TERRENCEW.DEACON The Symbolic Species THECO-EVOLUTIONOF LANGUAGEANDTHEBRAIN TerrenceW.Deacon W.W.NORTON&COMPANY~NEWYORKLONDON Copyright 1997 by Terrence W.Deacon Allrights reserved Printed in the United States of America For information about permission to reproduce selections fromthis book, write to Permissions, W.W.Norton &Company,Inc., 500 Fifth Avenue,New York,NY101I0. Thetext of this book iscomposed inNew Caledonia with the display set inTrade Gothic Condensed Bold.Compostion andmanufacturing by theHaddon Craftsmen, Inc. BOOKDESIGNBYBTDIROBINBENTZ. LIBRARYOFCONGRESSCATALOGING-IN-PUBLICATIONDATA Deacon, Terrence William. Thesymbolicspecies:the co-evolution oflanguage and the brain I by Terrence W.Deacon. p.cm. Includesbibliographicalreferences and index. ISBN 0-393-03838-6 1.Neurolinguistics.2.Brain-Evolution. I. Title. QP399.D43 1997 96-31 lIS CIP W.W.Norton &Company, Inc. 500Fifth Avenue,New York,N.Y.lOIlO http://www.wwnorton.com W.W.Norton &Company Ltd. 10 Coptic Street, London WCIAIPU 34567890 Inmerrwry of Harriet Deacon,my grandmother and first mentor; who taught me torecognize the miraculous in everyday things. Very special thanks are due to my family,Cris, Anneka, and John, who sup-ported me evenasIstole precious time away fromthemto complete this work,and tomy parents, whoseencouragement Icould alwayscount on. Thanks are also due to many who have directly inspired,assisted,and en-dured thisproject.These include:JosephMarcus,whoread,edited,and commented extensively on earlier drafts and who has consistently held me to the high standards he expects from a mentor; my own mentors, who have knowinglyor unknowingly contributedtheirinsightsand valuablecriti-cisms,butwhoarefartoonumeroustolist;myformerstudentAlan Sokoloff,whosePh.D.researchonoraltractinnervationunderliesmy thoughts on vocal evolution; David Rudner, Sandra Kleinman, Alan Aronie, and many other friends whose feedback has helped clear away some of the fog;my colleagues inthe lab,whounderstood and picked up more of the load asmy attentions were diverted from the lab bench; Robyn Swierk and JulieCriniere,who helped enter theendlesscorrections,and copyeditor AnnAdelman;Hoovertheseal,whoopenedmyearstothemysteryof speech; and my editor, Hilary Hinzmann, who was patient through my long spells of writer's block and overcommitment, an,d who helped nurse a rough collection of ideas and notes into a narrative.Ican never adequately repay these many gifts. Preface Part One: Language I.THEHUMANPARADOX AnEvolutionary Anomaly Technical Difficulties and HopefulMonsters The Missing Simple Languages 2.A LOSSFORWORDS Gymnastics of theMind InOther Words The Reference Problem 3.SYMBOLSAREN 'T SIMPLE The Hierarchical Nature of Reference The Symbolic Threshold Unlearning anInsight Contents 11 21 28 39 47 51 59 69 79 92 4.OUTSIDETHEBRAIN Chomsky'sHandstand The Other Evolution Emerging Universals Better Learning Through Amnesia Part Two:Brain 5.THESIZEOFI NT EL LIGENCE A GrossMisunderstanding Brains inBodies Thinking Your OwnSize 6.GROWINGAPART The Chihuahua Fallacy Using Fly Genes toMakeHumanBrains The Developmental Clock 7.A DARWINIANELECTRICIAN MusicalChairs Displacement An AlienBrain Transplant Experiment Beyond Phrenology 8.THETALKINGBRAIN Hoover'sBrain VisceralSounds Why Don't Mammals Sing LikeBirds? A Leveraged Takeover 9.SYMBOLMINDS Front-Heavy Making Symbols 10.LOCATINGLANGUAGE Forcing Square Pegs into Round Holes The Illuminated Brain Where Symbols Aren't Taking Sides 102 110 115 122 145 153 159 165 174 187 193 207 214 220 225 230 236 247 254 264 279 288 300 309 Part Three:Co-Evolution 11 .ANDTHEWDRDBECAMEFLESH The Brain That Didn't Evolve Language Adaptations H011W symbolicus The Co-Evolutionary Net The Writing on the Wall 12 .SYMBOLICORIGINS A Passion to Communicate Why Human Societies Shouldn't Work A Symbolic Solution Ritual Beginnings 13 .A SERENDIPITOUSMIND The Fail-Safe Computer The Sphinx No Mind I s an Island 14.SUCHSTUFFASDREAMSAREMADEON Ends To Be or Not toBe:What Isthe Difference? Reinventing theMind Notes Additional Readings Bibliography Index 321 334 340 349 365 376 384 393 401 411 416 423 433 438 455 465 485 489 511 Preface The conviction persists,though history shows it to be a hallucination,that all the questions that the human mind has asked are questions that can be answered in terms of the alternatives that the questions themselves present.But in fact,intellectual progress usually occurs throughsheer abandonment of questions together with both of the alternatives they assume,anabandonment that results from their decreasing vitalism and a change of urgent interest.Wedonot solve them, we get over them. -JohnDewey Anumber of yearsagoIwasgiving a brief talk about the evolutionof the brain when spmeone asked a question I couldn't answer. It was not acomplicatedquestion.It didn'tcomefromacolleague whohad found a weakness in my theory or a graduate student who had read about anew experiment that contradicted my data.It camefroma child inmy son'selementary school class.I had givenschool talksonbeing a scientist before, and I thought I knew what to expect. I never expected an eight-year old to stump me. Iwastalking about brainsand howthey work,and how humanbrains are different, and how this difference isreflected inour unique and com-11 plex mode of communication: language.But when I explained that only hu-mans communicate with language,I struck a dissonant chord. "But don't other animalshave their ownlanguages?" one child asked. This gaveme the opportunity to outline some of the ways that language isspecial:how speech isfarmorerapid and precise than any other com-munication behavior, how the underlying rules forconstructing sentences are so complicated and curious that it's hard to explain how they could ever be learned,and how no other formof animal communication hasthe logi-cal structure and open-ended possibilities that all languages have.But this wasn't enough to satiSfYa mind raised on Walt Disney animal stories. "Do animals just have SIMPLE languages?" my questioner continued. "N 0, apparently not," I explained. "Although other animals communicate with one another, at least within the same species, this communication re-sembleslanguageonlyinaverysuperficialway-forexample,using sounds-but nonethatIknowof hastheequivalentsof suchthingsas words,much less nouns, verbs, and sentences.Not even simple ones." "Why not?" asked another child. AtthisI hesitated.And the moreI thought about it,the more Irecog-nized that I didn't really have an answer. Asfar asI could tell no other sci-entistshad seriously considered the question in thisform.Why are there no simple languages, with simple formsof nouns,verbs,and sentences? It isindeed acounterintuitivefact.Myths,fables,fairytales,animatedcar-toons, and Disney movies portray what common sense suggests ought to be the case: that other animals with their Simpler minds communicate and rea-sonusingsimpler languages thanours.Why isn'tit so? I'm not sure why Ihadn't noticed this paradox before, or why other sci-entists hadn't.Most mammals aren't stupid. Many are capable of quite re-markable learning. Yet they don't communicate with simple languages, nor do they show much of a capacity to learn them-if our pets are any indica-tion.Perhaps we have been too preoccupied with trying to explain our big brains, or too complacent with the metaphoric use of the term animal lan-guage, to notice this contradictory little fact.But the question may also have been unconsciously avoided because of the intellectual costsof consider-ing it seriously.Indeed,the more deeply I have pursued thisquestion,the more it seemslike aPandora'sboxthatunleashestroubling doubtsabout many other questions that once seemed all but settled. This isn't the ques-tionwe had been asking,but maybe it should havebeen.AsDewey sug-gests,thealternativesweposeinourscientificquestionsmaynoteven address the most crucial issues. This book starts with thiscurious question, because it supersedes many t , . ~D ..........,'. _ ____________________________ l of the questions we thought were more important, and because it stubbornly refusesto resolve itself asa side effect of the superiority of human intelli-gence or the savantlike language ability of young children. But in my efforts toanswerit,Iamforcedtoreopenmanyquestionslong thought tohave been resolved,or at least reduced to a fewalternatives whichnow appear lessinformative than we once thought. I n the chapters that follow,I investigate how language differs from other forms of communication, why other species encounter virtually intractable difficultieswhenit comestolearning evensimplelanguage,how human brain structure has evolved to overcome these difficulties, and what forces andconditionsinitiatedandsteeredusalongthisunprecedentedevolu-tionary course.What resultsisa detailed reappraisal of humanbrainand languageevolutionthatemphasizestheunbrokencontinuitybetween human and nonhuman brains,and yet,at the same time, describes a sin-gular discontinuitybetweenhumanand nonhumanminds,or tobemore precise, between brains that use this form of communication and brains that do not.My somewhat unprecedented approachto these questionsunfolds asa step by step argument, in whicheach chapter builds on the questions, analyses,and evidence provided inprior chapters. At almost every step of the argument, I arrive at different interpretations from what might loosely be called the accepted theories inthe field.SoI suggest approaching this narrative asone might approach a mystery novel, where the order and pre-sentation of cluesare criticaltothe plot,becauseencountering the clues and unexpected conclusions out of context might require some tricky men-tal gymnasticsto discover how they fitback together. The presentation isbrokenup into three major sections.The firstpart of thebook-Language-focuses onthe nature of languageandtherea-sonsthat it isvirtually confined tothe human species. The second part of the book-Brain-tackles the problem of identifYing what is unusual about human brain structure that correspondswiththe unique problemsposed by language. T.he third part of the book-Co-Evolution-examines the pe-culiar extension of natural selection logiCthatisbehind humanbrainand language evolution, and tries to identifY what sort of communication "prob-lem" precipitated the evolution of our unprecedented mode of communi-cation. The book ends with some speculations on the significance of these new findingsfor the understanding of human consciousness. A major intent of the book isto engage the reader in a reexamination of manytacit assumptionsthat liebehind current views.In service of these aimsI have tried tomake mypresentation accessibletothe broadest pos-sible scientific audience and,I hope, to a scientifically interested lay audi-ence. Whenever possibleI havetriedto explaintechnicalpointsinnon-technical terms, and although I have not avoided introducing biological and neuroanatomicalterminology when it isrelevant,I have tried to illustrate some of the more technical pOintsin graphic form.Some readers may find the middle section of the book-Brain-a bit rough going, but I believe that a struggle with thismaterial willbe rewarded by seeing how itleads to the novelreassessment of human origins and human consciousness that I offer inthe more accessible and imaginative last section of the book. Almost everyone who has written on the origins oflanguage recounts with a sense of irony how the SocieteLinguistique de Paris passed a resolution in1866 banning allpapers on the origins of language. Thisresolution was meant tohalta growing flowof purely speculative papers,whichtheSo-cietedeemedtobe contributinglittleof substanceandoccupyingother-wisevaluabletime and resources. A young diSCiplineeager tomodel itself afterothernaturalsciencescouldillaffordtosponsorresearchinatopic that wasalmost entirely without empirical support.But the shadow of sus-picion thatloomsover language origins theories isnot just due to this his-toricalreputation.Languageoriginsscenariosnomoreempirically grounded than their banned predecessors stillabound inthe popular sci-ence literature,and they provideaperennialtopicforcocktail party dis-cussions. What's worse, assumptions about the nature oflanguage and the differencesbetweennonhumanand humanmindsareimplicitinalmost every philosophical and scientific theory concerned with cognition, knowl-edge, or human social behavior.It istruly a multidisciplinary problem that defiesanalysisfromanyone perspectivealone,and where thebreadth of technical topics that must be mastered exceeds even the most erudite schol-ars' capabilities. So it ishard tooverestimatetheimmensity of the task or the risksof superficial analysis, and it isunlikely that anyone account can hope to achieve anything close to a comprehensive treatment of the prob-lem. I take this as a serious caution to my own ambitions, and must admit from the outset thatthe depth of coverageand degree of expertise investedin the topics considered in this book clearly reflect my own intellectual biases, drawing primarily onmy training in the neurosciences and in evolutionary anthropology and supplemented by a dilettante'straining inother impor-tantareas.Consequently,the bookfocusesonthe variousimplicationsof onlythisonehuman/nonhumandifferenceinmentalabilities-particu-larlyneurologicalimplications-andignores ,manyotheraspectsof the brain-language relationship. I havenot attempted in any systematic way to review or compare the many alternative explanations proposed for the phe-1A ..D r a f ~ , . a nomena Iconsider,andI have only discussed specific alternative theories asthey serve ascounterpointsto makemyownapproachclearer.Anex-haustivereview of competing explanations would require another book at leastaslong asthisone.I apologize to my many scientific colleagues, who alsohave laboredovertheseissues,formakingthisarather personalex-ploration that does not do full justice to other theories, and does not explain why I do not evenmentionmany of them.My own contributions are only possiblebecauseof thelaborsof untolddozensof previousresearchers whose work has informed and influenced my own, and to whose contribu-tionsIhave added only a handful of new findings.For those interested, I havetried to provide areference to other approaches to these same puz-zlesin the end notes. In what follows,it may appear asthough I am a scientist with a naturally rebelliousnature.Imust admit thatIhaveanattractionto heresies,and that my sympathies naturally tend to be withthe cranksand doubters and againstwell-establisheddoctrines.But thisisnot becauseIenjoy contro-versy.Rather, it isbecause, like Dewey, I believe that the search for knowl-edge isasoftenimpeded by faultyassumptionsand by a limited creative visionfor alternatives asby a lack of necessary tools or critical evidence. So I will have achieved my intent if,in the process of recounting my thoughts on this mystery,I leave a few unquestioned assumptions more questionable, makesomecounter-intuitivealternativesmoreplausible,andprovidea new vantage point from whichto reflect upon human uniqueness. The Symbolic Species PARTONE Language CHAPTERONE TheHuman Paradox ... the paradoxisthe source of the thinker'spassion,and thethinker without a paradox islike a lover without feeling:a paltry mediocrity. -S_ren Kierkergaard An Evolutionary Anomaly As our speciesdesignation-sapiens-suggests,the defining attribute of human beings is anunparalleled cognitive ability. We think differ-entlyfromallothercreaturesonearth,andwecansharethose thoughts with one another in waysthat no other species even approaches. In comparison, the rest of our biology is almost incidental. Hundreds of mil-lions of years of evolution have produced hundreds of thousands of species with brains, and tens of thousands with complex behavioral, perceptual, and learning abilities.Only one of these hasever wondered about itsplace in the world, because only one evolved the ability to do so. Though we share the same earth with millions of kinds ofliving creatures, we also live in a world that no other species has access to. We inhabit a world 21 fullof abstractions,impossibilities,and paradoxes.Wealonebrood about what didn'thappen,and spend alargepart of each daymusing about the waythingscouldhavebeenif eventshadtranspireddifferently.Andwe alone ponder what it will be like not to be. Inwhat other species could in-dividuals ever be troubled by the factthat they do not recall the way things were beforethey were born and willnotknow what willoccur after they die? We tell stories about our real experiences and invent stories about imag-ined ones,and we even makeuse of these stories to organize our lives.In a real sense, we live our lives in this shared virtual world.And slowly,over the millennia, we have come to realize that no other species on earth seems able to followus into thismiraculousplace. Weareallfamiliarwiththisfacetof our lives,but how,youmight ask, could Ifeelso confident that it is not part of the mental experience of other species-so surethatthey do not sharethesekindsof thoughtsandcon-cerns-when they cannot be queried about them? That's just it!My answer, which willbe argued in detail inthe follOwingchapters,has everything to do withlanguageand the absence of it inother species.The doorway into thisvirtual world wasopened to usalone by the evolution of language,be-causelanguageisnotmerely amodeof communication,it isalsothe out-ward expression of anunusualmode of thought-symbolic representation. Without symbolization the entire virtual world that Ihave described isout of reach: inconceivable.My extravagant claimto know what other species cannot know rests on evidence that symbolic thought does not come innately built in,but developsby internalizing the symbolic processthatunderlies language. So species that have not acquired the ability to communicate sym-bolically cannot have acquired the ability to think this way either. The way that language represents objects, events, and relationships pro-vides a uniquely powerful economy of reference.It offers a means for gen-eratinganessentiallyinfinitevarietyof novelrepresentations,andan unprecedented inferentialengine forpredicting events, organizing mem-ories, and planning behaviors.It entirely shapes our thinking and the ways we know the physical world. It is so pervasive and inseparable from human intelligence ingeneral that it isdifficult to distinguish what aspectsof the human intellect have not been molded and streamlined by it.To explain this differenceanddescribetheevolutionarycircumstancesthatbroughtit about are the ultimate challenges inthe study of human origins. Thequestionthatultimatelymotivatesaperennialfascinationwith human origins is not who were our ancestors, or how they came to walk up-right,or evenhow they discovered the use of stone tools. It isnotreally a questionthat has a paleontological answer.It isa question that might oth-erwisebeaskedof psychologistsorneurologistsorevenphilosophers. Where dohumanminds come from?The missing link that we hope to fill inby investigating human origins isnot somuch agapinour familytree, but agapthat separatesusfromother speciesingeneral.Knowing how something originated often isthe best clue to how it works.And we know that human consciousnesshad abeginning.Those featuresof our mental abilities that distinguish usfrom allother species arose within the handful of millionyearssincewesharedacommonancestor withtheremaining African apes,and probably canmostly be traced to events that took place only withinthe last 2million.It wasaRubiconthat wascrossed at a defi-nite time and in a specific evolutionary context.If we could identifY what wasdifferent oneither side of this divide--differences inecology,behav-ior, anatomy, especially neuroanatomy-perhaps we would find the critical change that catapulted us into this unprecedented world full of abstractions, stories,and impossibilities,that we callhuman. It isnot just the origins of our biological species that we seek to explain, but the origin of our novelformof mind.Biologically, weare just another ape.Mentally,we are a new phylum of organisms.Inthese two seemingly incommensuratefactsliesa conundrum thatmust be resolved before we havean adequate explanation of what itmeansto be human. Advancesinthestudyof humanevolution,thebrain,andlanguage processes have led many scientists confidently to claimto be closing in on the finalcluesto this mystery.How close are we? Many lines of evidence seem to be converging on an answer.Withrespect to our ancestry, the re-maining gaps in the fossil evidence of our prehistory are being rapidly fIlled in.Withinthe last fewdecades a remarkably richpicture of the sizesand shapes of fossilhominid bodies and brains has emerged.It isprobably fair to say that at least with respect to the critical changes that distinguish us in this wayfromother apes,there are few missing links yet to be found, just particulars to be filled in. That crucial phase in hominid evolution when our ancestors' brains began to diverge in relative size fromother apes' brains is well bracketed by fossils that span the range. Asfor the inside story, the neu-rosciences are providing powerful new tools with which it has become pos-sibletoobtaindetailedimagesfromworkinghumanbrainsperforming language tasks, or to investigate the processes that build our brains during developmentanddistinguishthebrainsof differentspecies,orevento modelneural processesoutside of brains.Finally,linguists'analysesof the lOgicalstructure of languages,their diversity and recent ancestry, and the pattems that characterize their development inchildrenhaveproVideda wealth of information about just what needs to be explained, and compar-ative studies of animals' communications in the wild and their languagelike capacities inthe laboratOlY havehelped toframethesequestions withex-plicit examples. Despite allthese advances,some critical pieces of the puzzle still elude us. Even though neural science haspried ever deeper into the mysteriesof brainfunction,westilllacka theory of globalbrain functions.Weunder-stand many of the cellular and molecular details,we have mapped a num-berofcognitivetaskstoassociatedbrainregions,andweevenhave constructed computer simulations of networks that operate in ways that are vaguely like parts of brains; but we stilllack insight into the general logic that ties such details together. On the whole,most neuroscientists take the prudent perspective that onlybycontinuing tounmask thedetailsof sim-pleneuralprocessesinsimplebrains,andslowly,incrementally,putting these pieces together, will we ever be abletoaddress such global theoret-icalquestions asthe neural basisforlanguage.Wemust add to thismany new problems arising out of the comparisons of animal communication to language.If anything,theseproblemshavebecomemorecomplexand more confusing the more we have learned about the sophistication of other species' abilities and the paradoxes implicit in our own abilities. But the most criticalmissing piece of the puzzleisaccesstothe brainsinquestion:an-cestralhominidbrains.Thoughwehaveconsiderableinformationabout brain sizes infossil species,and a little information about brain shapes, the relevantanatomicalinformation,theinternalmicroarchitectureof these brains, has left no fossiltrail. With respect to fossil brains, we will never find the "smoking gun"-the firstbrain capable oflanguage. We will only have accesstocircumstantial information. So,what business do wehavespeculating about the beginnings of lan-guage? Given the complexity of the human brain, our current ignorance of many of its crucial operating principles, and the fact that neither languages nor the brains that produce them fossilize,there would appear to be many more immediate questions to be answered before even considering this one. There seem to be too many loose ends and gaps in the supportive evidence to proVide solid leads in the search for clues to the nature of the human mind inthe originsof language. But this ignores the Significance of the factthat languageisa one-of-a-kind anomaly. Often the most salient and useful hints about the underlying logicof nature'sdesignsare provided when unique or extremefeaturesin twodifferentdomainsarefoundto be correlated.Some notable examples include the correlation between superconductivity and extreme cold; be-tween greater cosmic distances and the increasing redness of starlight; be-tween the massive extinctions of fossilspecies and evidence of extraterres-trial impacts; between the peculiarity ofhaplo-diploid genetics and war, sui-cidal defense, and infertile castes in social insects; and so on.Each of these correlationscried out foran explanation and insodoing offered a critical cluetoamore generalprinciple.Themore tworelatedfeaturesdiverge fromwhat istypical in their respective domains, the more penetrating the insight that can be gleaned fromtheir underlying relationship. In thiscontext,then,consider the case of human language.It isone of themostdistinctivebehavioraladaptationsontheplanet.Languages evolved inonly one species,inonly one way, without precedent, except in the most general sense. And the differences between languages and all other naturalmodes of communicating are vast.Such a major shift inbehavioral adaptation could hardly failtohave left its impression on human anatomy. Evensuperficialappearancesbearthisout.Wehumanshaveananom-alously large brain and a uniquely modified vocal tract. Though these clues offer no more than a starting point, they suggest that the structural and func-tional relationships underlying these superficial correlations are likely to be robust and idiosyncratic tous. Ironically,then,the problem oflanguage originsmay actually offer one of the most promising entry pointsinthe searchforthe logiClinking cog-nitivefunctionsto brain organization. To the extent that theunique men-taldemandsoflanguagearereflectedinuniqueneuroanatomical differences, we may find an uneqUivocal example of how nature maps cog-nitive differences to brain structure differences. Though the details of this mysteryarechallenging,no critical piecesof thispuzzle lieburied inthe deep evolutionary past or inaccessible to current technology. They are ob-servable in the differences in cognitive abilitiesand brain structures to be found inliving species. Ithinkthatthe diffkulty of thelanguageoriginsquestionisnottobe blamedonwhatwedon'tknow,butratheronwhatwethinkwealready know. We think we know that what keeps language from being a widespread phenomenonisitsbyzantinecomplexityandtheincredibledemandsit places on learning and memory.We think weknow that language became possibleforour ancestorswhentheseimpedimentstolanguagelearning were overcome by some prior change in the brain. Depending on which as-pects of language are deemed tobe most complex,different prior adapta-tionsareinvokedtoexplainhowlanguagebecamepOSSible.Perhapsit required an increase in intelligence, a streamlining of oral and auditory abil-ities,aseparationof functionstothetwosidesof thebrain,or the evolu-tion of a sort of built-in grammar.Ithink we can be sure of none of these things.In fact,I thinkthat the problem ismorebasic andfarmorecoun-terintuitive than isassumed by any of these accounts. There are a few common assumptions shared by all of these explanations that I think are the root of a deeper problem. In general, these arguments parallel many others that continually resurface along that age old divide be-tween nature and nurture.Is language imposed fromthe outside or does it reflectwhat isalreadyinside?For decades,the superficiality of thisstale dichotomy has been evident,exposed by researchin the psychological and biologicalsciences that demonstrateshow truly complex and interdepen-dent the biological and environmental contributions to development can be; but westill findit difficult to conceive of these phenomena in other terms. We have reinvented the same old answers in new guises in each generation, stubbornly insisting that the answer to the question of language knowledge must be foundinone of just a fewmajor alternative paradigms(depicted incartoon fashioninFigure 1.1). A t one end of thisspectrum is the assumption that the architectureof language originates entirely outside (simple associationism); at the other end isthe assumption that it originates entirely inside (mentalese) . What other alternatives could there be, that are not captured between these extremes? And if there are no other alternatives,then shouldn't answering this ques-tionalso point tothe solution to the language origins question? Discover-ing whichaspectsof languageknowledgearecontributed bynatureand whichbynurtureoughttotelluswhatdifferenceinuswasnecessaryto bridge the original language acquisition gap. If the answer lies more toward the associationist end of the spectrum, then evolutionmust have given us languageby endowing uswith exceptionally powerful learning and mem-ory.If the answer lies more toward the mentalese end, then evolution must haveendoweduswithremarkablysophisticatedinstinctualknowledgeof language that made learning completely unnecessary. In lightof theseintuitively compelling alternatives,the approachI am about to take may seem misguided.Not only doI think that these alterna-tivesconfusethenature/nurtureproblemmorethanthey illuminate it,I think that the whole question of where language knowledge originates dur-ing development issecondary. Though a young child'salmostmiraculous development oflanguage abilities isindeed a remarkable mystery-one that willbe considered insome detail later (inChapters 4 and ll)-I think that thecauseof language originsmustbe sought elsewhere,and by pursuing some very different kinds of research questions. While wehave been wor-rying about where knowledge of language comes from, we have been avoid-ingamorebasicquestion:Whatsortof thingisknowledgeof language

- 'ABCD':: :: I........ .__ _ meanings asmental images meaningsasassociativemappings

-tt.:II. ABC0,':, I, II IS y45 , innate grammatical knowledgeinnate mental language Figure11Four cartoon depictions of some of the mfljor theoretical paradigms pro-posed to explain the basis ofhunum language.Topleft:The notion that word mean-ing is created when the perception ofthe sound of a spoken word is associated with an object both asperceived and as stored in the mind in the form ofa mental image. In this simple common sense view, stringing together words in a sentence leads the listener tobring together imagesin the mind.Topright: The notion that both word meaning and knowledge oflanguage stmcture are learned by internalizing patterns ofthe associative probabilities linking words to one another, and linking words and objects. B.FSkinner wasthemost prominentdefender of thisview,but recently more sophisticated versionsof thisbasicideahave been refonnulated with the aid ofinsights gained by studying parallel distributed learning processes.Knowledge of language is depicted as analogous to the distributed connection patterns in a neural net. Bottom left:One of the most influential views of grammatical knowledgecon-ceives of it as built in prior to language experience, like finl1warein a desktop com-puter (depicted as a computer chip inserted in the brain)_The structure oflanguage is imposed on strings ofwords (that presumably would still be meaninf!ful,just less useful, without this stmcture). This view was first explicitly fonnulated by the lin-guist Noam Chomsky. Bottom right: The extreme innatist view ofknowledge oflan-guage conceives ofit as an external reflection ofan internal lingua franca ofthe brain called "mentalese." InSteven Pinker's words (The Language Instinct, p. 82), "Know-ing a language,then,isknOWinghow totranslatementalese into stringsof words and vice versa.People without language would still have mentalese, and babies and manynonhuman aninwls presumably have simpler dialects. Indeed, babies did nothavea mentalese to translate toand fromEnglish,it isnotclear howlearning English could take place, or even what learning English would mean." None ofthese views prOVides a satisfacto11J explanation of the paradox explored inthis chapter. anyway?Beforeturningtothisquestion,however,itisworthwhilere-flectingonsome of theequallymisleadingevolutionaryassumptionsthat reinforce the traditional theoretical alternatives. Technical Difficulties and Hopeful Monsters Oneof the most common views about language evolution isthat it isthe inevitable outcome of evolution. Evolution washeaded this way, our way. Asthe only species capable of conceiving of our place among all others, we see what lookslike a continuous series of stagesleadingup to one species capable of suchreflections.It goeswithout saying that amore articulate, more precise, more flexiblemeans of communicating should alwaysbe an advantage, allother things being equal.Interms of cooperative behaviors, abetterabilitytopassoninformationaboutdistantor hiddenfoodre-sources,or to organize labor for ahunt,or to warnof impending danger, would be advantageous for kin and the social group as a whole.Better com-munication skillsmight alsocontribute tomoresuccessful socialmanipu-lation and deception. The ability to convince and mislead one's competitors or cooperate and connive\vithone'ssocial and sexual partners could also haveprovidedSignificantreproductiveadvantages,particularlyinsocial systems where competition determines accesstodefendableresources or multiplemates.Infact,it'sdifficulttoimagineany human endeavor that would notbenefitfrombetter communication.Looked atthisway,it ap-pears that humans have just developed further than other species along an inevitable progressive trend towardbetter thinking and better communi-cating. Surely we must be part of a trend of better communication in some form? Itseemstobe anunstated assumptionthat if biological evolution contin-ues long enough, some formoflanguage will even tually evolve in many other species. Are chimpanzees the runners-up, lagging only a little behind on the road to language? Asin Planet ofthe Apes, a science fictionmovie in which our more hairy cousinscatchuptoahuman level of mentaland liguistic abilities, we imagine that if givensufficient time,something like language is prefigured in evolution. We even imagine that if there is life on other plan-ets, and if it hasbeen evolving aslong aslife on earth, or longer, there will be "intelligent" specieswithwhomwemaysomedayconverse.l TheRe-naissancenotionof a"GreatChainofBeing"gaverisetonineteenth-century theories of phylogeny that ranked species from lower to higher, from mechanismtogodly,withhumansjustbelowangels.Thoughlaternine-teenth- andtwentieth-century evolutionistsrejectedthestaticranking of phylogeny and replaced it with the theory of evolutionary descent, the an-thropocentric perspective was simply rephrased in evolutionary terms. H u-mans were presumably the most "advanced" species.Carrying thisnotion to itsextreme, some people now suspect that there may be spaceships vis-iting earth, carrying beings that are "more highly evolved" than we are. On the surface, progress seems to be implicit in natural selection. Grad-ual improvement of adaptations seems toimplythat the longer evolution continues the better thedesign willbecome.Indeed,many scientists talk asthough a special kind of retrograde selection would be necessary to halt the progress of inevitably increasing intelligence. Small-brained species are oftenconsideredprimitiveor throwbackstoearlierforms,left out of the main trend. From an anthropocentric perspective, it seems unquestionable that more intelligent species willoutcompete lessintelligent ones.Intelli-gence isalways an advantage, right? Brain over brawn. We rank genius and mentalretardationonasinglescale,andpresumablyrankchimpanzees, dogs, and rats on the low end of the same scale.Human evolution isoften termed an "ascent" to imply a climb from lower to higher intelligence. And fromthisit seemstofollowthat humansare just the pinnacle example of an inevitable trend.The winner in a war of neurons. Theapparentreasonablenessof thisviewreflectsourfamiliaritywith technological progress in Western societies. The interchangeability of terms likeconsciousness expansion,social progress, and evolutionisnow almost commonplace in the popular press, and these ideas are seldom entirely dis-entangled even in the most sophisticated accounts of human evolution. But the idea of progress in evolution is an unnoticed habit left over from a mis-informed common sense,fromseeing the worldintermsof deSign.The problem isthat our intuitive model for evolution is borrowed from the his-tory of technological change, which has been a cumulative process, adding more and more tidbits of know-how to the growing massof devices, prac-tices, and records each day.In contrast, biological evolution isnot additive, except in some \lery limited ways. The human repertoire of expressed genes isabout the same asthat ina mouse or frog,and the body plans of allver-tebrates seemto bemostly modificationsof thesameshared plan----even for the brain. Though we are on the large end of the range of body and brain sizes, thisisnot the result of adding new organsbut merely enlarging ex-isting ones with slight modifications. Evolution isan irreversible process, a process of increasing diversifica-tion and distribution. Only in this sense does evolution exhibit a consistent direction.Like entropy, it isa process of spreading out to whatever possi-bilities are unfilled and within reach of a little more variation. Evolution does not continue to chum out ever better mousetraps,even if it hasproduced some remarkable examples along the way.But this pattern of spreading into unfilled niches does place usinone of the more extreme niches. Evolution isdiversification in alldirections, but there aremore options availableinsome directionsthan others.Organisms started out smalland short-lived and couldn't get much smaller,but they could always get larger and more long-lived.For the smallest organisms,the resources that can be devotedtointernalrepresentations of the worldare limited,thougheven bacteriaappearabletousetheironeinformation-storagesystem,their genes,totakeininformationfromaround themandmodifYtheir behav-iors appropriately.But the upper end of the range of information-handling abilities wasnot Similarly bounded, and so thedifference between the low end and the upper end of this range has increased over the hundreds of mil-lion yearsof animal evolution aspart of thisdiversification.Nevertheless, the number of small-brained creatures has not diminished because of com-petition with those with big brains, and theno-brainers-all the plants and single-celledorganisms-vastly outnumber therest of us.It just happens thatonevery,veryminorevolutionarydirectionistowardnicheswhere doing a lot of information processing during one's life is a good way to pass on genes.Inevitable? Well,it's about asinevitable a direction in evolution asthe development of arctic fishwithantifreeze intheir blood or electric eels who use electricity to sense their way throughmuddy Amazonian wa-ters. The niche was just there, and was eventually filled.Still,in some mea-sure, we arenear the extreme of this distribution. The question,then, iswhether the evolution oflanguage wassomehow prefigured in thistrend.Isthere a general trend toward better communi-cation? It'seasy inhindsight to arrange the history oflong-distancecom-municationfromtelegraphs totelephones to cellular phones toStar Trek communicators. It isnot so easy to determine if animal communication has been steadily getting better and if human language is a part of such a trend. Certainly,there were advancesindistanceand signal clarity inevolution, but even if wenarrow our comparison to vocal communication, there isno evidenceinlivingspeciesthatsomeinevitable progressive trend leadsto us.Apparentlysimplespeciescanusehighlycomplexmethodsof sound communication, and some highly complex species can be oblivious to their advantages.There are alsomany great sound tricks,such asecholocation, that are completely beyond human ability.Among our closest relatives, the great apes, there are both highly vocal(chimp) and nearlysilent (orangutan) species.Infact,most birds easily outshine any mammal in vocal skills,and though dogs, cats, horses, and monkeys are remarkably capable learners in many domains, vocalization isnot one of them. Our remarkable vocal abil-ities are not part of a trend, but anexception. We also tend to underestimate the complexity and subtlety of much non-humansocialcommunication.Inrecentdecades,fIeldstudiesof social communication innonhuman species have demonstrated that many birds, primates, and social carnivores use extensive vocal and gestural repertoires to structure their large social groups. These provide a medium for organizing foragingand group movement, for identifYing individuals,formaintaining andrestructuringmultidimensionalsocialhierarchies,formediatingag-gressive encounters, for soliciting aid, and for warning of dangers.2 Indeed, even in our own species, a rich and complex language isstillno substitute for a shocked expression,a muffled laugh,or a floodof silent tears, when it comes to communicating some of the more important messages of human social relationsru ps. However, although they are complex, these elaborate repertoires of calls, displays,and gesturesdonot seemtomaponto any of the elementsthat compose languages. Although various researchers have suggested that par-allels to certain facets of language are to be found in the learned dialects of birdsong,theexternalreferenceevidentinvervetmonkeyalarmcallsor honeybeedances,andthesociallytransmittedsequencesof soundsthat make up humpback whale songs(eachof which will be considered in some detail in Chapter 2), these and many other examples like them only exhibit a superficialresemblance to language learning, word reference, or syntax, respectively.Even if we were to grant these parallels, no nonhuman species appears to put thesefacetsof languagetogether into a coordinated,rule-governed system. Could wehavemissedrecognizing nonhuman languages becausethey areasalientousasour speech isto them? People havelong entertained thispossibility at least in mythology and children'sstories.They offer the fantasy that wemight someday overcomethecommunication boundaries that separate !1Umans and other animals and share memories, beliefs, hopes, and fears with them.In the popular children's book made into a movie, Dr. Doolittle enlists the aid of a "multilingual" parrot to translate between an-imal and human speech. But iseven a very superfIcial "translation" possi-ble?Whatdoyoutellachildwhoasks,"What isthekittysaying?"Do animals' vocalizations and gestures explain, describe, ask, or command? Do theyargue,disagree,bargain, gossip,persuade,or entertain oneanother with their thoughts? Are there any corresponding elements in animal com-munication that map onto the elements of human language? Unfortunately, animal calls and displays havenothing that corresponds to noun parts or verb T .....................,.,n._ ____A~ parts of sentences, no grammatical versus ungrammatical strings,no mark-ing of singular or plural,no indications of tense,and not even any elements that easily map onto words, except inthe most basic sense of the beginning and ending of a sound. One quite reasonable caution against making strong claims about the ab-senceof nonhuman languages isthat our study of other species'commu-nication systems is still in its infancy. Isn't it more prudent to remain agnostic untilwehavelearned a great dealmore aboutother species communica-tions? It is always wise not to prejudge the evidence, especially with respect to a subject about whichwe haveundeniable prejudices. And there are far more species about whose communicative behavior we know next tonoth-ing than there are whose communication hasbeen studied.Nevertheless, I think that wehave suffIcient information to make a reasonably confident claimeven aboutspecieswhosecommunicative behaviors havebeen only superflcially studied.What makesthisa fairly safe guessisnot the sophis-tication of our behavioral analyses, but rather the striking characteristics that wouldbe evident ina nonhuman language.Where the differences should be glaring, thesensitivity and sophistication of observationsand testscan be minimal. What would be the characteristics of anonhumanlanguage that would allow usinstantly to recognize it as a languagelike form of communication, even if it were quite alien with respecttoallhuman languages? This isthe sort of question that scientistsscanning the heavens withradiotelescopes listeningforsignalsfromunearthly speciesmustaskthemselves,or that might be asked by those engaged in electronic surveillance interested in dis-tinguishingthetransmissionof codedor encrypted signalsfromrandom noise.Interestingly, many of thesecharacteristicsare exhibited inthe sur-face structure of the signals, and require no special insight into meaning or referential function, and no obvious correspondence with natural language grammars,to discern.3 Hereare a few general features that ought to stand out.A languagelike signal would exhibit a combinatorial formin which dis-tinguishable elements are able to recur in different combinations. It would exhibit a creative productivity of diverse outputs and a rather limited amount of large-scaleredundancy.And although there would be a high degree of variety inthe possible combinations of elements, the majority of combina-torial possibilities would be systematically excluded.In terrestrial examples, where it would be possible to observe the correlations between signals and contextual events, there should be another rather striking and counterin-tuitive feature. The correlations between individual signals and the objects and eventsthatformthe context of theirproductionshould not exhibita simpleone-to-onemapping.The correlations between signelements and their pragmatic contextshoulddiffer radically yet systematically fromoc-casionto occasion,depending onhowthe signalsarearranged incom bi-nation with respect to one another. These, of course, are all general features associatedwithsyntax,thoughnotjustlanguagesyntax.Humangames, mathematics, and even cultural customs exhibit these features. If a radio-telescope observer identified a signal emanating fromdistant space with these characteristics it would make world headlines, despite the fact that themeaning of the signal would remain completely undecodable. With farmore to goon than this,in even superficially studied animal com-munications, we can be reasonably sure thatforthe vastmajority of likely candidate species such a signal hasnot yet been observed.Instead, though highlycomplexandsophisticated,thecommunicativebehaviorsinother species tend to occur asisolated signals,in fixed sequences, or in relatively unorganized combinationsbetter described by summation than by formal rules.Andtheir correspondences with events and behavioral outcomes, in the cases where this can be investigated, inevitably tum out to be of a one-to-onecorrelational nature. Thoughanas yet undescribed exampleof an animal communication systemthat satisfies these criteria cannot be ruled out, it seems reasonable to conclude that the chances are poor that it would have gone unobserved in common animal species, any more than we would missitin acosmicradio signal. My pOint is not that we humans are better or smarter than other species, or that language isimpossibleforthem.It issimply that thesedifferences are not a matter of incommensurate kinds oflanguage, but ratherthat these nonhuman forms ofcommunication are something quite different from lan-guage.For thisreason I think that the comparison ismisguided, and use-f ulonlyata very superficiallevel.Thisfactshouldnot betoodifficultto appreciatebecause weallhavepersonalexperienceof justthe sort of in-commensurability Iamtalking about.There arenumeroushuman coun-terparts too t h ~ ranimals'nonlinguistic communicativebehaviors. We too have a widerange of innately produced and universally understoodfacial expressions, vocalizations, and gestures. Asin other species, they are an ir-replaceablecomponentof humansocialcommunication.Yetthisisnot analogoustobeingbilingual.Thisother repertoire of communicativebe-haviors isnot a language of gestures instead of words.It issomething else. And although these human calls and gestures comprise an entirely familiar system, we find the same djfficulty translating theminto word equivalents as we do withanimal calls and gestures with which wearefarlessfamiliar. The problemisnottheir unfamiliaritybutrather that it simplymakesno sense to ask what kind of word a laugh is, whether a sob isexpressed in past or present tense,or if a sequence of facialgestures iscorrectly stated. The problemisn'tadiffIcultymappinghumantononhumanlanguages,but rather a diffIculty mapping languages to any other form of naturally evolved communication, human or otherwise. Of noother natural formof communicationisitlegitimate tosay that "language isa more complicated version of that." It isjustasmisleading to call other species' communication systems simple languages asit isto call them languages. In addition to asserting that a Procrustean mapping of one to theother ispossible,theanalogy ignores thesophisticationand power of animals'nonlinguistic communication,whosecapabilitiesmayalsobe without language parallels. Perhaps we are predisposed to see other species' communicationsthroughthefilterof languagemetaphorsbecauselan-guage istoo much a natural part of our everyday cognitive apparatus to let useaSily gainanoutside perspective on it.Yetour experience of itsnatu-ralness, its matter-of-factness,belies its alien nature in the grander scheme of things.Itisan evolutionaryanomaly,notmerelyanevolutionary ex-treme. This lack of precedent makes language a problem for biologists.Evolu-tionary explanations are about biological continuity,soa lack of continuity limits the use of the comparative method in several important ways. We can't ask,"What ecologicalvariablecorrelates with increasing language use ina sampleof species?"Norcan we investigate the "neurological correlates of increased language complexity." There isno range of species to include in our analysis.Asa result, efforts to analyze the evolutionary forcesrespon-sible for language have often relied on crude substitutes to make up for the lack of homology between language and nonhuman forms of communica-tion.It is tempting to try to conceive oflanguage as an extrapolated extreme of something that other speciesproduce, such ascalls,grunts,gestures, or social grooming.4 It isalso tempting to tum to some other feature of human anatomy that can be more eaSily compared to other species asan index of language evolution.Humans can, for example, be ranked along with other species with respect to brain size, group size, social-sexual organization, for-aging strategy, etc. But even if humans are at the extreme in many of these measures,thecorrelationsamongtheseattributesarenotobvious,and their linkage to language is dubious, since trends in these attibutes in other species occur irrespective of language. Interpreting the discontinuity between linguistic and nonlinguistic com-municationasanessentialdistinctionbetweenhumansandnonhumans, however, has led to an equally exaggerated and untenable interpretation of languageorigins: the claimthat languageisthe product of auniqueone-of-a-kind piece of neural circuitry that provides all the essential features that make language unique (e.g.,grammar).But this doesnot just assume that there isauniqueneurological featurethat correlates with thisunique be-havior, it also assumes an essential biological discontinuity. In other words, thatlanguage issomehow separatefromthe restof our biology andneu-rology. It isas though we are apes plus language-as though one handed a language computer to a chimpanzee. Thisremindsme of a wonderful piece of modem mythology froma re-cent filmentitled Short Circuit. A sophisticated robot isaccidentally trans-formedfromamechanismthat"justrunsprograms"intoaconscious, self-aware being asa result of being struck by lightning. The power surge damageditscircuitsinjusttherightway.Thenowconsciousrobot,of course, does not think of this as "damage." From his perspective, the light-ning bolt corrected a design limitation.Asa cinematic device,thebolt of lightning accomplishes two important things.The catastrophic and unpre-dictable nature of lightning provides a vehicle for invoking drastic and un-precedentedchange,anditsintrinsicallychaotic-and,bytradition, miraculous-character obviatesany possibility of describing exactly what alterationschangeda computer mechanism intoa human-typemind.For the sake of the story, we suspend critical analysis and allow thismiraculous accident to stand in place of anotherwiseinexplicable transformation. As an allegory of human mental evolution, it offers a paradigm example of what biologists call a "hopeful monster" theory: the evolutionary theorist's coun-terpart to divine intervention, in which a freak mutation just happens to pro-duce a radically different and serendipitously better-equipped organism.5 The Single most influential "hopeful monster" theory of human language evolution wasoffered by the linguistNoamChomsky,and hassincebeen echoed by numerous linguists, philosophers, anthropologists, and psychol-ogists. Chomsky argued that the ability of children to acqUire the grammar of their first language, and the ability of adults effortlessly to use this gram-mar, can only be explained if weassumethat all grammars are variations of asinglegeneric"UniversalGrammar,"andthatallhumanbrainscome with a built-in language organ that contains this language blueprint.This is offered as the only plausible answer to an apparently insurmountable learn-ing problem. Grammars appear to have an unparalleled complexity and sys-tematic logical structure, the individual grammatical "rules" aren't explicitly evident in the information available to the child, and when they acquire their firstlanguage children are still poor at learningmany other things. Despite these limitations children acquire language knowledge at a remarkable rate. Thisleadstotheapparentlyinescapableconclusionthatlanguageinfor-mationmustalready be "inthebrain" before the processbeginsinorder foritto be successful.Childrenmust already "know" what constitutesan allowable grammar, in order to be able to ignore the innumerable falsehy-potheses about grammar that their limited experience might otherwise sug-gest. Thisdevice,a "language organ" unique to the human brain,could also account for the failureof other species to acquire language. The appeal of thisscemuioisthatiteliminatesmany troublesomequestionsinonefell swoop:the discontinuity between human and nonhuman communication, the larger human brain (adding a new part enlarges it), the systemic inter-dependentnatureof grammaticalrules(they allderivefromoneneuro-logicalsource),thepresumeduniversalfeaturesof languagestructure (ditto), the intertranslatability of languages (ditto), and the ease with which languageisinitiallyacquireddespiteaninsufficientinputandalackof grammatical error correction by adults. Another appeal of thehopefulmonster story isthat it promisesa defi-nite and dramatic transition fromone stage to another in the evolutionary sequence.It offers asingle-stepevolutionary account that ismuch easier to comprehend and organize inone's thinking than continuous changes in-volVingmultiplefactorsinteractingandoverlappingintimeincomplex ways. It tantalizes the imagination to hear that the story of human origins was written in the course of a single dramatic and decisive prehistOlic event. In one step,someancestor crossed the threshold into humanity.But such a crucial transition could hardly occur without leaving a trail of evidence at-testing to its discontinuity.If modem language abilities appeared all of a sud-den in human prehistory, then we ought to find numerous other correlates of a radical reorganization of human behavior and biology. Inspired by this possibility, researchers in many fields have combed through their own data for signs of sudden transitions that might be the result of such anincredi-ble language mutation.Not surprisingly,many have been "discovered" in the record of human prehistory. They include: abrupt technological transi-tions(e.g.,thefirst appearance of stone tools or of extensive cultural vaIi-ationsintooldesign);possiblepunctuatedspeciationevents(e.g.,the OIigination of anatomically modem humans froma "mitochondrial Eve"); rapid population changes (e.g., the demise of the Neanderthals); and signs of major innovations incultural artefacts(the firstappearance of durable representative art, such as carvings and cave paintings inEurope).But be-cause they offer evidence that isindirect, at best,and sosparse and frag-mentary, paleontological finds canappear irregular for many other reasons, notleastof whichisour predispositiontoorganize theevidenceincate-gorical terms. Anaccidentallanguage organrequires no adaptive explanationforthe structure of language.If this hypothetical organ was plugged into the brain in a single accident of prehistory, rather than evolving bit by bit with respect to itsfunctionalconsequences,thenno functionalexplanationswouldbe necessary.If it wasjustanaccident,anyutility would be entirely acciden-talaswell,discovered after the fact.Thistoomightaccount forthe many little idiosyncrasies oflanguage and its discontinuities when compared with other nonhuman formsof communication.But I think that this story isfar tooneat andtidyprecisely becauseitsuggeststhatsomany questionsdo notneed tobeaddressed. Theaccidentallanguageorgantheory politely begsustoignorethemessydetailsof languageorigins,abandonhope of finding precedents in the structure of ape brains or their cognitive abilities, and stop looking for any deep design logic tothe structural and functional relationships oflanguage grammars and syntax. This is a lot to ignore. What does this hypothesis provide instead? One of the characters inMoliere's play The Imaginary Invalid,6 is asked by hisphysician-examiners to explain themeansby which opiuminduces sleep.He replies that it induces sleep because it contains a "soporific fac-tor." This answer isapplauded by the doctors. The playwright is, of course, satiIizing the falseexpertiseof these apparently learnedmenby showing their knowledgetobe nomore thansophistry.The answerisa nonexpla-nation.It merely takes what isinneed of explanation and givesit a name, as though it were some physical object. Like phlogiston, the su bstance once hypotheSized bypre-atomicchemistry to be the essencethat determined flammability, the "soporific factor" fails to reduce the phenomenon in need of explanation to any more basic causal mechanisms. Formanylinguists,grammaticalknowledgeiswhatneedstobeex-plained,and what islacking isanadequate account of thesourceof chil-dren'sgrammatit:alandsyntacticabilitiesintermsof antecedentsinthe child's expeIience oflanguage. We are thus like the characters inMoliere's play, who know what isproduced but don't know how itisproduced.Fail-ing to discover a satisfactory explanation forhow grammatical knowledge couldbe impressed uponchildren'sminds fromthe outside, wenaturally tum to the possibility that it does not come from the outside at all.But sim-ply assuming that this knowledge is already present, and so doesn't need to pass fromoutside to inside, only restates this negativefindinginpositive terms.A grammar instinct or a universalgrammar serve asplaceholders for whatever couldnotbelearned.Thenatureof thispresumed innate knowledge of language isdescribed only in terms of itsconsequences. Lin-guistshave progressively redefined what supposedly cannot be learned in evermoreformaland preciseterms,and so wemayhavethefeelingthat these accounts are approaching closer and closer to an explanation.But al-though the description ofwhat is missing has gotten more precise, ultimately itisonlyamoreandmoreprecise versionof whatismissing.These"ex-planations" of the nature of a language instinct are inevitably presented in the guise of elaborate defInitions of grammatical principles or else as some-thing akin to computer programs, and in this way they are only more for-mal restatements of the problem of the missing information. Saying that the human brain alone produces grammar because it alone possesses a gram-mar factor ultimately passesthe explanatory buck out of the hands of lin-guistsand into the handsof neurobiologists. To be fair, the intent oflanguage organ theories is not to address the ques-tion of initial language origins, but rather to explain the source of language competence in development.For this reason, it isnot wedded to the hope-ful monster assumption. Steven Pinker, a proponent of the Universal Gram-marviewoflanguageabilitiesandanarticulatechampionof manyof Chomsky'soriginal inSightsabout the uniqueness of language,argues in a recent book (The LanguageInstinct) that innate grammatical knowledge is not at all incompatible with an adaptationist interpretation of its origins.He argues that a language instinct could have gradually evolved through the ac-tion of natural selection. On the one hand, this is a far more biologically plau-siblealternativeto miraculousaccidentsand itchallengesusto facesome of the difficult problems ignored by theories relying on miraculous accidents to fillin thegaps.On the other hand,anadequateformalaccount of lan-guagecompetencedoesnotprovideanadequate account of how itarose throughnaturalselection,andthesearchforsomenewstructuresinthe humanbraintofulfHthistheoreticalvacuum,likethe searchfor phlogis-ton,hasno obvious end point.Failure to locate it in such a complex hier-archyof mechanismscanalwaysbedismissedwiththeinjunction:look harder. A fullevolutionary account cannot stop with a formaldescription of what ismissing or a scenario of how selectionmight havefavoredthe evolution of innate grammatical knowledge. It must also provide a functional account of why its particular organization was favored, how incremental and partial versionswerealsofunctional,andhowstructurespresentinnonhuman brains weremodifiedto prOVidethisability.The languageinstinct theory proVidesanend point,anassessment of what alanguage evolution theory ultimatelyneedsto explain.It rephrasestheproblembygivingitanew f t ~I ~ ' .__~ .________. ________________________________________ name. Butthis offers little more thanthemiraculous accident theory pro-vided: a formalredescription of what remains unexplained. Unfortunately, I think italsomissesthe forestforthe trees,even intlusendeavor. I don't thinkthatchildren's grammatical abilitiesarethecrucialmysteryof lan-guage. TheMissing Simple Languages Thetwodominantparadigmsforframingthelanguageoriginsques-tion-the evolutionof greater intelligenceversusthe evolutionof aspe-cializedlanguageorgan-have onethingincommon:botharestatedin terms of the problem of learning a very largeand complex set of rules and signs.They assume that other species arepoor language learners because language isjust too complicated for them to learn,and too demanding for themtoperform.It requiresrapidandefficientlearning,demandsim-mense memory storage, takes advantage of almost supernatural rates of ar-ticulation and auditory analysis, and poses an analytic problem that is worthy of a linguistic Einstein. Both approaches assume that the difficulty for other speciesisthe complexity of language,but they disagree onthe sourceof this difficulty and on what isrequired to overcome it.Do human children merely need to be very much more intelligent than other species in order to learn language, or is language so complicated that it is impossible to learn without some built-in language information to "jump start" the process? Ac-cepting one or the other assumption leads to opposite claims about the na-ture of the evolution of language and the human mind.If language is just difficult to learn, then the neural adaptation that supports it could be quite general in its effect on cognitive abilities. Iflanguage is,for all practical pur-poses, impossible to learn,then the neural adaptationnecessary to support it wouldneed to bequite specific.However one looksatthese problems, itappearsthatovercomingthelimitationsimposedbytheobviouscom-plexity oflanguage is a prerequisite to language evolution. I say it "appears" this way,because I think that something has been missed in both views of the problem,something fundamental.These alternatives,and many plau-sible intermediates, only address one of the main problems inneed of ex-planation, and it isnot the critical one. A task that is phYSically too difficult to perform may exceed our strength, our endurance, our rate of performance, our precision of action, our capacity to do manythingsatthe sametime,etc. In cognitiveterms,thesecorre-spond to our ability to focus attention, the persistence of our memories, our rate of learning,and our short-termmemory span,etc. When we say that askillisdifficulttolearn,wemeanthat thedesiredmovementsequence severely taxesour ability precisely to time, control, or coordinate the com-ponent movements. When we say that a perceptual task is difficult to learn, wemean that it requires utilizing cues too subtle or fleeting to detect, too irregularto discover theircommonalities,or embedded intoo many dis-tracting cues to sort out. And when we say that a cognitivetask isdifficult to learn, we mean that there are too many associations to be held in work-ingmemory at one timeor too many to be considered intoolittletimeor simply too many to be remembered. Each demands too much, from too lit-tle,in insufficient time. Both the complexity of the task and the resources one hasavailable willdetermineits relative difficulty. Clearly, languageiscomplicated inallthese ways.Linguistic communi-cation requires us to learn and perform some remarkably complicated skills, both inthe production of speech and inthe analysisof speech sounds.In addition,there isa great deal to learn:thousands of vocabulary itemsand an intricatesystemof grammatical rulesand syntactic operations. And it's not enough that language iscomplicated. According to many linguists, we aren't even offered sufficient outside support to deal with it. We are forced to figureout the underlying implicitrulesof grammar and syntax without good teaching and with vastly inadequate examplesand counterexamples. This apparent lack of adequate instruction adds insult to in jury, so to speak, by making a too complicated task even harder. The degree to which the sup-portforlanguage learningundershootsthisneedis(exponentially)pro-portional to how complicated the task is to begin with, and so the complexity of language isdoubly limiting. How could anyone doubt that language complexity isthe problem? Lan-guages are indeed complicated things. They are probably orders of magni-tudemorecomplicatedthanthenext-most-complicatedcommunication system outside of the human sphere.And they areindeed almost impossi-bly difficult for other species to acquire. The question is whether this com-plexity is the source of the difHculty that essentially limits the use of language to our species alone.Althoughthiswouldseemto be the obviousconclu-sion, it isnot quite so obvious asit might first appear. The most crucial dis-tinguishing featuresoflanguage cannot be accounted formerely interms of language complexity. The challenge tothe complexity argument forhuman language origins rests on a simple thought experiment.Imagine a greatly Simplified language, not a child'slanguagethat isa fragment of a morecomplicated adultlan-guage, but a language that is logically complete in itself, but with a very lim-ited vocabulary and syntax, perhaps sufficient for only a very narrow range of activities.I do not mean "language" in a metaphoric sense, the way that all communicationsystemsare sometimes glossed aslanguages.ButIalso donot restrict my meaning to speech, or to a systemwhose organizational principles are limited to the sorts of grammatical rules found in modem lan-guages. I mean language in the following very generic sense: a mode of com-munication based upon symbollc reference (the way words refer to things) and involving combinatorial rules that comprise a system forrepresenting synthetic logical relationships among these symbols.Under this definition, manual signing, mathematics, computer "languages," musical compositions, religious ceremonies, systems of etiquette, and many rule-governed games might quality as having the core attributes oflanguage. More important, no morethan atiny "vocabulary"of meaningfulunitsand only two or three typesof combinatorialruleswould be necessary tofulfillthesecriteria. A five- or ten-word vocabulary and a syntax assimple astoddlers'two- and three-word combinations would suffice.Reducing the definition oflanguage tosuchminimal conditionsallowsustoconceive of languagelikesystems thatarefarsimplereventhanthecommunicativerepertoiresfoundto occur in the social interactions of many other species. So this is the real mystery.Even under these loosened criteria, there are nosimplelanguagesusedamongotherspecies,thoughtherearemany other equally or more complicated modes of communication. Why not? And the problem is even more counterintuitive when we consider the almost in-surmountable difficulties of teaching language to other species. This issur-prising, because there are many clever species. Though researchers report thatlanguagelikecommunicationhasbeentaughttononhumanspecies, even the best results are not above legitimate challenges, and the fact that it isdifficult to prove whether or not some of these efforts have succeeded attests tothe rather limited scope of the resulting behaviors, aswellasto deep disagreements about what exactly constitutes languagelike behavior. Both the successes and failuresthat have come of thisresearchare never-thelesshighly informative withregard toboth what animalscan and can't do and how we ~ o n c e i v eof language itself, but the fewarguable successes must be seen against the background of domesticated animalsand family pets that never seem to catch on,despite being raisedin a context where they are bombarded with a constant barrage of commands, one-sided con-versations, and "rhetorical" questions. For the vast majority of species in the vast majority of contexts, even simple language justdoesn't compute. This lack of simple languages in the wild and inability to learn simple languages under human tutelage don't make sense!Many of these species engage in natural communicative behaviors that are farmore complex than a simple language,and they arecapableof learning larger setsof associationsthan are necessary for constructing a simple language.So why islanguage such a problem? The difference cannot be simple versus complex. The complexity of language isimportant.Itdemands an explanation, as doestheabilityof youngchildrentomakesenseof it,seeminglywithout sufficient feedback or time at their disposal. These are remarkable aspects of the language mystery, but they are secondary to a more basic mystery that has a lot more to do with the human/nonhuman difference.Despite the in-telligence of other species, and the fact that they engage in communicative behaviors that are as complex inother waysasa simple languagemight be, no other language systems exist. And it's not just a matter of their not being needed.For somereasonevena simple languageseemsimpossibly diffi-cult fornonhumans.Thisposesa profound riddle.So why hasitbeen ig-nored? Perhaps wehavebeen too preoccupied by thedetailstorecognize thissimpler problem.Or maybe we have been tooeager tocast the prob-lem interms of progress in communication, with humans in the lead. What-everthereason,it'stimewerecognizedthatthequestionswethought needed tobe explained bya theory oflanguage origins were secondary to a more fundamental mystery: Why aren't there any simplelanguages? And whyareevensimplelanguagessonearlyimpossibleforotherspeciesto learn? This changes everything. If complexity isnot the problem, then theories thatpurport toexplainlanguageevolutionintermsof overcomingcom-plexity losetheir justifIcation.A smallvocabulary shouldnotrequirevast intelligenceor memory capacity or articulatory skilltomaster.Lower in-telligence of our primate and mammalian relatives cannot, then, be the rea-son they don't catch on. A simple grammar and syntax should also be a trivial matter to learn.NospecialbUilt-inencoder-decoder for grammars should benecessaryif the combinatorial analysesaresimpleand the possibleal-ternativesarerelativelyfew.Evenminimalpowersof inductivelearning would suffice.The whole raison d'etre of an innateUniversalGrammar or languageorganevaporateswhenitcomestosimplelanguages.Finally, complex phonology, rapid articulation, and automated speech sound analy-sisareequally unnecessary.The learning problems addressed by allthese theories do not explain the absence of nonhuman languages,they only ex-plainwhynonhumanlanguagesshouldnotbeascomplicatedashuman ones. They point to issues that are relevant to complex modem human lan-guages, but they donot illuminate the phenomenon weoriginally thought they explained. They don't provide any clue to why language evolved in the humanlineage andnowhereelse.Thisisanapples-versus-orangesprob-lem,not acomplicated-versus-simple one. It'snot just curiousthatother species haven't started onthisevolutionary path; itdefiescommon sense. What isleft that isdifficultabout learning language,if itscomplexity is not at issue? When westrip away the complexity,only one significant dif-ferencebetween language and nonlanguage communicationremains: the common, everyday miracle of word meaning and reference. Neithergrammar,norsyntax,norarticulatesoundproduction,nora huge vocabulary have kept other species fromevolving languages. Just the simple problem of figUring out how combinations of wordsrefer to things. Why should this be so difficult? Why should the curiously different way that languages represent things have imposed such an almost impenetrable bar-rier in evolution? If we succeed in explaining this one paradoxical difficulty, we may catch a glimpse of the critical evolutionary threshold that only our own ancestors managed tosurmount . The firstmajor taskof thisbook,then,istodescribe precisely the dif-ferencebetweenthisuniquehumanmodeof reference,whichcanbe termed symbolic reference, and the forms of nonsymbolic reference that are found in all nonhuman communication (and in many other forms of human communication aswell).The second task isto explainwhy itissoincredi-bly difficultfor other speciesto comprehend thisformof reference.And the third task isto prOVideanexplanationfor how wehumans(and a few other animalsincarefully structured language learning experiments)have managedtoovercomethisdifficulty.Even thoughthisaspectof thelan-guage originsmystery isonly a part of the story oflanguage evolution, and understandingthisdifferenceoffersnoimmediateanswerstowhylan-guages are ascomplex asthey are today, or why they obey seemingly inex-plicabledesignrules,or howitispossibleforhumanchildrentomake sense of these otherwise byzantine and atypicaldetails, none of these other questionscanbeanswered without taking symbolicreferenceasagiven. But it isnot a given.Grammatical rulesand categories are symbolicrules and categories.Syntactic structure isjust physical regularity when consid-ered irrespective of the symbolic operations itisintended to encode. The-ories oflanguage and mind that fail to address this issue head on, or suggest that it needs no explanation, ultimately assume what they set out to explain. Wemust explainthe curious difficulty of symbolic reference fIrst. InhindSight,the centrality of this problem wasrecognized allalong,at least impliCitly. It would not be an exaggeration to suggest that more philo-sophicalink hasbeen spilt over attempts toexplainthe basisforsymbolic reference than over any other problem. Yet despite its intuitive familiarity (orbecause of it),and notwithstanding the effOlts of some of the greatest TorranN::!1IWnO'!ll,..nn...A' minds of each century, it remains curiously unresolved. Linguists, too, have struggled withthis problem inthe formof semantic theories, with parallel difficulties.For this reason, we should not be surprised to find that it resur-faces as the central riddle in the problem oflanguage origins. Linguists, psy-chologists,andbiologistscannotbeblamedforfailingto solvethisbasic mystery of mind before turning their efforts to other aspects of the language problem.Grammarandsyntaxcanbestudiedandcomparedfromlan-guagetolanguageandthecorrelations between language processesand brain functions can also be identifled irrespective of solving the problem of symbolic reference.Even many facets oflanguage learning can be studied without considering it in any depth. But theories that purport to explain the human/nonhuman difference in language abilities cannot ignore it, nor can accounts of what makeshumanmindsdifferent fromnonhuman minds. But if the way languagerepresents things hasbeen the primary barrier to language evolution inother species,then we willalsoneed to rethink a great many other aspects ofhuman mental evolution. Iflanguage complexity isasecondarydevelopmentwithrespecttothismoreprimary cognitive adaptation, then most theories have inverted the evolutionary cause and ef-fectrelationshipsthathavedrivenhumanmentalevolution.They have placed the cart (brain evolution)before thehorse(language evolution). If neither greater intelligence, facile articulatory abilities, nor preSCient gram-matical predispositions of children were the keysto cracking thissymbolic barrier, then the evolution of thesesupports for language complexity must have been consequences rather than causes or prerequisitesof language evolution.More important, these adaptations could not have been the most critical determinants of brain evolution in our species. Approaching the lan-guage origins mystery fromthis perspective is like stepping out of a mirror to findeverything to be the reverse of what we assumed. From this perspective language must be viewed as its own prime mover. It is the author of a co-evolved complex of adaptations arrayed around a sin-gle core semiotic innovation that was initially extremely difficult to acquire. Subsequent brainevolution wasaresponse tothisselection pressure and progressively made this symbolic threshold ever easier to cross. This has in turn opened the door for the evolutionof ever greater language complex-ity.Modem languages, with their complex grammars and syntax, their mas-sivevocabularies,andtheirintensesensorimotordemands,evolved incrementally fromsimplerbeginnings.Though simplelanguages existin no SOCietyfoundtoday,they almostcertainly existedatsomepoint inour prehistory.These simple languageswere superseded bymodemcomplex languages,and thebrainsthatoriginally struggled tosupportsimplelan-guages were replaced bybrainsbetter suited to this awkward adaptation. Somehow, despite their cognitive limitations, our ancestors found a way tocreateandreproduceasimplesystemof symbols,and onceavailable, these symbolic tools qUickly became indispensable. This insinuated a novel mode of information transmission into the evolutionary process for the first time in the billions of years since living processes became encoded in 0N A sequences.Becausethisnovelformof informationtransmission waspar-tially decoupled from genetic transmission, it sent our lineage of apes down anovelevolutionary path-a paththathascontinued todivergefromall other species ever since.Thisinversion of cause and effect hasenormous consequences.If thehumanpredispositionforlanguagehasbeenhoned by evolutionfora Significantfractionof our prehistory,thenour unique mentality must also be understood inthese terms.The incessant demands of efficiently reconstituting a symbolic system in each generation would have created selection pressures to reshape our lineage's ape brains to fitthisnew function. The implications for brain evolution are also profound. The human brainshouldreflectlanguage initsarchitecture the waybirdsreflectthe aerodynamicsof flightintheshapeandmovementsof their wings.That which is most peculiar about language processing should correspond to that which ismost peculiar about human brains. So if what ismost unusual about language isitssymbolicbasis, what ismost unusual about our brains? Human brains are unusually large:three times larger thanthey should be foran ape with our sizebody.But large brain sizeisonly the mostsu-perficialsymptomof asubstantialreorganizationatdeeperlevels.Un-packingthiscomplicatedanatomicalproblemandmappingitonto the specialcomputationaldemands posed by languageisthe purpose of the middle sectionof this book.Looking more closely,we willdiscover that a radicalre-engineering of thewholebrainhastakenplace,andona scale that isunprecedented.Interpreting these differencesasconsequences of the functional demands imposed by eons of language processing may offer new insight i ~ t othe relationship between differences in cognitive function and differences in large-scale brain organization.In the co-evolution of the brainand language two of themostformidablemysteries of science con- I verge, and together they provide a substantial set of clues about their rela-tionship to one another. Thoughneuroanatomists have been searching fora "RosettaStone"of human brain functionfor centuries, it has been far froma trivial task to sort out the significantfromthe incidental differences inbrain structure and it willtakeconsiderable effort just to identifY exactly what haschanged and how.Aswiththe problem of determining what isfundamentally different about language, an analysis of how the brain has responded to these influ-ences willrequire us to delve well below brain sizes and superficial differ-encesinbrainstructures,toprobetheprocessesthatbuildbrainsin embryos.Brainsare themostintricate and powerful computing devices on the planet. Linguistic communication is the most complex behavior known. Evolution is the epitome of inscrutability, indirectness, and opportunism-seldom following an obvious or elegant path.Now wemust throw into this already dauntingmixof problemstheequally perplexingproblemof ex-plaining symbolic reference. A puzzle of such magnitude is unlikely to have aneasy solution,nor do I imagine that what little evidence we have issuf-ficient to do any more than begin the process of hunting in the right places formoreclues.But arranging the cluenthe appropriate order isa first step, and considering an old problem fromovel perspective isoften the best way to escape themaze of assumptionstlireventsusfromrecog-nizing the obvious. Perhaps by juxtaposing these linKesteries from dif-ferentdomainswewillrecognizethecommonthreadof logicthatruns throughthemall.Likethe famousRosettaStone,on which the same text was written in three radically different scripts,these pieces of the cognitive and neural puzzle, aligned side by side,may enable us to discover how each translates the other. CHAPTER TWO ALoss for Words Ifthe only tool you have is a hammer,youtend totreat everything asif it were a nail. -Abraham Maslow Gymnastics of the Mind Weoften findthat anapparently simpletask isdifficult,not because it is cOIpplicated but because it is awkward. Sometimes, for instance, you just lack the right toolsfor the job.Nomatter how easy it isin principle to tighten a screw, if it has a slotted head (-)and all you have isa Phillipsscrewdriver (+),forgetit.Thisisafamiliarsourceof difficulty in physical activities. We humans are unprepared to perform a number of tasks that other species perform withease.Other species can boast such adap-tations as streamlining and finsfor swimming, large flexiblemobile surfaces forglidingand flying,clawsfor clinging to thetrunks of treesor snagging prey,or sharp caninesfor tearing flesh.Attempt a task that youare poorly suited for and your performance isat best inelegant and clumsy, and often it just doesn't succeed. Suchawkwardnessis,inessence,theoppositeof preadaptation-that lucky chance of evolution when preexistingbody parts COincidentally hap-pentobe predisposed for anoveladaptational challenge.Incomparison, thiskindof built-inineptnessmightbe called"pre-maladaptation."Pre-maladaptation accounts for our difficulties trying to sleepstanding up,cut meat without a knife, or distinguish friends from foes by their smell, though certain other species do each well.In the same way that wefind certain ac-tions or movements to be impossible unaided,or find ourselves maladroit atmanipulating certain objects,certainkindsof mental taskscanalsobe unwieldy for a brain predisposed to different sorts of analyses.Inaddition, certain mental predispositions that serve well insomedomains can get in thewayof accomplishing otherwisetrivialtasksthat requireanew per-spective or logiC.Cognitive pre-maladaptations might also include predis-positions to behave inappropriately, in ways that are opposite to how events tend to occur,or preferences that lead usto pay attentionto irrelevant or misleading details.Thisis,of course,the secret to many magicians' tricks and the basisfor animals' protective coloration. Thesecouldmake anoth-erwisesimple task difficultby misdirecting attentionand interfering with appropriate actions. Learning is not anyone general process. Learning always occurs in a par-ticular context, involving particular sensesand types of motor actions,and certain waysof organizing the information involved.The process of learn-ing isalsonot just committing anassociationtomemory under the influ-ence of reinforcement.Leaming involves figuring out what isrelevant and then figuring out how the relevant variables are associated. It involves sort-ing and organizing, and sometimes recodingwhat we have already learned. When a pigeon learns that pecking at a red-lighted button willcause food tobedelivered and that pecking at ablue-lighted button willshut off an aversive noise, it does not just commit these links to memory, it also learns to ignorea great dealthat isirrelevant:the talking inthebackground, the timeof day,the temperature of the room,the pecking of neighboringpi-geons, odors that periodically waft by,and the other lights and buttons and structures of the cage.Successor failureat learning and problem solving depends on habits of attention, what we find salient and what we don't tend to notice, and how eaSily one can override and reprogram these tendencies. Therearealso internal distractions,inparticular thosethat arisefrom past learning experiences.Interference effectsof thissort are a common experience. Old associations provide a sort of repository of hypotheses.In many circumstances in the real world, events with similar features will pro-duce similar outcomes.Generalizing fromresemblances of present to past associations can often provide useful shortcuts that avoid wasting long pe-riods of trial-and-error learning.Sometimes, however,superficial similari-ties are just that, and the responses they elicit are off track.If the habits are strong, the opportunitiesto compare different habits are few,or the feed-back about incorrect responses is weak, forgetting the old in order to learn the new can become a practical impossibility. Asa result, certain problems can be difflcultbecause they require usto think in waysthat are not typi-cal:to infer what's missing, to work back from an end, to assume the oppo-site of what we want to demonstrate, and so on. Trick questions and riddles stumpus,notbecausethe questioniscomplicated but becausethesolu-tioniscounterintuitive. Jokesareendlesslyfascinatingbecausethey pro-videaninsidiously logicalpunchline thatwe are unableto anticipate or predict. A successful punchlinecannot be just a nonseqUitur,itmust fol-low ineVitably and obviously, if only after the fact.The minor cognitive im-plOSion summed up by "Now I get it" is in part an experience of admiration at being fooled so well, so simply, and yet so logically. As with a magic trick, the key to a good joke ismisdirection; and the mosteffective misdirection isself-imposed. The ways we naturally think about the subject of the joke are preCisely the causes of our capacity to be fooled. Intelligence tests and "brain teasers" often challengeour capability to think outside of self-imposed and overly narrow contexts, and we consider the facility to make leaps of logic a mark of intelligence.Indeed, the crite-rionforcallingsomething a work of geniusisseldomitscomplexity,but rather how innovative the approachand how many others were unable to see it. This iswhy some of the great discoveries inmany fleldshavebeen made by people outside the inner circle of true experts, because the experts are often too acquainted with a problem to see it in a novel way..As the great biologistThomasHuxleyissaidto haveexclaimedafter learning of Dar-win's theory of natural selection, "How stupid of me not to have thought of that!"what we do best worksagainst us. The difficulties that a problem poses are a functionof both its intrin,sic complexity and the leamer's predispositions to attend to itsmost relevant aspects.Comparative psychologists have long struggled tountangle ques-tions of general learning ability from specific learning abilities in different species.Asthey havebecomemoresensitivetotheproblemof under-standing a species' capabilities in the context of its natural environment, they are finding correlated differences in learning. Animal breeders have noticed these patterns for generations. Some breeds of dogs make b