Self-Representation in Nervous Systems Author(s): Patricia S. Churchland Source: Science, New Series, Vol. 296, No. 5566 (Apr. 12, 2002), pp. 308-310 Published by: American Association for the Advancement of Science Stable URL: http://www.jstor.org/stable/3076513 Accessed: 19/09/2010 22:27 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://links.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://links.jstor.org/action/showPublisher?publisherCode=aaas. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science.
http://links.jstor.org
REFLECTIONS been noted by immunologistsgrapplingwith the origin of adaptiveimmunity(1, 31), are a consequenceof similarselective pressuresfor diversificationand co-evolutionof recognition functionsto retainaffinitybetween interaction partners. A hallmarkof these specific recognition systems is that their genes are subject to intense diversifyingselection. Largenumbers of alleles are commonly found, and extraordinarilyhigh levels of intraspecificpolymorphism are typically achieved, in some cases resulting from acceleratedrates of evolution (18, 32). Due to balancing selection, polymorphismsin these genes can persist for long periods of time and often predate species diversification.Trans-speciespolymorphisms have been describedin the MHC (33) and in SI systems (34), and in both cases, divergence of some allelic lineages appears to have occurredat least 20 million years ago. Another emerging commonality between recognitionloci is their structuralheteromorphism, which apparentlyreduces intralocus recombinationevents and preventsdisruption of the co-adaptedgene complex. The crucifer S locus has been extensively restructuredby expansion or contractionof the physical distance between SRK and SCR, gene duplication, as well as rearrangementof these two genes relative to each other and to flanking markers(Fig. 1) (18, 35). Similarly,the MHC
ON SELF: IMMUNITY
AND BEYOND
has undergonefrequentgene duplicationsand deletions during its evolution (33), and the mating-type locus of Chlamydomonascontains a highly rearrangedregion that causes suppressionof recombinationover a 1-megabase chromosomalregion (36). Thus, in many respects, the challenges facing researchin the crucifer SI system are similar to those facing researchersof other recognition systems. Comparisons of these different systems should lead to insight into common selective pressures that drive the diversification and co-evolution of self/nonself recognitiongenes and shape the structure of their controlling loci. References and Notes 1. F. M. Burnet,Nature 232, 230 (1971). 2. D. De Nettancourt, Incompatibilityand Incongruityin Wild and Cultivated Plants (Springer-Verlag,Berlin, 2001). 3. J.J. Rudd,V. E.Franklin-Tong, New Phytol.151, 7 (2001). 4. K. Ida et al.,J. Mol. Biol. 314, 103 (2001). 5. I.S. Nou, M.Watanabe,A. Isogai,K.Hinata,Sex. Plant Reprod.6, 79 (1993). 6. C. R.Schopfer,M. E.Nasrallah,J. B. Nasrallah,Science 286, 1697 (1999). 7. H. Shiba et al., Plant Physiol. 125, 2095 (2001). 8. T. Takasakiet al., Nature 403, 913 (2000). 9. Y. Cui, Y.-M. Bi, N. Brugiere, M. Arnoldo, S. J. Rothstein, Proc. Natl. Acad. Sci. U.S.A. 97, 3713 (2000). 10. A. P. Kachroo,C. R. Schopfer, M. E. Nasrallah,J. B. Nasrallah,Science 293, 1824 (2001). 11. S. Takayamaet al., Nature 413, 534 (2001). 12. M. Kusabaet al., Plant Cell 13, 627 (2001). 13. J. C. Stein, B. Howlett, D. C. Boyes, M. E. Nasrallah,
14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38.
J. B. Nasrallah,Proc. Natl. Acad. Sci. U.SA. 88, 8816 (1991). S. Takayamaet al., Proc. Natl. Acad. Sci. U.SA. 97, 1920 (2000). S.-H. Shiu, A. B. Bleecker,Proc. Natl. Acad. Sci. U.SA. 98, 10763 (2001). A. L.Hughes, Cell. Mol. LifeSci. 56, 94 (1999). V. Vanoosthuyse, C. Miege, C. Dumas, J. M. Cock, Plant Mol. Biol. 46, 17 (2001). J. B. Nasrallah,Curr.Opin. Plant Biol. 3, 368 (2000). J.-L.Giranton,C. Dumas, J. M. Cock, T. Gaude, Proc. Natl. Acad. Sci. U.SA. 97, 3759 (2000). D. Cabrillac,J. M. Cock, C. Dumas, T. Gaude, Nature 410, 220 (2001). S. L.Stone, M.Arnoldo,D. R.Goring,Science 26, 1729 (1999). C. Azevedo, M.J. Santos-Rosa,K.Shirasu,TrendsPlant Sci. 6, 354 (2001). T. Nishio, M. Kusaba,Ann. Bot. 85 (Suppl. A), 141 (2000). M. Watanabe et al., FEBSLett. 473, 139 (2000). K. Hatakeyamaet al., PlantJ. 26, 69 (2001). M.Kusaba,C.-W.Tung,M.E.Nasrallah,J. B. Nasrallah, Plant Physiol.128, 17 (2002). M. K. Uyenoyama, Y. Zhang, E. Newbigin, Genetics 157, 1805 (2001). D. P. Matton et al., Plant Cell 11, 2087 (1999). D. Charlesworth,Curr.Biol. 10, R184 (2000). A.J. Brown,L A. Casselton,TrendsGenet.17, 393 (2001). J. Klein,Natural Historyof the MajorHistocompatibility Complex (Wiley, New York,1986). P. J. Ferris,C. Pavlovic,S. Fabry,U. W. Goodenough, Proc. Natl. Acad. Sci. U.S.A.94, 8634 (1997). J. Klein,A. Sato, S. Nagl, C. O'hUigin,Annu.Rev.Ecol. Syst. 29, 1 (1998). M. Uyenoyama, Genetics 139, 975 (1995). D. C. Boyes, M. E. Nasrallah,J. Vrebalov,J. B. Nasrallah, Plant Cell 9, 237 (1997). P. J. Ferris,E. V. Armbrust,U. W. Goodenough, Genetics 160, 181 (2002). J. B. Nasrallah,data not shown. Supportedby grants from the NIH, NSF,and USDA.
in Nervous Self-Representation Systems Patricia S. Churchland* The brain's earliest self-representationalcapacities arose as evolution found neuralnetworksolutions for coordinatingand regulatinginner-body signals, thereby improvingbehavioralstrategies. Additionalflexibility in organizingcoherent behavioraloptions emerges from neuralmodels that represent some of the brain's inner states as states of its body, while representingother signals as perceptions of the external world. Brains manipulate inner models to predict the distinct consequences in the externalworldof distinct behavioraloptions. The self thus turns out to be identifiablenot with a nonphysicalsoul, but ratherwith a set of representational capacities of the physicalbrain. What Is "the Self"? Descartes proposed that the self is not identical with one's body, or indeed, with any physical thing. Instead,he famously concluded that the essential self-the self one means when one thinks, "I exist"-is a nonphysical, consciousthing. At this stage of scientificdevelopment,the Cartesianapproachis unsatisfactory for three reasons: (i) psychological PhilosophyDepartment0119, Universityof California, San Diego, LaJolla,CA92093, USA. *To whom correspondenceshould be addressed.Email pschurchland@ucsd.edu
308
functions generally, including conscious thoughtssuch as "I exist,"are activitiesof the physicalbrain(1, 2); (ii) aspectsof self-regulation (e.g., inhibitingsexual inclinations),and self-cognition(e.g., knowing where I stand in my clan's dominancehierarchy),may be nonconscious(3); and (iii) as the Scottishphilosopher David Hume (1711-1776) realized,there is in any case no introspectiveexperienceof the "self" as a distinctthing apartfrom the body (4). Introspection,Hume concluded, reveals only a continuouslychanging flux of visual perceptions,sounds, smells, emotions,memories, thoughts,feelings of fatigue,and so forth.
12 APRIL2002
VOL296
To identifythe phenomenonthat we want explained,it is useful to startwith the idea that one's self-conceptis a set of organizational tools for "coherencing"the brain'splans, decisions, and perceptions.Thus, if a brick falls on my foot, I know the pain is mine. I know without pausingto figure it out that "this body is my own,"andthata decisionto fightratherthanflee is a decision affectingmy body's painful encounterwith the body of another.If I scold myself aboutjaywalking,I know that it is me talkingto myself. We know that if we fail to plan for futurecontingencies,our futureselves may suffer,and we care now aboutthat future self. Sometimeswe use "myself"to mean"'my body,"as when we say "I weighedmyself."By contrast,when we say "I deceivedmyself,"we arenot referringto ourphysicalbodies.We talk of our social and our privateselves, of discovering and realizing ourselves, of self-control, and self-denial(5). self-improvement, Thisremarkablydiverserangeof uses of the self-conceptmotivatesrecastingproblemsabout ca"the self" in terms of self-representational pacitiesof thebrain.Doing so deflatesthe temp-
SCIENCEwww.sciencemag.org
REFLECTIONS tationto thinkof the self as a singularentityand encouragesthe idea that self-representinginvolves a pluralityof functions,each having a rangeof shades,levels, and degrees.Further,it broadensthe inquirybeyond humansto other species,suggestingthatvaryinglevels of coherencing operatein all nervous systems of any significantcomplexity.The reformulationalso sets the stage for designingexperimentsto determinemore preciselythe types of self-representationsnervoussystemshave, how they are connectedto one another,andthe natureof their neuralsubstrates(6). The expectationthatthe brainandbehavioral scienceswill eventuallyunderstand the nature of self-representational capacitiesis not universally shared.Traditionalists preferto hive off the fundamentalquestions about the self or consciousness as philosophicalin the "armchairsenses of the only"or "forever-beyond-science" term (7). The dominantideology in academic philosophy, functionalism,acknowledges the relevance of the behavioralsciences, but discountsthe neurosciencesas largelyirrelevantto making progress in understandingthe higher functions (8). The functionalistrationaledepends on an allegedly close analogy between psychologicalprocesses and runningsoftware on a computer.Accordingto the analogy,the brainis only the hardwareon which the cognitive softwarehappensto run (9). The brainis thus deemeda mereimplementation of the software. The corollaryis that understandingthe hardwareis thereforeunimportant, by andlarge, in figuringoutthe software.Thoughthe analogy between cognitive functionsand runningsoftware is not close but feeble, and though the corollaryfails to follow, functionalismretains considerablepopularitybeyond the bordersof neuroscience(10). Mysticismand functionalismnotwithstanding, questions about self-representationare steadily shiftinginto the provinceof the brain and cognitive sciences. This shift is part of a generaltrendenabledby the scientificadvances in the 20th centuryat all levels of brainorganizationfromsynapsesto systems.Theseadvances, alongwith improvementsin technology,data analysis, and computationalmodeling, have meant that virtuallyall topics concerningthe mind are now vigorouslyexploredat the interface of neuroscience,cognitivescience,andphilosophy.Thishas been the fortune,for example, of colorperception(11), autobiographical memory (12, 13), the emotions(3, 14-16), decisionmaking(3, 12, 14, 17, 18), sleep and dreaming (19), and consciousness(6, 12, 15, 20). As in any science, some discoveries force a more enlightened articulationof the very questions themselves. For example, the splitbrain studies revealed that interruptinginformation flow between the two hemispheresby surgical section of the cerebralcommissures gives rise to striking disconnection effects; that is, the perceptions and decisions of one
ON
SELF:
IMMUNITY
AND
hemisphereare disconnectedfromthose of its counterparthemisphere (21). This implied that the "unity of self," advertisedby some philosophersas a "transcendental"necessity, was actually subjectto anatomicalmanipulation. The results implied that such unity and coherence as exist in one's self-conception depend not on transcendental necessity, whatever that might be, but on neuronal organization. A recharacterizationof the phenomenonof unity of the self was consequently occasioned by the new empirical data. A standard principle, illustrated by the split-brainresults, is that the definition of the phenomenonto be explained coevolves with experimentaldiscoveries. In the early stages of the scientific attackon any problem,accurate definition of the phenomenon is hampered precisely because not enough is known to permitan accuratedefinition.A pragmatic strategy is to begin by studying those cases agreed to be obvious examples of the phenomenon. Powered by this agreement,provisional, rough characterizationscan leverage the science's first stages, with refinementsin the phenomenon'sdefinition emerging as the surroundingfacts become clear. From a historical perspective, the interdependenceof definition and discovery typifies the transformation of assorted problems of what was originally "pure"philosophy (e.g., the nature of fire, space, matter, life, the cosmos) into problems of the experimentalsciences (2, 6). Self-Representational Capacities In the brain, some networksare involved in representingthings in the externalworld, such as the face of GrouchoMarxor a loomingbus. Other networksrepresentstates of the body, such as its postureor its need for water.Some networks operate on other representations, such as knowing yielding meta-representations that my need to flee is more urgentthan my need for water,knowingthatJohndislikesme, or rememberingthat John hit me. Neural networks engagedin integratingsuch meta-representationsareprobablytheones mostrelevantto questionsaboutself-representation. Self-representations may be widely distributedacrossbrainstructures, coordinatedonly on an "as-needed"basis, and arrangedin a loose and loopy hierarchy.We see the slow emergence and elaborationof self-representational capacitiesin children(22), andthe tragicfading of these capacitiesin patientswith dementia. Despite largegaps in ourknowledge,humanas well as animalstudieshave made it possibleto begin to distinguishdifferenttypes of self-representationalfunctions,and in some instances, to identify,albeit in generalterms,theirneural dependencies. Self-representational capacitiesincluderepresenting the internalmilieu and viscera via chemicaland neuralpathwaysaimedlargelyat the brainstemand hypothalamus;representing
www.sciencemag.org SCIENCE VOL296
BEYOND musculoskeletalstructuresvia the somaticsensory system; representing autobiographical events via medialtemporallobe structures;deferringgratificationandcontrollingimpulsesvia andrepreprefrontallobe andlimbicstructures; sentingthe sequenceof actionsto take next, as well as representingwhere one is in space-time andthe social order. Studiesof humanpatientsreflectthe multidimensionalityof self-representation by showing thatparticularfunctionscan be sparedwhen othersare impaired.For example,a subclassof amnesic subjects with bilaterallesions in the hippocampaland associatedcorticalstructures areunableto acquirenew knowledgeandhave lost essentially all autobiographicalinformation. For example,the patientR.B. lives essentially within a moving 40-s time bin (3). AlthoughR.B. does sufferdiminishedself-understanding,he neverthelessretainsmanyelements of normalself capacities,includingself-control in social situationsand the fluent and correct use of "I." He also knows his currentbody configurationand status,and he can engage in self-imagery,identify feelings such as happiness, and show sympathywith the distressof others. The existence of such amnesics is a counterexampleto the seemingly obvious hypothesisthatone's self is constitutedby personal narrative(9). Schizophrenia,known to involve decreased prefrontalactivityand increasedstriatalactivity (23), presentsa differentdimensionof self-dysfunction.Duringa floridepisode,a schizophrenic mayhavegood autobiographical memory,but sufferdeep confusionaboutself/nonselfboundaries, e.g., respondingto a tactile stimulusby claimingthatthe sensationbelongsto someone else or thatit exists somewhereoutsideof him. Auditoryhallucinations,often considereddiagnostic of schizophrenia,exemplify integrative failure.The "voices"appearto be the patients' own thoughtsor innerspeech,but they are not represented,and thus not recognized,as such (24, 25). The anestheticketamine and drugs such as LSD can triggersimilarphenomena. A patient with lesions in right parietal cortex, resulting in loss of sensation and movement on the left side of the body, may firmlydeny thather left limbs are in fact hers. On occasion, a patient with limb denial will use the normal right arm to try to throw the paralyzedleft leg out of the bed, insisting it is alien. Despite suffering compromised bodyrepresentation,the patients may nevertheless have normal autobiographical memory as well as other self-representationalfunctions such as knowing whether they feel bored or hungry. Patients with lesions in the anterior cingulate region may exhibit alien hand syndrome.In these cases, the contralesionalhand will sometimes behave as though it is independentlycontrolled.Patientswith alien hand syndromesometimes controltheir embarrassing alien hand with verbal commands.
12 APRIL2002
309
REFLECTIONS
ON SELF: IMMUNITY
AND BEYOND
Self-regulatingfunctions can also be selectively impaired.Lesions in prefrontalcortex, especially in the ventromedial region, have been followed by significantchanges in self-control, and particularlyin the capacity to inhibit unwise impulses, despite normal functioning of many other self-representational capacities. Personality changes commonly occur with prefrontaldamage. Hitherto quiet and self-controlled, a person with lesions in the ventromedialregion of frontal cortex is apt to be more reckless in decisionmaking, impairedin impulse control, and socially insensitive (3, 17, 18).
answer to this question: What motor command should be issued to get my arm to contact the plum? Taking the command-proposal, the forwardmodel calculates the error by runningthe command on a neuronalemulator,and the inverse model respondsto the errorsignal with an upgradedcommand.Emulation is faster and safer than real-world feedback. Assuming the forwardand inverse models are also capable of learning, this organizationcan be very efficient in acquiringa wide range of sensorimotorskills. With sufficient access to backgroundknowledge, goal priorities, and current sensory information, emulatorscan compute accuratesolutions to Evolutionof Self-Representational complex motor problems. Capacities Rudimentaryneuronalemulators,groundThe most fundamentalof the self-representa- ed in the basic coordinatingand self-regulattional capacitiesprobablyarose as evolution ing functions,can in turnbe upgradedto yield stumbledon solutions for coordinatinginner- fancier inner models of planning. Emulators body signals to generate survival-appropriate can facilitate making an appropriatemoveinnerregulation.The basic coordinationprob- ment after the target has become invisible, lems for all animalsderivefromthe problemof perhapsbecause the prey is in a cavity or the whatto do next. Pain signalsshouldbe coordi- predator is sneaking up on the prey. More nated with withdrawal,not with approach. generally, with appropriateconnectivity, an Thirstsignalsshouldbe coordinatedwithwater- emulator could run off-line to plan for the seeking,not with fleeing,unlessa presentthreat long-term future, thus deploying extended takeshigherpriority.Homeostaticfunctionsand body-image manipulation.Additional modithe abilityto switchbetweenthe differentinter- fication permits off-line emulation of cogninal configurationfor fight and flight from that tive states. For example, when planning the needed for rest and digest requirecoordinated details of a raid, one may imagine oneself controlof heart,lungs, viscera,liver, and adre- feeling anxiety while stalking the enemy nal medulla.Body-statesignalshave to be inte- camp, assessing the attentivenessof the camp grated, options evaluated,and choices made, guards,formulatingspecific intentionsto outsince the organismneeds to act as a coherent fox wary guards,and so on. Like body-image whole, not as a group of independentsystems manipulationused in planninga climb, this is with competinginterests. mind-imagemanipulationused in planninga complex, extended me-them encounter(27).
The Neural Platform
The most basic level of innercoordinationand regulationoccurs in the brainstem,anchoring whatDamasiorefersto as "theprotoself"(12). In vertebrates,the brainstem-hypothalamic axis is the site of convergenceof signals from the viscera,internalmilieu,andthe somaticsensory system.Also locatedin the brainstemarenuclei thatregulatevital functions,sleep-wakefulness cycles, arousal, attention,and the emotions. This level of integration,sharedacross many species, is the nonconscious neurobiological platformforhigherlevels of self-representation. Other, more complicatedand flexible aspects of the self demandgreatercomputational resources.Wolpert(26) and Grush(27) have proposedthat increasedaccuracyin planning and execution of movement in space-timeis achieved by cortical models of the body in relationto its environment.Roughly, a somewhat sloppy inversemodel is connectedto an error-predictingforwardmodel, and the two convergeon a good answerto the problemof how to move a many-limbedbody in just the rightway at just the righttime. If, for example, the goal is to reach a plum, the inverse model gives a first-pass 310
Consciousness and Self-Representation An appealing hypothesis defended by Damasio (12) is that the self/nonself distinction, though originally designed to support coherencing, is ultimately responsible for consciousness. According to this view, a brain whose wiring enables it to distinguish between inner-worldrepresentationsand outer-worldrepresentationsand to build a metarepresentationalmodel of the relation between outer and inner entities is a brain enjoying some degree of consciousness. Thus, such a system could representthe relation between the thistle and itself as "that (outer) thing causes me (inner) pain." Conceivably, as wiring modifications enable increasinglysophisticatedsimulationand deliberation, the self-representationalapparatus becomes correspondingly more elaborate, and therewith the self/not-self apparatus. On this hypothesis, the degrees or levels of conscious awareness are upgraded in tandem with the self-representational upgrades. Thus, chimpanzees, but not frogs, know whether they can be seen by a sub-
12 APRIL2002
VOL296
ordinate female but not the dominant male. Infant human development studies and nonhuman primate studies support these hypotheses (28, 29). Whetherneuroscience can build on these foundationsto discover full and detailed explanations of all self-representationalphenomena remains to be seen. Still, unpredictability obscures the destiny of essentially all neurobiologicalpuzzles, includingnoncognitive functions such as thermoregulation.An abiding challenge in neuroscience is to discover the basic principlesgoverningthe integration of information at various levels of brainorganizationand at various time scales. This challenge is not confined to the neuroscience of self-representation,but confronts neuroscience generally. References and Notes 1. P. M. Churchland,Matter and Consciousness (MIT Press,Cambridge,MA,ed. 2, 1988). 2. P. S. Churchland,Neurophilosophy(MITPress, Cambridge, MA, 1986). 3. A. R. Damasio, Descartes' Error (Grossett/Putnam, New York,1994). 4. D. Hume,A Treatiseof HumanNature (1739); modern edition by L. A. Selby-Bigge, Ed. (Clarendon Press, Oxford, 1888). 5. G. Lakoff,M. Johnson, Philosophyin the Flesh (Basic Books, New York,1999). 6. P. S. Churchland,Brain-Wise:Studiesin Neurophilosophy (MITPress, Cambridge,MA,in press). 7. See the essays by Z. Vendlerand by C. McGinn,in The Mind-BodyProblem:A Guide to the CurrentDebate, R. Warner,T. Szubka,Eds. (Blackwell,Oxford, 1994). 8. J. A. Fodor,Synthese28, 97 (1974). Forcriticismof this T. J. Sejnowski,TheCompuview, see P. S. Churchland, tationalBrain(MITPress,Cambridge,MA,1992). 9. D. C. Dennett, ConsciousnessExplained(LittleBrown, Boston, 1991). 10. S. Pinker,How the Mind Works(Norton, New York, 1997). 11. S. E. Palmer, Vision Science (MITPress, Cambridge, MA, 1999). 12. A. R. Damasio, The Feeling of What Happens (Harcourt Brace,New York,1999). 13. L. R. Squire, E. R. Kandel, Memory: From Mind to Molecules (Freeman,New York,1999). 14. J. Le Doux, The EmotionalBrain(Simon & Schuster, New York,1996). 15. R. R. Llinas,I of the Vortex(MITPress,Cambridge,MA 2001). 16. J. Panksepp, Affective Neuroscience (Oxford Univ. Press, New York,1998). 17. S. W. Andersonet al., NatureNeurosci.2, 1032 (1999). 18. A. Becharaet al., Science 275, 1293 (1997). 19. A. R. Hobson, Sleep (Freeman,New York,1989). 20. F. Crick, C. Koch, in Problems in Systems Neuroscience, J. L.van Hemmen, T. J. Sejnowski,Eds. (Oxford Univ. Press, Oxford, in press). 21. R. W. Sperry,Science 217, 1223 (1982). 22. S. Harter,The Constructionof the Self: A Developmental Perspective (Guilford,New York,1999). 23. A. Meyer-Lindenberget al., Nature Neurosci. 5, 267 (2002). 24. C. D. Frith.TheCognitiveNeuroscienceof Schizophrenia (Eribaum,Hillsdale,NJ, 1992). 25. G. L.Stephens, G. Graham,WhenSelf-Consciousness Breaks(MITPress,Cambridge,MA,2000). 26. D. M. Wolpert et al., Science 269, 1880 (1995). 27. R. Grush,Philos. Psychol. 10, 5 (1997). 28. A. N. Schore,Affect Regulationand the Originof the Self (Eribaum,Hillsdale,NJ, 1994). 29. M.Tomasello,J. Call,PrimateCognition(OxfordUniv. Press, New York,1997). 30. Specialthanks to P. M. Churchland,M. Churchland,A. Churchland,J. Cohen, F. Crick,A. Damasio, D. Eagleman, and R. Grush.
SCIENCEwww.sciencemag.org