorconstructionofnewresponsesinappropriatesituations.
47"TheProblemofSerialOrderinBehavior,"inL.A.Jeffress,ed.,HixonSymposiumonCerebral
MechanismsinBehavior(NewYork:JohnWiley&SonsInc.,1951).ReprintedinF.A.Beach,D.O.Hebb,
C.T.Morgan,H.W.Nissen,eds.,TheNeuropsychologyofLashley(NewYork:McGrawHillBook
Company,1960).Pagereferencesaretothelatter.
48Thereisnothingessentiallymysteriousaboutthis.Complexinnatebehaviorpatternsandinnate
"tendenciestolearninspecificways"havebeencarefullystudiedinlowerorganisms.Manypsychologists
havebeeninclinedtobelievethatsuchbiologicalstructurewillnothaveanimportanteffectonacquisitionof
complexbehaviorinhigherorganisms,butIhavenotbeenabletofindanyseriousjustificationforthis
attitude.Somerecentstudieshavestressedthenecessityforcarefullyanalyzingthestrategiesavailabletothe
organism,regardedasacomplex"informationprocessingsystem"(cf.J.S.Bruner,J.J.Goodnow,andG.A.
Austin,AStudyofThinking[NewYork,1956]A.Newell,J.C.Shaw,andH.A.Simon,"Elementsofa
TheoryofHumanProblemSolving,"Psych.Rev.,65,[1958],15166),ifanythingsignificantistobesaid
aboutthecharacterofhumanlearning.Thesemaybelargelyinnate,ordevelopedbyearlylearningprocesses
aboutwhichverylittleisyetknown.(ButseeHarlow,"TheFormationofLearningSets,"Psych.Rev.,56
(1949),5165,andmanylaterpapers,wherestrikingshiftsinthecharacteroflearningareshownasaresult
ofearlytrainingalsoD.O.Hebb,OrganizationofBehavior,109ff.).Theyareundoubtedlyquitecomplex.
Cf.Lenneberg,op.cit.,andR.B.Lees,reviewofN.Chomsky'sSyntacticStructuresinLanguage,33(1957),
406f,fordiscussionofthetopicsmentionedinthissection.

http://cogprints.org/1148/1/chomsky.htm

24/24

Vous aimerez peut-être aussi