Bodily mimesis as “the missing link” in human cognitive ... - CiteSeerX

Bodily mimesisas“the missing link”inhuman cognitive evolutionJordan Zlatev, Tomas Persson and Peter GärdenforsLund University Cognitive Studies 121,2005

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 3Finally, the units of language stand in complexinternal semantic relations involving e.g.complementarity (Deacon 1997) and grammaticalrelations involving hierarchical structure andrecursivity (Hauser, Chomsky and Fitch 2002;Givón 2002). Nothing of the kind has beenshown to exist in natural animal communication,though aspects of both semantic relations andgrammatical patterns are not beyond the graspof language-taught apes (Savage-Rumbaugh andLewin 1994).We may conclude that the six general differencesbetween language and animal communicationdiscussed in this section and summarized inTable 1 define two very different types of semioticsystems. Even though one may argue thatfor each of the individual features there may besome forms of animal communication thatwould appear to be “intermediary”, e.g. theflexible interpretations of chimpanzee alarm callsby Diana monkeys (Zuberbühler 2000), no animalin the wild has anything approaching the socially transmitted,voluntarily controlled, contextually flexible, triadicsemiotic system that is language.As pointed out in the introduction, the difficultquestion is to account why and how such adistinctively “odd” (from a biological perspective)communication system emerged in the firstplace. Proposals in the currently exploding literatureon language origins vary widely on whatthey find to be the crucial feature of humancognition that could “bridge the gap” to language:processing of recursivity (Hauser, Chomskyand Fitch 2002; Bickerton 2003), enhancementsin social cognition (Tomasello 1999;Tomasello et al. in press), mechanisms supportingthe learning and use of symbolic relations(Deacon 1997, 2003). In the present quest forthe Holy Grail of explaining human uniquenessthat tends to focus on the origin of languagedifferent researchers tend to offer their favoritecandidate for “the missing link” and do theirbest to support their hypothesis, often at theprice of disregarding a good deal of evidence(see Johansson 2005). Our overall project similarlyaims to resolve the puzzle of what makeshuman beings cognitively and communicativelyspecial, but differs from the majority in (a) notfocusing on language but on a skill that we argueis a prerequisite for language and (b) drawingcognitive interpretations and toward more low-levelmechanisms linked to the particular stimulus featuresof the environment.” (Hauser 1996: 59)from a large number of empirical and theoreticalstudies.Table 1. Six characteristics of animal communicationand human language1. Degree oflearning2. Consciouscontrol3. Contextuality4. Interpretation5. Communicativerelations6. Systematicity(internal relationsandcombinatorics)Animal communication(inthe wild)None (“innate”)orhighly limitedNone orhighly limitedTied to aparticularcontext(stimulussetting)(Relatively)fixed responseDyadic:- Cue (environment,subject)- Signal (subject,recipient)None, orvery limitedHuman languageExtensiveHighFlexible, relativelyindependent fromspecific contextFlexible, “negotiable”Triadic:Speaker-addresseereferentHighAnimals and human beings differ not only intheir means of communication, but in (apparently)non-communicative aspects of cognitionsuch as planning, symbolic thought, “theory ofmind”, etc. In other words, there is a parallelbetween the “communicative gap” and the“cognitive gap”. What is the source of this parallelism?When it comes to ontogeny, a possibleexplanation for the source of the “higher functions”of human beings is the classical proposallaid out by Vygotsky (1978) that the latter can beseen as resulting from the internalization of interpersonalrelations and representations, or as expressed byVygotsky in an often quoted passage:Every function in the child’s cultural developmentappears twice: first, on the sociallevel, and later, on the individual level; firstbetween people (interpsychological), and theninside the child (intrapsychological). This appliesequally to voluntary attention, to logicalmemory, and to the formation of concepts.

4Bodily mimesis as “the missing link” in human cognitive evolutionAll the higher functions originate as actual relationsbetween human individuals. (Vygotsky1978: 57)We wish to suggest that an evolutionary analogto this process, i.e. a process of hominid selfenculturationin which more complex forms ofcommunication created selection pressures formore complex forms of cognition, can help usunderstand the close connections between thetwo “gaps”. Such a socio-cultural perspective onhuman cognition and its evolutionary originsdoes not conflict with, but rather presupposesthe crucial underlying role of our general primatecognition (Donald 1991; Tomasello 1999; Zlatev2003). In this article we will adopt such a bioculturalperspective as a general “working hypothesis”,without thereby arguing against moreindividual-based alternatives (e.g. Gärdenfors2003). Ultimately, we will need a synthesis of thesocial and the individual aspects of human cognitionin order to make sense of the nature ofour major theoretical concept that is both interandintra-personal, as we show below.3. Bodily mimesisThe concept of mimesis goes back at least to Aristoteles,but as far as human cognitive evolutionis concerned, we take our departure from themimesis hypothesis of human origins presented byDonald (1991, 1998, 2001). This hypothesisstates that a specific form of communication andcognition (and corresponding culture) mediatedbetween those of the common ape-ancestor andmodern humans based on “the ability to produceconscious, self-initiated, representationalacts that are intentional but not linguistic.”(Donald 1991: 168) In brief, Donald proposesthat while ape culture is based on “episodic cognition”and associational learning, early Homo –most likely Homo erectus/ergaster considering therelative jump in brain size and material culturewitnessed at this stage of hominid evolution –evolved a new form of cognition. What supportedthis was the fact that the body could beused volitionally to do what somebody else isdoing (imitation), to represent external eventsfor the purpose of communication or thought(mime, gesture) and to rehearse a given skill bymatching performance to a goal.The most important features that Donald attributesto mimesis are: (a) reference: mimetic representationsstand for something (for someone);(b) intentionality: this referential relationship isgrasped both by the “mimer” and the interpreter;(c) communicativity: the relationship in (b)can be used for the purpose of conveyingthoughts; (d) autocuing: production is voluntary asopposed to instinctive; and (e) generativity: the“ability to “parse” one’s own motor actions intocomponents and then recombine these componentsin various ways” (Donald 1991: 171).It can be observed that features (a)-(e) definea system of communication that is intermediatebetween animal communication and language, ascharacterized in Section 2 and Table 1. Like thelatter it is learned, flexible and possibly triadic.At the same time, it lacks the following criticalfeatures of human language:• Conventionality: not just shared, but knownto be shared and thus a part of mutualknowledge (Itkonen 1978; Clark 1996); Notthat this is not the same as arbitrariness (seebelow), even though to notions are oftenmistakenly conflated. 2• Arbitrariness: the semiotic ground for theexpression-content relation requires neithersimilarity, nor contiguity (Peirce 1931-1935;Sonesson in press)• Extensive systematicity of the internal relationsamong signs (Deacon 1997; Zlatev2003).Thus the hypothesis that mimesis plays the roleof a “missing link” (in the words of Donaldhimself) in human cognitive evolution is a prioriattractive. Furthermore, it has been backed upby evidence from archeology, neurobiology,cognitive psychology and developmental psychology,such as the presence of a “mimeticstage” in human ontogeny and the presence ofmimetic thought in patients with total aphasia,the homology between “mirror neuron” systemsin monkeys used for action recognition, andstructures for the control of imitation, mentalizingand language in human beings (e.g. Donald1991, 2001; Nelson 1996; Zlatev 2002, 2005;Corballis 2002).2 It is also completely distinct from the so-calledconventionalization of behaviors discussed in theethological literature, for which we reserve the termritualization (see Section 6).

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 5Nevertheless, the mimesis hypothesis hasbeen met with much resistance and criticism(Mitchell and Miles 1993; Tomasello 1993; etc)that can be summarized as follows: On the onehand, Donald’s theory underestimates the cognitionof non-human primates with respect totool-making and cultural traditions (Boesch2003) and their ability to understand others’intentions (Tomasello, Call and Hare 2003). Onthe other hand, features (a)-(e) attribute so muchrepresentational complexity to mimesis, that thisobviates the need for a second cognitive transitionto language (Laakso 1993). If this criticismis correct, then mimesis can hardly “bridge thegap” as required, since it, so to speak, both givestoo little to the apes and too much to Homo erectus,thus effectively creating a second gap: betweenapes and Homo erectus, without any indicationsof how it was bridged in evolution. Finally,Donald’s theory does not address the questionwhy mimesis evolved in the first place, and whyit later became superseded by language.Apart from these issues, Donald’s definitionof mimesis is unclear with respect to at least thefollowing points: Are the features (a)-(e) mentionedabove necessary and jointly sufficient orare some more central than others? Is there apriority for the communicative or the noncommunicativeforms of mimesis? In which wayshould one interpret the notoriously ambiguousterms “reference” and “intentionality”? Furthermore,if mimesis possesses all the features oflanguage except for phonetic realization asclaimed by Laakso (1993), it remains unclearwhat exactly makes it “not linguistic” in thewords of Donald quoted earlier. Hence, our firststep in elaborating the mimetic hypothesis ofhuman origins is to offer a more specific(re)definition of the concept, which we refer toas bodily mimesis in order to distinguish it fromthe broader concept with Aristotelian roots:(Def) Bodily mimesis: A particular act ofcognition or communication is an act of bodilymimesis if and only if:1. It involves a cross-modal mapping betweenproprioception (kinaesthetic experience) andexteroception (normally dominated by vision),unless proprioception is compromised.(Cross-modality)2. It is realized by bodily motion that is, or canbe, under conscious control. (Volition)3. The body (part) and its motion correspondto – either iconically or indexically – some action,object or event, but at the same time aredifferentiated from it by the subject. (Representation)4. The subject intends for the act to stand forsome action, object or event for an addressee,and for the addressee to appreciate this(Communicative sign function)But it is not an act of bodily mimesis if:5. The act is fully conventional, i.e. a part ofmutual knowledge, and breaks up(semi)compositionally into meaningful subactsthat relate systematically to each otherand to other similar acts. (Symbolicity)This definition resolves some of the problemsmentioned earlier. By stating that bodily mimesisinvolves a mapping between proprioception, involvingkinesthetic sense and the input from skinreceptors and serving as our basic form of selfknowledge(e.g. Edelman 1992; Gallagher 1995,2005) and exteroception (above all vision, butalso hearing and touch), this explains the sensein which bodily mimesis involves the body, includingspecific organs such as the vocal tract –to the extent that they are involved in proprioceptive-exteroceptivemapping. Since speechproduction, and possibly even speech comprehensionaccording to the “motor theory ofspeech perception” (Liberman and Mattingly1985) involve such a mapping, this implies thateven speech involves bodily mimesis, at thesame time as it transcends it due to symbolicity.By focusing on full conventionality and systematicityas distinguishing language from bodilymimesis, it becomes clear in which sense mimesisis “not linguistic”. Most importantly, however,the five criteria in the definition of bodilymimesis presented above allow us to classifyhuman and ape socio-cognitive skills along ascale consisting of at least four distinct stages.We refer to this as the mimesis hierarchy (see Table2).Acts which only fulfill the first condition, butnot the others can be defined as proto-mimetic.Neonatal face-mirroring and cross-modal matching(Meltzoff and Moore 1977, 1983, 1994) canbe regarded as such, and these have been witnessedin newborn chimpanzees as well (Myowa-Yamakoshi et al. 2004). On the other hand, bothdeferred imitation and mirror self-recognition fulfillconditions (1)-(3) and thus show a full form ofbodily mimesis. Skills such as these presupposethat the subject can both differentiate betweenhis own (felt) bodily representation, and the

6Bodily mimesis as “the missing link” in human cognitive evolutionentity that this representation corresponds to,and to see the first as standing for the latter. Thisis what Piaget (1945) called ‘the symbolic function’and argued that it appears at the end of thesensorimotor period of the child’s development.Sonesson (1989, in press) uses differentiation as thecrucial criterion to distinguish between truesigns, which display it, and pre-sign meanings,which do not. Note, however, that such representation,as in Piaget’s example of the childmodeling the opening and closing of a matchboxwith his mouth, is not intentionally communicative.Therefore, if condition (4) is not fulfilled itis bodily mimesis only of a dyadic sort. If it isfulfilled, then we have triadic mimesis, such as thatinvolved in pantomime or declarative pointing. Finally,condition (5) distinguishes between intentionalgestural communication, and signed language,which also possesses the properties ofconventionality and systematicity (Singleton,Goldwin-Meadow and McNeill 1995).Table 2. The mimesis hierarchy. Definitions of the fourevolutionary stages (for terms in italics, see the definitionof bodily mimesis in the text) and examples ofcorresponding types of acts.Stage Definition ExamplesProtomimesisDyadicmimesisTriadicmimesisJoint attention,declarativepointing, pantomimePostmimesisA bodily act involvingCross-modality withproprioception, butlacking Volition orRepresentation (or both)An interpersonal orintrapersonal bodilyact displaying Volitionand Representation, butnot Communicative signfunctionAs dyadic mimesis butalso involving Communicativesign functionAs triadic mimesis,but also involvingSymbolicityFacial expressions,bodilysynchronizationShared attention,imperativepointing, mirrorself-recognition,do-as-I-doimitationSigned languageIn the remainder of this article we shall apply theproposed mimesis hierarchy to the evolution(and to a lesser degree, the ontogeny) of threecognitive-communicative capabilities: imitation(Section 4), intersubjectivity (Section 5), and gestures(Section 6). By analyzing evidence from primatology,and to a smaller extent child development,we will identify levels within these domainsthat correspond to the levels of the mimesishierarchy. In other words, we considerwhether there are simple forms of these capacitiesthat are in essence proto-mimetic; whetherthere are forms that are dyadic mimetic; whetherthere is furthermore evidence for communicativeintentions, and thus for triadic mimesis.Finally we ask whether there are specific formsof imitation, intersubjectivity and gesture thatdepend on the acquisition of a conventionalsymbolic system, i.e. language, and are thus postmimetic.We will show that the progression protomimetic> dyadic-mimetic > triadic-mimetic > postmimetic,instead of the dichotomy animal signalingvs. human language outlined in Section 2,and in particular by distinguishing between dyadicand triadic mimesis, has implications for atheory of the origins of human cognitiveuniqueness. It may seem that such a theory goesagainst the stream of much current theorizing inevolutionary science since it invokes relativelydiscrete “stages” requiring qualitative transitions,while both the nature of evolutionary processesinvolving many small mutations and the availablefissile evidence seem to speak in favor of aprolonged gradual process (Johansson 2005).However, we hasten to note that there is nocontradiction between the stage-like theory thatwe will pursue and the evolutionary evidence.Firstly, it is completely possible to have relativelydiscrete transitions, or “punctuated equilibria”(Eldredge and Gould 1972) within a basicallygradual framework, i.e. without this implying anyform of “saltations” or sudden “macromutations”(Dessalles 2004). The likelihood forfairly rapid and abrupt changes is even higherwhen biological and cultural evolution interact,as has most likely been the case with humancognitive evolution including language (Deacon1997). Secondly, our notion of a mimesis hierarchy,as other similar constructs (Dennett 1996;Zlatev 2003), is meant to serve primarily as aconceptual role in helping us relate and compareprimate skills in somewhat different domains: itis not an evolutionary ladder for classifying differentspecies, which may very well have differentdegrees of the mimetic skills under discussion,and thus form much more of a cline. Withthese caveats we proceed with our analysis.

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 74. Mimesis and imitationChimpanzees, bonobos, gorillas and orangutanscan copy behavior to some extent and in seeminglyvarious ways. The literature on ape imitationin the broad sense of the term (see Whitenet al. 2004 for a recent review) can for our purposesbe grouped into the following main experimentalparadigms: Neonatal mimicking Instrumental tool-use Experiments with “artificial fruits” (puzzleboxes) “Do-as-I-do” tasks- in respect to body movements (“Simonsays”)- in manipulation of objects Ethological methods (i.e. field observationsof food processing techniques or indirectevidence from local traditions)Most obviously relevant for our discussion ofstages of bodily mimesis are those types of observationallearning in which an observer copiessomething of a model’s bodily actions, may it be in alaboratory tool-using situation, natural foragingsituation or mimicking in a do-as-I-do game.Other forms of observational or social learning,like emulation, available in several versions (Byrne1998; Custance, Whiten and Fredman 1999),response facilitation (see Byrne 1999), and stimulusandlocal enhancement (see Tomasello 1996), caninform observers about properties of the environmentand objects, or the connection betweencertain manipulations and outcomes, but do notinvolve copying bodily actions per se and thus itis not clear if they involve bodily mimesis.For our purposes, the collective term imitationpertains most clearly to the copying of bodilymotions (the do-as-I-do experimental paradigm)as well as to imitation on the action level in objectmanipulation. The notion of action level imitationwas originally contrasted to that of programlevel (Byrne and Russon 1998), the latter beingthe hierarchical organization of a series of actionsand the former the execution of an individualcomponent action. Note that imitation onthe program level can take place without imitationon action level. The program level has beenfurther divided by Whiten (1998) into beingeither of a sequential or a hierarchical kind. Anexample of sequence imitation from the artificialfruitexperiments would be to manipulate obstacleA before obstacle B, while an example ofimitation on the action level would be poking vs.pulling to remove a plug. If the subject copiesthe manual movement of the model, it is regardedas an instance of action-level imitation,but if the ape instead copies the movement ofthe plug, it is categorized by Custance et al.(1999) as object movement re-enactment. The latter isusually regarded as a form of emulation, wherethe end state of an event is achieved (or copied)but the subject invents his or her own way toachieve this end and the model is in principlesuperfluous. However, since object movementre-enactment is a special kind of emulationwhere at least the direction of the movement ofthe object is copied, in our analysis it wouldqualify as a form of bodily mimesis.There is also the hypothetical case where thesubject only learns that the plug can be movedand then proceeds to bring about that effect inwhatever way he or she prefers. To determinewhat is what, however, is a tricky business. Inthe latter case the subject might happen to bringabout by chance the same movement of the plugas the model showed, and thus give the impressionof object movement re-enactment. Furthermore,in both a false and in a true case ofobject movement re-enactment, the subject canalso happen to perform the same manual actionsas the model by chance and give the impressionof bodily imitation on the action level. At leastthree forms of observational learning can thusyield the same expression: simple emulation,object movement re-enactment and action levelimitation. The experiments with artificial fruits,some of which we will review below, are designedto tease apart these copying strategies.The failures of great apes to fully copy behaviorin several early studies led Tomasello,Kruger and Ratner (1993) to conclude that apesemulate, but cannot imitate, since imitation intheir view entails copying both of the goal andthe method to bring about that goal. ForTomasello and his collaborators imitation thushas a “theory of mind” component by definition,and on the basis of the imitation data, the grouphas argued that apes lack an understanding ofothers as intentional agents (e.g. Tomasello1999). Since then both the field of primate imitationstudies has expanded and the MPI Leipziggroup has modified their views on apes’ abilitiesto understand intentional agency (Tomasello etal. in press, see Section 5 for more discussion).

8Bodily mimesis as “the missing link” in human cognitive evolutionDonald’s definition of mimesis (presented inSection 3), comprising reference, intentionality, communicativity,autocuing and generativity, is difficult toapply to non-human primate imitation becauseobservational learning through copying has beenstudied in non-communicative tool using andpuzzle box situations, with few exceptions suchas the study by Tomasello et al. (1994) on chimpanzeegestures in which the authors did notfind any observational learning involved in functionalgesture acquisition. However, using ourdefinition (see Section 3), it is possible to applythe concept of bodily mimesis to findings in thefield of ape imitation. As we will show in thissection, these findings can be interpreted asevidence that apes are capable of at least protomimeticand dyadic mimetic imitation, and possiblyeven triadic mimetic imitation when a communicativesign function is added.4.1 Proto-mimetic imitationIn what we call proto-mimetic imitation there is(iconic) resemblance between the copy and thecopied, which is probably based on a basic formof identification with the model (see Section5.1). For this to be classified as proto-mimeticrather than dyadic-mimetic, there should be noclear evidence for volition or for differentiationbetween oneself and the model.The prime candidate for proto-mimetic imitationis neonatal facial mirroring. In the case ofhuman neonatal mirroring, the field is dividedbetween those who prefer a minimalist interpretationand those who argue for a more mentalisticone. For example, Anisfeld (1996) in a metaanalysisof experiments by Meltzoff and Mooreand Heimann argues that the infant’s response islimited to one facial movement, tongue protrusion,and not several. If so, it would be a largelyreflexive form of mirroring unlikely to requirevolition on the part of the infant, and thus wouldbe proto-mimetic. On the other hand Meltzoffand Moore (1977, 1983, 1994) have argued thatthe human infant not only displays volition (i.e.choice on whether to imitate or not, and abilityto mirror more than tongue protrusion), but thathe/she differentiates between him/herself and themodel. Meltzoff and Moore (1994) argue thatthere is evidence for deferred imitation in 6-week-old infants who imitated the action whenseeing the model again after a delay of 24 hours.While differentiation in itself does not requirethe kind of displacement that is witnessed indeferred imitation (see Sonesson in press), weagree with Piaget (1945) that the capacity fordeferred imitation rests on at least a simple formof representational activity. Thus, if the Meltzoff-Mooreinterpretation of neonatal imitationis correct, we must conclude that human childrenperform dyadic imitation (see below) froma very early age.In the case of apes, newborn chimpanzeeshave been shown to have a period of innatemirroring responsiveness to human facial stimuli(mouth opening and tongue protrusion) that ispresent to 11 (Myowa-Yamakoshi 2001) or 9(Myowa-Yamakoshi et al. 2004) weeks of age.This shows that neonate chimpanzees have (atleast) the capacity for proto-mimesis with respectto faces. It is to our knowledge not yettested whether this capacity also involves differentiationor volition. But different sources forhighly similar responses in human and ape newbornsseem unlikely.4.2 Dyadic-mimetic imitationIn what we call dyadic-mimetic imitation, thesubject differentiates between self and modeland understands that a bodily posture or motioncan correspond to something else, like an objector action. However, there is no attempt tocommunicate this “something else” to anotherindividual by means of the bodily motion.In a non-communicative setting, like in theexperiments on social learning studied with nonhumanprimates the “something else” mentionedabove would be to understand the meaningof a model’s actions: what the model is tryingto achieve. As mentioned above, imitating thegoal as well as the means to bring about that goalis the hallmark of (true) imitation according toTomasello, Kruger and Ratner (1993), and involvescertain levels of intersubjectivity involvingunderstanding of intentions.When trying to implement the imitation ofgoals in an experimental setting the goal is oftento get hold of a food item, and these experimentshave tended to either foster innovation oruse of old reliable methods. This is not surprising.If you truly want to get hold of a food item,and apes clearly do so, the most effective means

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 9one can come up with is used, and that is notnecessarily the one witnessed.The well-known rake experiment of Nagell etal. (1993) is an example of imitation studies usingan instrumental tool. If there is any learninginvolved in the case of apes it is generally consideredto be in the form of emulation: the subjectslearn something new about tools, theirmovements or the environment, but not aboutthe bodily actions the model is using to bringabout the intended goal.Better imitation results have been obtained inexperiments with so called “artificial fruits”, i.e.boxes with obstacles that the subject has to workaround in order to get to the food content.These experiments address the question: Whatdo apes spontaneously copy when they get anopportunity to observe a model? Which strategiesand mechanisms of social learning are involved?The artificial fruit experimental paradigmhas yielded mixed results, as strategies ofresult emulation, object movement reenactment,action-level imitation and sequenceimitation have been recorded (Whiten et al.2004). The results seem to vary over fruit usedand species tested and/or age of subjects (Stoinskiand Whiten 2003; Custance et al. 2001; Stoinskiet al. 2001; Whiten et al. 1996). Action-levelimitation implies that the subject can represent amodel’s manual movements and re-enact thesefrom memory. Performing sequence imitationmeans that the subjects are also able to parse astring of component actions and later reassemblethese in the same fashion from memory,which again show their ability to re-enact detachedrepresentations voluntarily.Apes have demonstrated some skills of action-leveland sequence imitation, but at thesame time they experience some difficulties withthem as evident by their lack of full copying.Imitating the use of familiar motor actions innovel situations seem to come about more easilyin apes than copying new motor actions altogether(Myowa-Yamakoshi and Matsuzawa1999). Especially for young individuals actionsare more likely to be copied if they are close tothe ones already in the subjects’ repertoire(Bjorklund et al. 2002). It is not yet settledwhether apes’ relatively poor performance ondetailed imitation of bodily actions is due tomotor-perceptual aspects or skills in intersubjectivity.However, since recent studies have shownthat at least chimpanzees have a degree of understandingof the mentality of others (see belowand Section 5.2), the motor-perceptual explanationshould not be underestimated.Goal directedness is lacking in the most bodilymimetic paradigm, the “do-as-I-do” experiments,where part of making sure that the copyis acquired through observation is to performnever-before seen, or at least unusual, movementswithout any practical function. There isno food item to get to at the end of the actionsequence. The subject is verbally asked to do thesame as the model on cue. The do-as-I-do paradigmis an example of studies that ask the question:Can apes copy behavior? The answer ispositive. Both cross-fostered (or “enculturated”)apes, e.g. the orangutan Chantek (Miles et al.1996) and the chimpanzee Viki (Custance et al.1995), and chimpanzees with more moderatehuman upbringing (Custance et al. 1995) canpass these tests. Once they understand themeaning of the “do the same” command, anyfaults in the copy is bound to be motorperceptualand not due to intersubjectivity issues.The do-as-I-do experiments prove thatapes are capable of both volition and representation(they represent the model’s body), which qualifythem as dyadic-mimetic imitators. However,learning the concept of SAMENESS may influencethe performance of apes profoundly. Thompson,Boysen and Oden (1997) found that language-naïvechimpanzees learned to judge relationsbetween relations (A is to A as B is to B,and not as C is to D) in a matching-to-sampletask only if the subjects had previously learned atoken for the concept SAME. Ordinary matchingto-sampletraining could not yield the same result.Experiments that hinge on the subject “gettingthe point of the game”, so to speak, haveimplications for ecological validity since theexperimenter has to plant certain knowledge inthe subject that it might not have stumbled uponnaturally. This is especially problematic for evolutionaryreasoning around comparative experimentalpsychology, since it is not clear whatcircumstances would have “planted” suchknowledge in the hominid line.Deferred imitation on objects, involving doas-I-doimitation, is perhaps the most promisingevidence for dyadic mimetic capacities in apes.In these experiments the subject is shown actionson objects and is then either asked to “dothe same thing” (if the verbal command islearned) or (if not) the spontaneous handling of

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 11taking have shown that chimpanzees know theimportance of others’ visual attention, both incommunicative contexts (Tomasello et al. 1994;Krause and Fouts 1997; Pika, Liebal andTomasello 2003, 2005) e.g. use of attentiongettinggestures and competitive contexts (Hareet al. 2000; Hare, Call and Tomasello 2001).Apes also seem to be able to distinguish intentionalfrom accidental acts (Call and Tomasello1998) and to judge an experimenter as eitherunwilling or unable to give them a treat (Call etal. 2004). The conclusion is therefore that apesdo understand second-order mentality, at least inthe form of second-order attention and (possibly)second-order intentions. However, understandingthe communicative sign function, andthus triadic mimesis, requires third-order mentality– as we will argue in Section 5 – and there isyet no clear evidence that (non-enculturated)apes are capable of this.Possible instances of triadic mimetic imitationin non-human primates would be copyingof referential signs in language-taught apes.However, most of the learning of signs by apesthat have been taught a signed language, such asthe gorillas Koko and Michael (Bonvillan andPatterson 1993, 1999), the chimpanzee Washoe(Fouts 1972) and the orangutan Chantek (Miles1990) takes place by getting the hands moldedinto the signs and therefore does not qualify aseither imitation or bodily mimesis. Still, Gardnerand Gardner (1969) also used the “do-as-I-do”paradigm to teach new signs to Washoe withsome success. A more general problem, however,is that it is difficult to distinguish the imitativelearning of such signs from post-mimesissince these apes were taught a simplified form ofAmerican Sign Language (ASL), and that impliesthat they were exposed to, and possibly evenacquired at least a degree of symbolicity (involvingconventionality and systematicity). In a studyof the functions of the repetition of lexigrams bylanguage-trained chimpanzees and bonbobosGreenfield and Savage-Rumbaugh (1993) reportat least some clear cases of the acquisition oflexigrams through imitation. At the same timeSavage-Rumbaugh and Lewin (1994) emphasizethat only when the chimpanzees Sherman andAustin acquired the referential meaning of thelexigrams, after a prolonged and laborious processof training, could they match lexigrams tosamples.Perhaps the most convincing evidence fortriadic-mimetic imitation in apes involves theinvention of iconic signs by Koko (Patterson1980) and Chantek (Miles 1990). Such sign areexamples of representation (the ape representssomething in the environment) and communicativesign function (the ape addresses this to his or hercaregivers). Since the imitation is creative, involvingthe resemblance (iconicity) between thesign and the way the referent is typically used, itcannot be solely a product of being taught asymbolic code. The fact that Koko’s and Chantek’ssign inventions build on iconicity suggestthat they are cases of triadic-mimetic rather thanpost-mimetic imitation. We return to these casesin Section 6.4.4 Post-mimetic imitationWhile speech may be regarded as in part postmimetic(see Section 3), the paradigm exampleof post-mimesis is signed language: it is based oncross-modal mapping between proprioceptionand vision, and possesses the other features ofmimesis: volition, representation and communicativefunction. Furthermore, it is symbolic, not inthe Peircian sense of lacking any motivationalrelationship between the expression and contentpoles of its signs – over 50% of the signs of ASLare judged to be iconic (Woll and Kyle 2004) –but in the sense of consisting of conventional signs(presupposing third-order mentality, see Section5) and systematicity (i.e. internal relations betweenthe signs). Other instances of postmimesiswould be emblems like the thumbs-up“OK” sign, or nodding instead of saying “yes”,which fulfill the property of conventionality, andat least a limited form of systematicity.Accordingly, post-mimetic imitation wouldconsist of skills involved in the learning of suchsigns. As pointed out above, some apes thathave been taught a signed language did acquiresome of their signs in this manner, though themajority of signs were acquired through moldingof the hands. Even for those that were acquiredby imitation, there may be doubt concerning thedegree of their conventionality and systematicity,e.g. deaf native signers have difficulties interpretingapes’ signing (Pinker 1994).One form of imitation that is clearly symbolic(and hence post-mimetic) is what Tomasello(1999) calls role-reversal imitation: imitation of sign

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 13of language, it is not so much the syntax of imitationthat is difficlt for apes, but its semantics. It isconceivable how ape copying ability, coupledwith a motivation to share one’s mental states(which can, of course, also mean attempting toimpose one’s view on others), can give rise tomimetic-communicatory interactions, in whichlevels of mimesis and intersubjectivity are closelyconnected, as we will argue in the followingsection.5. Mimesis and intersubjectivityWe understand the notion of intersubjectivitybroadly (and rather literally) as the sharing andunderstanding of others’ mentality. The term “mentality”is taken here to involve not only beliefs andother forms of conceptual knowledge, but allforms of consciousness, including emotions,desires, attentional foci and intentions (seeGärdenfors 2003). Not all of these need to beunderstood conceptually, but may involve preconceptual,non-reflective understanding.While the close connection between bodilymimesis and imitation discussed above, andgesture (to be analyzed in Section 6) is fairlystraightforward, this may not be the case withrespect to intersubjectivity. The reason for thislies, we believe, in the dominant cognitivist viewof mentality as consisting entirely (or at leastpredominantly) of propositional representations.In this view, to understand another’s mindwould imply having beliefs concerning theirbeliefs. This conception can easily be framed ashaving some sort of “theory of mind” and, indeed,this has been the dominant conception inthe field (e.g. Baron-Cohen 1995).A great deal of the debate about the cognitiveabilities of apes has concerned whether or not theyhave a “theory of mind” (Woodruff andPremack 1979; Byrne 1995; Gómez 1994; Heyes1998; Tomasello 1999). Also in the discussion ofhuman cognitive development, the question whenand how children develop one has been a centralfocus of content (Perner et al. 1987; Gopnik andAstington 1988; Gopnik and Meltzoff 1997;Mitchell 1997).But regarding intersubjectivity as a “theory”involving (primarily) beliefs has had some negativeimplications for understanding this fundamentalsuite of socio-cognitive capacities: (a) thequestion is typically posed as involving a discrete,all-or-nothing property: either the child oranimal “has a theory of mind” or doesn’t; (b) therole of one’s own body and actions – and those ofthe other – is minimized, and (c) “theorizing”about the minds of others (or even oneself) isconsidered prior to acting in consort with them,giving intersubjectivity a rather static, detachedquality.Recently, however, these implications havebeen seriously questioned on the basis of bothnew empirical evidence from child developmentand primatology and older philosophical, aboveall phenomenological, insights. With respect to(a) it has become clear that “the generic term‘theory of mind’ actually covers a wide range ofprocesses of social cognition” (Tomasello, Calland Hare 2003: 239). Consequently one mustdistinguish between these different processes inorder to understand the cognitive capabilities ofanimals, as well as children of different ages. Forexample, Gärdenfors (2003) splits the so-called“theory of mind” into six levels, where the middlefour clearly correspond to different sociocognitivecapabilities: understanding others’emotions, attention, intentions and beliefs, respectively.Such a division can be complementedwith the hypothesis that the different capabilitiescorrespond to different phylogenetic and ontogeneticstages in evolution and development, andindeed Gärdenfors (2003) makes this proposal.The criticism concerning (b) and (c) was developedfirst within the phenomenology ofHusserl some one hundred years ago and targetedthe intellectualist perspective on the “understandingof other minds” (the usual term forintersubjectivity in philosophy) as a matter ofinference or analogy from the knowledge ofone’s own mind. As summarized by a moderninterpreter: “For Husserl, understanding anotherperson is not a matter of intellectual inference,but a matter of sensory activations that are unifiedin or by the animate organism or lived bodythat is perceiving another animate organism.”(Gallagher in press).This perspective on intersubjectivity was furtherdeveloped by Merleau-Ponty (1962), inparticular through the notion of the “corporealschema” which serves as “a normal means ofknowing other bodies” (Merleau-Ponty 2003:218). More recently, Gallagher (1995, 2005, inpress) has elaborated the distinction betweenbody schema and body image where the first is preconsciousand serves as a precondition and

14Bodily mimesis as “the missing link” in human cognitive evolutionbackdrop for intentionality (cf. Searle’s (1992)notion of the Background), while the latter is “a(sometimes conscious) system of perceptions,attitudes, beliefs and dispositions pertaining toone’s own body” (Gallagher 2005: 37). Empiricalneuroscience has recently given support for therole of the body in understanding others throughmechanisms involving “mirror neurons” (Rizzolattiet al. 1996; Gallese, Keysers and Rizzolati2004; Arbib in press), that mediate between theperception of another and the subject’s ownproprioception and action.Within this “embodied” and action-orientedperspective on the understanding of otherminds, unlike in the intellectualist cognitivistone, bodily mimesis becomes clearly relevant.With respect to distinguishing different levels ofintersubjectivity, we can apply the mimesis hierarchydefined in Section 3 to yield a possibleevolutionary and developmental hierarchy ofintersubjectivitity.It should be noted that this way of “splittingthe theory of mind” is different from that proposedby Gärdenfors (2003) since the differentcapabilities distinguished by the Gärdenfors(understanding emotion, attention, intention andbelief) are in a sense orthogonal to the mimesishierarchy. For example, the understanding ofothers’ emotions occurs on each one of the levelsof the mimesis hierarchy, with increasingcognitive complexity. However, if we considerwhat may be regarded as a typical manifestation ofthe respective capability (emotion, attention andintention), there turns out to be considerableoverlap between the levels of intersubjectivitydefined in this article and those of Gärdenfors(2003): Emotion can be shared on a protomimeticlevel, sharing attention involves (atleast) dyadic mimesis, understanding communicativeintentions is centrally implicated in triadicmimesis (but “simple” intentions are understooddyadically), and understanding beliefs appears torequire explicit symbolic skills, which can beargued to be post-mimetic since they depend onlanguage (see Section 5.4).A central question in relating the proposedmimesis hierarchy and intersubjectivity iswhether the respective mimesis level serves as aprecondition and a causal factor for the developmentof corresponding skills of intersubjectivity. Or isit rather that independently reached insights intothe mind of others makes increasingly complexforms of mimesis possible? Our general Vygotskyanapproach suggests the first scenario:bodily mimesis is a fundamentally interpersonalactivity that exercises (in ontogeny) and providesselection pressures (in phylogeny) for developingmore refined skills in mind reading. Thus, ourpredominant take is that mimesis drives intersubjectivityrather than the other way around.At the same time we acknowledge that the “driving”metaphor is not completely adequate sincethe causality runs in both ways and is possiblybest described in terms of co-evolution. We willreturn to this issue in the summary at the end ofthis section.5.1 Proto-mimetic intersubjectivityGiven our definition of bodily mimesis we canregard some of the most basic forms of intersubjectivityas “proto-mimetic” to the extentthat they (a) consist of forms of interpersonalinteraction that involve cross-modal mappingbetween proprioception and the (visual) perceptionof others, (b) do not fulfill either one of thecharacteristics volition and representation. This canbe made more precise using the distinction madeby Gallagher (1995, 2005) between body schemaand body image. The first is characterized as “asystem of sensory-motor processes that constantlyregulate posture and movement – processesthat function without reflective awarenessor the necessity of perceptual monitoring” (Gallagher2005: 37-38). The second is, as pointedout, “a (sometimes conscious) system of perceptions,attitudes, beliefs and dispositions pertainingto one’s own body” (Gallagher 2005: 37).Given these definitions, we can state that acts ofproto-mimesis involve (above all) the body schema,which is largely innate (in the sense of beingpresent at birth) and pre-conscious, rather thanthe body image, which is gradually constructedwith experience and is accessible to consciousness.Furthermore, the forms of intersubjectivitydescribed here do not require a conceptual differentiationbetween self and other, which isnecessary for establishing a correspondencerelation between them, i.e. what we refer in ourdefinition as representation. This is not to say thatthe young infant lives in a completely undifferentiatedworld in which there is no awareness ofself whatsoever, as in classical accounts of infantcognition (e.g. James 1890). Nevertheless, even a

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 15modern developmental psychologist who emphasizesthe role of awareness of others’ attentionand the presence of affective selfconsciousnessin the first months of life pointsout that “older infants reveal a greater focus onthe self and the younger ones reveal a more immersed,less detached focus on the other” (Reddy2003: 401). This “more immersed, less detached”quality of the earliest forms of intersubjectivitysupports our classification of them asproto-mimetic.Can this analysis be extended to the (early)interpersonal relations among apes? As pointedout in Section 4.3 “neonatal mirroring” has alsobeen observed in apes (though this has so faronly be attested with human, rather than apefaces as stimuli). Since this is typically attributedto a form of identification with the person imitated,serving as a basis for intersubjectivitywhen children are concerned (Meltzoff andGopnik 1993; Gallagher 2005), it can be viewedas evidence that at least apes, and possibly othermammals too, have such basic proto-mimeticintersubjectivity. The function of such “mirroring”can be related to what is possibly the mostbasic form of intersubjectivity, both ontogeneticallyand phylogenetically: the ability to shareemotions, or empathy (Einfühlung). As a protomimetic,non-representational capacity, this istestified in early infancy and sometimes referredto as interaffectivity (Stern 2001). The well-knownexperiments described by Trevarthen (1992)show that parent-infant interactions in the firstfew months take the form of a reciprocal rhythmic“dance”, and that frustration follows if this“attunement” is disrupted (notice the musicalmetaphors). The suggestion is that emotionssuch as joy and suffering are perceived directly,possibly involving mirror-neuron structuressimilar to those involved in action recognitionand imitation, rather than involving inferences tounderlying states (Gallese, Keysers and Rizzolati2004).Preston and de Waal (2002) have argued persuasivelythat as a basic mechanism involving thelinkage of perception and action, a basic form ofempathy is available to most if not all mammalspecies. Defining empathy as “any processwhere the attended perception of the object’sstate generates a state in the subject that is moreapplicable to the object’s state or situation thanto the subject’s own prior state or situation”(ibid: 4), they see a clear evolutionary motivationfor its emergence in the ability to recognize andunderstand the behavior of con-specifics. It is anopen question how much of such matching betweenthe visually perceived body of the otherand the proprioceptively perceived body of oneselfis domain-general – and thus can be expectedto be general across species – and howmuch is specialized in the form of speciesspecificcommunicative signals such as facialexpressions. It is characteristic that such signals,such as the famous “play-face” expression ofgreat apes (an evolutionary precursor to thehuman smile) typically carry emotional ratherthan referential meaning.A second socio-cognitive skill that relates tointersubjectivity and appears to have a protomimeticorigin, at least in human children, isattention. Reddy (2001, 2003, 2005) has arguedthat prior to awareness of the other’s attentionto an external object and much prior to jointattention appearing around 12 months (see below),children “show an awareness of others asattending beings, as well as an awareness of selfas an object of others’ attention” (Reddy 2003:357), displayed in phenomena such as eyecontact,intense smiling, coyness, ‘calling’ vocalizations,showing-off etc. Since awareness of self(at this stage) is largely proprioceptive, whileawareness of the attention of others (in seeingchildren) is mostly based on vision, this satisfiesour first criterion for bodily mimesis. Reddy’sclaim is that such dyadic (though not dyadicmimetic)interactions underpin later developmentsin intersubjectivity. Evidence for this is the observationthat autistic children show difficultieseven with such simple interpersonal engagements,“suggesting that whatever is going on indyadic attentional engagements may indeed becritical, not just as a source of information andexperience about attentional behavior, or as ascaffold for the subsequent development ofawareness of attention, but also as evidence ofawareness of attention” (Reddy 2005: 95).Until recently it has not been clear whethersuch awareness of another’s attention exists inthe interaction between infant apes and theirmothers, but in a recent study, Bard et al. (inpress) report that the rates of mutual gaze betweeninfants and their mothers are virtually thesame in 3-month old human children and 3-month old chimpanzees; 18-20 and 17 times perhour, respectively (though humans tend to engagein longer bouts of mutual gaze). Further-

16Bodily mimesis as “the missing link” in human cognitive evolutionmore, the authors noticed a “cultural” differencebetween the apes at Primate Research Institute,Japan and those at Yerkes National PrimateResearch Center, USA, with the ones in Japanengaging in mutual gaze at much higher rates (22vs. 12 times per hour), while the ape mothers inthe USA cradled their infants more often (71% vs.40% of the total time). Intriguingly, a similarinverse correlation between visual and tactilecontact has been observed in human societies,with traditional cultures favoring touch andWestern ones gaze: “With reduced physical contactfound in Western societies, mutual engagementshifts to the visual system, arguably a moreevolutionarily derived pattern” (Bard et al. inpress). Since apes do not seem to differ fromhumans in the capacity to perform this shift (andpossibly even transmit it culturally to their descendents),this confirms the conclusions fromneonatal imitation that there is no major differencebetween the species on the proto-mimeticlevel.5.2 Dyadic mimetic intersubjectivityIn our previous discussion, we proposed thatproto-mimetic intersubjectivity, without a cleardifferentiation between self and other is basedon the (mostly) unattended body schema andsimilarly unattended mechanisms for “bodycopying”. On the other hand, what we here referto as dyadic mimetic intersubjectivity is based on theconscious control of the movements of one’sbody and attention to their correspondence tothe body of another, whereby one can imaginewhat the other experiences on the basis of one’sown experiences in similar circumstances. In theterminology of Gallagher (1995, 2005), we proposethat the role that bodily mimesis properplays for the development of intersubjectivityimplies not the body schema but the body image.While the body schema and the body imagenormally interact, Gallagher (2005) shows howin certain pathologies they can be disassociated.The distinction between the dyadic form ofmimetic intersubjectivity, described here, and thetriadic one described in the following sub-sectionis that while the dyadic form involves the propertiesof volition and representation (see Section3), e.g. in the case of recognizing oneself in amirror or in shared attention, there is no understandingof communicative sign function. Thus thereis no basis for intentional communication on thebasis of shared representations. As we shall seehere and in the discussion of ape gestures inSection 6, it is the latter that appears to be mostdifficult for non-human primates. Below, webriefly describe how understanding others’ emotions,attention and intentions can be seen asintimately related to dyadic mimesis.Whereas (simple) empathy is proto-mimetic,what Preston and de Waal (2002) call cognitiveempathy requires a differentiation between subjectand object where “the subject is thought to useperspective-taking processes to imagine or projectinto the place of the object” (ibid: 18). Evidencethat this is not an isolated phenomenon,but shows a more advanced level of intentionalityis the fact that cognitive empathy “appears toemerge developmentally and phylogeneticallywith other ‘markers of mind’ … including perspectivetaking …, mirror self-recognition …,deception, and tool-use.” (ibid: 18). Researchconcerning cognitive empathy in apes has focusedon their consolation behavior, which iswell attested in at least chimpanzees, but has notbeen found in monkey species (de Waal andAureli 1996) or any other mammalian species.Consolation is cognitively more complex thansimple empathy since the consoling individualnot only feels that somebody else experiences anegative emotion, but also intends to help relievethis, implying an ability to imagine the morepositive emotional state. 3This supports the interpretation that cognitiveempathy involves a more sophisticated representationalcapacity than what is necessary forsimple empathy. Since dyadic mimesis involvesboth the ability to identify with the other, and atthe same time to differentiate between self andother, our (Vygotskyan) hypothesis is that it isdyadic mimesis, implicated in e.g. imitation (Section4) that scaffolds the development of suchrepresentational capacity.Since dyadic mimesis allows to “place oneselfin the shoes of others”, it also gives the oppor-3 An interesting question is what goes on in themind of an ape that is being consoled: If one couldidentify cognitive processes of the form “I can feelthat you feel that I feel bad”, this would be equivalentto “third order emotion”. However, given thelack of methodology for testing the existence ofsuch a process, we feel compelled to apply Occam’srazor and ascribe only second order emotionsof the form “I feel that you feel”.

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 17tunity to understand what someone else is attendingto. Such “second-order attention” is welltestified among great apes (Hare et al. 2000).When two individuals become aware that theother is attending to the same object, what resultsis shared attention. This comes a good dealtowards the construction of a “consensual reality”that can be communicated about, but doesnot quite reach it (cf. Brinck 2004). To make agiven object X fully intersubjective between youand me, I would need not only to “see that yousee X”, (second-order attention, see Figure 1a),but also “to see that you see that I see X” (thirdorderattention, see Figure 1b) and vice versa –which is our interpretation of what it means toengage in joint attention.Figure 1a. Shared attention: Second-order attention: “Isee that you see X” (and vice versa)there is a close connection between communicativeintention and the emergence of joint attention.Concerning the understanding of another’sintentions, it has been for sometime a predominantopinion that apes cannot do this, and indeedthis was one of the central claims ofTomasello (1999). Recently, however, there hasbeen mounting evidence that (at least)“[c]himpanzees understand psychological states– the question is which ones and to what extent”,which is the title of Tomasello, Call andHare (2003). In the experiments behind thisclaim a subordinate and a dominant chimpanzeecompeted for food placed on the subordinate’sside of two barriers, so that in some cases onlythe subordinate, but not the dominant could seethe food. The subordinates could also monitorthe visual access of their dominant competitor.The results showed that the subordinates preferentiallyretrieved the food that dominants couldnot see, suggesting that chimpanzees understandwhat con-specifics have and have not seen andeven have some awareness of the dominant’s goal(i.e. “a mental representation of a desired state”,Tomasello et al. in press) of obtaining the food.This does not have to be done on the basis ofexplicit reasoning or inference, but possiblythrough the “projection” of one’s own motivationalstate in a similar situation onto the other(“I would get the food if I were in his place!”), ina dyadic mimetic fashion. Apparently, even nonenculturatedand language-naïve apes are capableof this. 4However, it has not been shown that (nonenculturated)apes are capable of understandinganother’s mental states about their own mentalstates, which would involve, again, third-ordermentality. 5 As suggested earlier, joint attentionFigure 1b. Joint attention: Third-order attention: “I seethat you see that I see X” (and vice versa)Full joint attention thus involves third-ordermentality (cf. Brinck 2003, 2004), and (possiblybecause of this, see below) appears to be beyondthe cognitive capacities of apes (Tomasello1999). It also goes beyond dyadic mimesis, so wewill leave this capacity for the time being, butreturn to it in the following subsection, since4 While it still not conclusively shown that apes arecapable of such mental “projection”, and it is conceivablehow the evidence can be explained in amore behavioristic manner involving learning generalizationsover other individuals’ behavior inrelation to food, we contend that the mental explanationis (a) ultimately more parsimonious and (b)consistent with other mental skills in apes, as wellas in human children in roughly comparable stagesof development.5 Note that when we in this section link triadic mimesisand third-order mentality, we do not implythat an agent capable of triadic mimesis can handleall forms of third-order mentality, in particularthose which involve beliefs.

18Bodily mimesis as “the missing link” in human cognitive evolutionalso requires third-order mentality, and is similarlysomething that apes seem to find difficult.In natural settings, some cases of deceptionmay be interpreted in a way to involve thirdordermentality, but do not require this. Forexample de Waal (1998 [1982]) describes thechimpanzee Yeroen who has had a fight withNikkie, and continues to fake a limp only whenin the presence of Nikkie, apparently in an act ofwishing to provoke Nikkie’s empathy: “I wish tomake you see that I hurt”. The more parsimoniousexplanation, however, is that Yeroen haslearned from previous experience that he is notbothered by Nikkie when he is hurt: “He mayhave learned from incidents in the past in whichhe is seriously wounded that his rival was lesshard on him during periods when he was (ofnecessity) limping” (ibid: 35-36), so he mimes theappropriate behavior. Here we have a clear correspondencebetween dyadic mimesis and second-orderintersubjectivity. Notice that Yeroen’slimping was not a form of intentional communication– he did not wish Nikkie to understandthat he was faking a limp – if he did, that wouldbe a clear case of triadic mimesis.Another possible instance of third-ordermentality is that reported by Woodruff andPremack (1979) in which chimpanzees learned tosuppress glances and body orientation towardthe place where food was hidden, when a trainerwho would not share the food was present.However, since it took a long number of trialsfor the apes to learn this behavior, a simplerexplanation involving conditioning is possible.The conclusion that we can draw from thesevarious examples is that wild apes as well asthose who are exposed to different degrees ofhuman contact (captive, nursery-raised and laboratory-trainedapes), but are not raised in a“something like a human cultural environment”and thus enculturated (Call and Tomasello 1996)can indeed understand second-order mentality.However, such apes do not seem able to understandthird-order mentality – neither in the domainsof emotion, attention nor intention – inwhich their own mental state needs to be eitherintentionally communicated – in collaboration,or hidden – in competition. This correspondswell with the capacity of apes for dyadic mimesis,but their relative difficulty with triadic mimesis,which we discuss further below.5.3 Triadic mimetic intersubjectivityIn the case of triadic mimesis, there is by definitionnot just an understanding of representationitself, but an understanding that representationscan be used communicatively, or what we referto as communicative sign function. This implies anunderstanding that the sign has the same meaningfor the addressee as for the sender. This involves atleast second-order mentality, which was shownabove to correlate with dyadic mimesis. But“having the same meaning” is a reflexive notionand this implies at least some degree of thirdordermentality. Consider the simple example ofwhat knowing the meaning of the word cat implies:a) I know that cat means “a small furry animalthat meows”.b) I assume you know that cat means “asmall furry animal that meows”.c) I assume that you know that I know thatcat means “a small furry animal that meows”.While it is possible for intentional communicationto begin without a full realization of (c), itis practically inevitable that discursive experience(including failures in communication) will promotethe development of third-order mentality.Thus, it is possible that it was sign use itself thatwas the major driving force behind the developmentof intersubjectivity in hominid evolution.Unlike competing hypotheses related to“Machiavellian intelligence” (Byrne and Whiten1988), this puts the focus on cooperation ratherthan competition (see also Brinck and Gärdenfors2003; Tomasello et al. in press). A predictionfrom this hypothesis is that enculturatedapes – and these have all been taught at leastsome communication through signs – will develophigher-level skills of intersubjectivity.There is some support for this prediction. Insummarizing some 200 studies of the role ofhuman influence, Call and Tomasello (1996)conclude that “[t]he sociocognitive domains inwhich humans seem to have the highest effecton apes are intentional communication and sociallearning” (ibid: 391).In Section 6, we will focus more specificallyon the understanding and use of communicativegestures by apes. Here and in the following subsection,we will mostly review the evidence thatsign use in general, and the acquisition of (aspectsof) language in particular is the major fac-

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 19tor behind the development of higher forms ofintersubjectivity that may deserve the label “theoryof mind”.As pointed out earlier, wild apes do not seemto be capable of engaging in full, third orderjoint attention. Furthermore, as Tomasello(1999: 21) points out, wild apes (1) do not pointto objects; (2) do not hold up objects to showthem to others; (3) do not take someone along to aplace to show them something; (4) do not activelyoffer something to someone; and (5) do notintentionally teach other individuals new behaviors.Tomasello’s original account of these absenceswas based on the claim that apes are unableto understand other’s intentions. Given themore recent evidence, this explanation is nolonger tenable, and indeed Tomasello et al. (inpress), suggest instead that the crucial differencebetween apes and humans involves the motivationto participate in joint collaborative engagements,and the lack of this motivation preventsapes from constructing “dialogical cognitiverepresentations”.Our explanation is similar but more specific:we believe that non-enculturated apes fail todevelop the communicative sign function, andrelated to that, the ability to engage in third-ordermentality. The motivational difference betweenapes and humans appealed to by Tomasello et al.(in press) cannot be the full explanation sinceenculturated apes such as Koko, Kanzi andChantek manage at least (1) and apparentlycommunicative skills (2-5) above as well (Miles1999), even if in restricted forms. This seems toimply that the human cultural environments 6 ofthe enculturated apes have taught them the basicsof intentional, sign-mediated communication,and thereby (the roots of) third-order mentality.How this could occur can be seen again withrespect to joint attention, which can be seen asemerging from second-order attention combinedwith the recognition of another’s intention concerningmy attention: “I see that you see X”(second-order attention, Figure 1a), and furthermore“I realize that you want me to look atX”. In other words, joint attention can bebrought forth by understanding a simple form ofcommunicative intention, combined with already6 Consisting of what Wittgenstein (1953) called the“forms of life” that provide the necessary contextfor the emergence and functioning of intentionalcommunication and language.existing second-order attention (Brinck 2003,2004). Thus, communicating the intention tojointly attend may be said to involve the simplestkind of triadic mimesis: whatever kind of behaviorthat is used to convey that intention – someform of index (see the example in the next paragraph)– can be said to stand for that intention forboth sender and interpreter.Without enculturation, experiments indicatethat apes do not understand communicativeintentions. A rather typical example is an experimentby Tomasello, Call and Gluckman(1997), where the authors in different ways indicatedfor both chimpanzees and two- to threeyear-oldchildren which out of three containerscontained a reward: by pointing to the correctcontainer; by placing a marker on top of thecorrect container; and holding up a replica of thecorrect container. Tomasello (1999: 102) summarizesthe results of the experiment as follows:Children already knew about pointing, butthey did not know about using markers andreplicas as communicative signs. They neverthelessused these novel signs very effectivelyto find the reward. In contrast, no ape wasable to do this for any of the communicative signsthat they did not know before the experiment.One explanation of these results is that theapes were not able to understand that the human beingshad intentions toward their own attentional states.The apes thus treated the communicative attemptsof the human as discriminative cues onpar with all other types of discriminative cuesthat have to be laboriously learned over repeatedexperiences. The children, in contrast,treated each communicative attempt as an expressionof the adult’s intention to direct theirattention in ways relevant to the current situation.[our emphasis]In other words, while the children clearly understoodthe communicative intentions of the experimenter,the apes did not. This interpretationis supported by a similar experiment designed totest “false beliefs” (Call and Tomasello 1999), inwhich the enculturated and language-taughtorangutan Chantek clearly performed differentlyfrom all the other apes in understanding a humancommunicator’s signals. Even though thiswas not the goal of the experiment, and Chantekdid not score better than the other apes in thefalse beliefs task, his much better performancecould be explained by considering that he understoodthe signals as communicative signs (in this

20Bodily mimesis as “the missing link” in human cognitive evolutioncase indexes), rather than as “discriminativecues”.Finally, we should mention the case of captiveapes living in a zoo and thus involved withat least some degree of interaction with humanculture. In their study of spontaneous gesturalcommunication in a group of gorillas in the SanFrancisco zoo, Tanner and Byrne (1996, 1999),found a wealth of gestures used by severalmembers of the group, in particular by the adultmale Kubie. We return to this is Section 6, but itsuffices here to note that the gestures seemed tobe used in a communicative way so that:[w]hether the receiving partner was a humanor another ape, the signaling ape made surethat visual contact was established (except fortactile close gestures), and seemed to understandboth the other’s potential actions and what thepartner might, in turn, understand from his (the signaler’s)performance of gestures. (Tanner and Byrne1999: 231, [our emphasis])We can conclude that a triadic from of mimeticintersubjectivity, involving understanding notonly con-specifics’ intentions, but their communicativeintentions, and consequently a degree ofthird-order mentality, appears to be not completelybeyond the cognitive potential of apes. To realize thispotential, they need an environment that is richin opportunities for developing the communicativesign function, i.e. a particular form of enculturation.Thus triadic mimesis may be said to bewithin apes’ “Zone of Proximal Development”(ZPD) – the notion introduced by Vygotsky(1978) to refer to skills that children could acquirewith the help of adults, but not alone.If “enculturation” provides the ZPD for present-dayapes, it is reasonable to suppose that itdid the same for some particularly social groupof hominids through a form of “selfdomestication”giving rise to a bootstrappingspiral of sign use and intersubjectivity. In theterms of Donald (2001), triadic mimesis musthave been within the common ancestor’s “zoneof proximal evolution”.5.4 Post-mimetic intersubjectivityWhat differentiates post-mimetic, or symbolic,cognition from mimesis is the use of fully conventionalsigns, interrelated within a system (Deacon1997; Zlatev 2003). The most obvious exampleof post-mimesis, involving all the previous featuresbut also symbolicity is a conventional, institutionalizedsigned language such as ASL (Stokoe1960) or Swedish Sign Language (Ahlgren2003). What is the relation between acquiringsuch a system and intersubjectivity?A convention (Lewis 1969; Clark 1996) or anorm (Itkonen 1978) exists as a form of mutual(or common) knowledge among the members of thegroup that share the convention. A commonexplication of mutual knowledge is that it consistsof third-order knowledge: “I know that youknow that I know X”. Translated to the capacityto understand the minds of others, commonknowledge amounts not only to my understandingthe mentality of others, but also to my understandingthat others (can) understand mymentality. This has been expressed by Itkonen(1978: 96) as follows:[P]erson A cannot know that he is doing orthinking X, and thus cannot do or think X,unless he is able to know what it is for someother person B to do and think X. The sameis true, in turn of B’s knowledge of his ownactions with respect to A’s (possible) actionsand thoughts. Hence it can be shown onpurely conceptual grounds that A must beable, in principle, to identify B’s various mentalstates or processes, and vice versa. Thismeans that mental states and processes existonly at the level of common knowledge i.e. of acommon ability to identify them, whereverthey occur.Mutual knowledge is often stated as a form of“third-order belief”, but this is potentially misleadingsince it is not necessary that the understandingon all three orders of mentality is explicitenough to be a matter of “belief”, i.e. apropositional representation that is actively heldto be true. Consider again the three orders ofknowing the conventional meaning of cat mentionedearlier: I assume that you know that Iknow that cat means “a small furry animal thatmeows”. The highest order, my assumption thatyou know that I know, is not properly speaking abelief for the 4 year old child, since it is taken forgranted, without pondering on whether it is trueor not. At the same time, children at this agebecome (a) capable of understanding that otherslack knowledge or have “false beliefs” (e.g.,Perner 1991; Wimmer et al. 1988; Gopnik andGraf 1988; Mitchell 1997), implying metarepresentationalcapacity, and (b) at least somewhat fluentin their first language. It appears that these twodevelopments are closely connected, and that

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 21acquiring a language, spoken or signed, is a majorcausal factor for developing a fully-fledged“theory of mind”. Despite successes in understandingother’s attention and even intentions,apes, including enculturated ones, have so farconsistently failed so-called “false belief tests”(Premack 1988; Call and Tomasello 1996, 1999;Povinelli 2000). They have also failed in acquiringa full human language – despite successes inacquiring certain aspects of it. Three differentsides to language (use) combine to promotemetarepresentational capacity.First, as mentioned above, language is a conventionalsymbolic system, and as such its masteryimplies third-order knowledge, which wouldcarry with it training in the understanding ofothers’ beliefs. Second, two specific (universal)features of human languages are (a) mentalpredicates such as “think”, “believe”, “know”…and (b) sentential complement constructionssuch as “say that…”. If one can meaningfullyformulate sentences such as “I think that youthink that X”, then one should be able to thinkthe corresponding thought. Third, as pointed outby Tomasello (1999), not just the logical structureof language, but its use in discourse wouldpromote the understanding of others as “mentalagents”: There are at least “three kinds of discourse,each of which requires [children] to takethe perspective of another person in a way thatgoes beyond the perspective-taking inherent incomprehending individual linguistic symbols andconstructions.” (ibid: 173): disagreements, repairs/explanationsand meta-discourse.Empirical evidence for the existence of astrong connection from language to the understandingof beliefs has been accumulating duringrecent years and involves e.g. the following:• Deaf children who are not exposed to signlanguage at an early age pass false belief taskssignificantly later than those who are (Petersonand Siegal 1995);• There are correlations between parent use ofmental predicates in their child-directedspeech and the children’s performance infalse belief tasks (de Villiers and Pyers 1997);• Longitudinal studies indicate that languagedevelopment predicts false belief task performance,but not vice versa (Astington andJenkins 1999);• Training in (non-mental predicate) sententialcomplement constructions significantly improvesperformance on false belief tasks(Hale and Tager-Flusberg 2003; Lohmannand Tomasello 2003).• Training in perspectival discourse alone(without sentential complements or mentalpredicates) contributes to performance onfalse belief tasks (Lohmann and Tomasello2003).Our conclusion is therefore that the understandingof (false) beliefs is a form of post-mimeticintersubjectivity, in the sense that is based onlanguage, either spoken or signed. Notice thatthis does not contradict analyses by e.g. Bloom(2000) that the acquisition of language presupposes“theory of mind” skills, since we haveargued that the latter, including joint attentionand communicative intention are triadic mimeticphenomena, which also in our analysis are prerequisitesfor language.In one of the relevant studies, Call andTomasello (1999) used a non-verbal false belieftask with chimpanzees and orangutans as well aswith human children. The main results were thatthe children's performance on verbal and nonverbalfalse belief tasks was highly correlated,supporting the hypothesis of a possible causalconnection. At the same time, no ape could passthe nonverbal false belief task even though theysucceeded in all of the control trials indicatingmastery of the general task demands. However,as mentioned in Section 5.3, in this latter experimentthe enculturated and language-taughtorangutan Chantek performed differently fromthe other apes, displaying a better understandingof the experimenter’s signals as communicativesigns. A prediction from our analysis would bethat if Chantek, or any of the other “languageapes” that have been the subject of so muchcontroversy, could progress in their languagedevelopment to involve more mental terms, aswell as grammatical constructions involvingsentence complementation, they would also beable to pass false belief tasks.A counter-example to this Vygotskyan hypothesisof interpersonal communication precedingintrapersonal cognition would be to showthat non-enculturated apes are capable of understandingbeliefs (or at least third-order mentality),even outside of communicative settings. Acandidate domain is again deception, but asmentioned in Section 5.2, the kind of examplesinvolving deception, usually provided in supportfor theory-of-mind skills (e.g. Yeroen’s fake

22Bodily mimesis as “the missing link” in human cognitive evolutionlimp) can be explained by simpler skills. A furthercandidate is counter-deception. In a study performedwith chimpanzees by Menzel (1973,1974), only the female Belle was allowed to seewhere food was hidden within a large enclosedarea, and subsequently she led the group to thefood, and all were able to share it. When theexperiment was repeated, however, the maleRock began to take all the food for himself.After this, Belle did not reveal where the foodwas hidden if Rock was nearby. She sat furtherand further from the place where the food washidden, and it was only when Rock was lookingin another direction that she went to the hidingplace. Rock, however, countered this by pretendingto go away and then turning just as Belle wasabout to reveal where the food was. Byrne(1995) interprets this as implying that Rock hasan understanding that Belle is trying to deceivehim, i.e. third-order mentality. However,Tomasello and Call (1997) note that the experimentwith all its steps lasted several months soBelle and Rock may have successively learnedhow to predict the behavior of the other withoutunderstanding deceptive intentions. This hypothesisis supported by a later experiment withmangabey monkeys performed by Coussi-Korbel (1994), who replicated Menzel’s study,but performed a detailed day-to-day analysis.One of the informed mangabeys, step by step infour days, learned the strategy of leading hermain competitor away from the food.5.5 SummaryIn this section we have tried to show that thereis a close connection between different levelsof intersubjectivity and the different levels ofthe mimesis hierarchy, as summarized in Table4 below. Proto-mimesis is crucially implicatedin mutual attention and the awareness of others’feelings, through a species-general capacityfor empathy that has possibly been furtherdeveloped in the “ultra-social” species Homosapiens. Dyadic mimesis leads to the ability tomap between one’s own body and that of others,in a more detached, differentiated way, andin this way understand others’ emotions, i.e.cognitive empathy, shared attention and evenintentions through a conscious process of “projection”:what would I see/feel/wish if I wereyou.Unlike earlier claims to the contrary, newer evidenceand analyses show that apes do not havemuch difficulties with this level and that theyhave the capacity for second-order mentality.Table 4. The mimesis hierarchy and intersubjectivityLevel Intersubjectivity skills Type of mentalityProtomimesisDyadicmimesisTriadicmimesisPostmimesis(simple) empathymutual attentionCognitive empathyShared attentionUnderstandingother’s intention(in competitivecontexts)Joint attentionCommunicativeintentions(False) belief understanding1 st order: lack ofcomplete differentiationbetween selfand other2 nd order: understandingthe otherthrough “projection”(identification, butdifferentiation)3 rd order (attentionand intentions)3 rd order (beliefs)The crucial step in the evolution of intersubjectivityinvolves triadic mimesis, implying havingand understanding others’ communicativeintentions, which requires third-order mentality:“I want you to do X (e.g. share attention on anobject) by recognizing my intention that you dothis” from the sender’s perspective and “I understandthat you want me to do X” from therecipient’s.This is not something that comes easily toapes in natural conditions, but through enculturationand especially through extensive signuse, some understanding of communicative intentionsseems to be within the reach of apes’“Zone of Proximal Development”, even thoughin may be in its periphery. Evidence for this isthe relative mastery of joint attention by enculturants,and as argued this can be seen to originatein (dyadic mimetic) second-order attentioncombined with the understanding of the other’sintention that I attend. Finally, post-mimesis,with its features of conventionality and systematicitynecessarily brings with it sufficientunderstanding of third-order mental states, and(at least) second-order beliefs, e.g. “I think thatyou know (or don’t)”. The latter have beenshown in numerous false belief tasks, which areregularly failed by apes. On the other hand, there

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 23is mounting evidence that the acquisition oflanguage, with its structural properties and discursiveuse significantly contributes to higherforms of “theory of mind” development.In summary, in this section we have arguedthat bodily mimesis, in its proto, dyadic andtriadic forms, is a major factor in the developmentof intersubjectivity. Sign use itself wassuggested to be a driving force in the developmentof an understanding of third-order mentality,and the performance of enculturants showsthat this achievement is within the reach of apes.Therefore one can conclude that (mimetic) signuse was possibly within the “zone of proximalevolution” (Donald 2001) of the common apehumanancestor, considerably more so thanfully-fledged language, characterized by full conventionalityand systematicity.Finally, while we have focused on the mimesis-to-intersubjectivitydirection of causality, thisdoes not mean that there is no “feedback” linkand thus we assume a co-evolutionary scenario.It was implicit in the discussion that achievingdifferent levels of intersubjectivity is necessaryfor moving to still higher levels on the mimesishierarchy, e.g. joint attention for the acquisitionof language. It is also possible that in the presentanalysis we have overemphasized the Vygotskyanprinciple of the “interpsychologicalpreceding the intrapsychological” (see Section 1),but this has come rather naturally by defining thedevelopmental levels or stages in terms of bodilymimesis, with its originally interpersonal character.6. Mimesis and gestureOf the three socio-cognitive domains we discussin this article, gesture is the one most obviouslyrelated to bodily mimesis: at least some gesturesposses all the properties emphasized by Donald(1991): reference, intentionality, communicativity,autocuing and generativity (see Section 3).However, not all bodily acts that are called “gestures”have these properties. Furthermore, in thecurrently highly active field of gesture studies (e.g.Kendon 1981; McNeil 1992; Goldin-Meadow1999; Kita and Özyürek 2003), studying amongother things the spontaneous co-speech gesturingof human children and adults, it is oftenpointed out that gestures and language interactin complex ways. Thus, not all gestures are easilysubsumed under the label “non-linguistic” or“non-verbal”. In such a case, it becomes questionablewhether they are truly mimetic ratherthan post-mimetic.Perhaps in the most general sense gesture hasbeen defined as “a functional unit, an equivalenceclass of coordinated movements thatachieve some end” (Armstrong, Stokoe andWilcox 1995: 46; Wilcox 2004: 44), which encompassescommunicative as well as noncommunicativebody movements as long as theyposses the property of goal-directedness. This ismuch too general for our purposes and we willconsider as gestures only goal-directed communicativebody movements, i.e. such that require interpretation froman audience for achieving the gesturer’s goal. This includesa variety of acts such as attention getting,pointing and miming. At the same time, ourdefinition excludes bodily signals such as laughter,which is not goal-directed, or acts thatachieve their ends through “mechanical causality”rather than interpretation, e.g. pulling someone,as opposed to beckoning him. Notice thatwhile we focus on communication, we do notrequire all gestures to be intentionally communicative:what is essential is that the bodily motion isinterpreted by an observer as revealing somethingabout the state of mind of the gesturer.There are numerous classifications of gestures(see Kendon 1981 for a well-known taxonomy),but none, to our knowledge, that has explicitlytaken an evolutionary perspective – apartfrom the analysis of the origins of pointing (e.g.Brinck 2001). Our goal in this section is to applythe mimesis hierarchy to gesture, and in showinghow different types of gestures correspond tothe different stages of the mimesis hierarchy,suggest such an evolutionary taxonomy. We willalso show how the four stages of the hierarchycross-cut with two basic semiotic distinctions:indexicality and iconicity, which differ with respectto the semiotic ground (Peirce 1931-1935; Sonesson1989) for the relationship between the “signvehicle” (the expression) and the meaning of thegesture:• Indexical gestures, where the ground is oneof spatio-temporal contiguity, the prototypicalcase being pointing.• Iconic gestures, where the ground is ofsimilarity, with the prototype pantomime.Figure 2 illustrates the two types schematically:what is essential for indexical gestures, as with all

24Bodily mimesis as “the missing link” in human cognitive evolutionindexical signs, is that expression and contentare closely related in time-space. An iconic gesture(sign), on the other hand, requires similarity(often of a quite schematic type) but on theother hand it can be removed in time-space fromtheir meanings.Figure 2. Indexical signs, based on contiguity vs. iconicsigns, based on similarity. In both case expression andcontent need to be differentiated and perceived asstanding in a representational relationship by an interpreterfor there to be a (genuine) sign. Withoutthis, there are only indexicalities and iconicities.In agreement with Sonesson (1989, in press), wedistinguish between iconicity and indexicality asa type of semiotic ground, and the sign function(Piaget 1945). The latter presupposes an interpreter(signified by the “happy face” in Figure 2)who differentiates between the expression andthe content and is furthermore aware of therepresentational relationship between them:Only in this case do we have iconic and indexicalsigns. In the case of gestures, this means that onlyif the gesturing individual is aware of the expression-contentstructure of his gesture are thegestures intentionally used as signs. If not, thegestures are only a matter of indexicality or iconicityfor the gesturer; however, they may still beseen as signs by an interpreter, as is the case withthe first types of gestures discussed below.6.1 Proto-mimetic gesturesWe have defined proto-mimesis as involving across-modal mapping between exteroperceptionand proprioception, but lacking (clear) differentiationand or/volition. From this perspectivethe spontaneous reaching behavior of the childobserved soon after birth (von Hofsten 1983)and various forms of phylogenetic ritualizationmay be classed as proto-mimetic (for the gesturer),involving indexicality and iconicity, respectively,but not the sign function.6.1.1 Proto-mimetic indexicality: reachingIn the first 6 months of life, the human childcoordinates its hand movements and reachestowards interesting objects, including those thatit desires but cannot grasp because they are toofar away. At the same time there is no gaze alternationbetween the object and a third party, andno attempt to get the other’s attention – andpossibly help – in obtaining the object. Clearlythis is not a communicative gesture from thepoint of view of the child. However, at least intypical Western caregiver-child interactions atthis period of development, the adult often interpretssuch reaching as communicative andoffers the desired object. This appears to havean effect on the child’s interpretation of its ownaction through a process of internalization of theinterpersonal to intrapersonal (Vygotsky 1978,see Section 1). Eventually, this transforms reachinginto (proto-) imperative pointing, which, asdiscussed below, is a dyadic mimetic gesture.Ape infants reach out to objects that attracttheir attention similarly to human children, butless so to distal objects due to the fact that theybecome capable of locomotion much sooner.Furthermore, ape mothers are less “interventionist”and less likely to oblige their infants, evenwhen the latter do reach out to distal objects(Bard and Vauclair 1984). In general they interfereonly when they deem that there is a risk oftheir offspring getting in harm’s way. Such differencesmay be the major reason why feral apespoint much less (if at all) than human beings,since they are quite capable of acquiring pointinggestures when in contact with humans (see below).6.1.2 Proto-mimetic iconicity: phylogenetic ritualizationPhylogenetic ritualization is often proposed asthe basic mechanism for the emergence of animalsignals, as explained by e.g. Knight (2002:147):Over evolutionary time, certain aspects ofnon-communicative behaviour assume a signallingfunction, becoming correspondinglyspecialized through a process known as ‘ritualization’.Signal evolution begins when othersread some aspect of normal behaviour as significant.If the subject of such surveillance canbenefit from having its mind read or its behaviouranticipated, then over evolutionary

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 25time natural selection will accentuate the cues,reducing any ambiguity. By definition, suchphylogenetic ritualization entails special elaborationand added costs.Gestures such as gorilla chest beating (and otherforms of display) and the “submissive gesture”of apes, in which they crouch and “make themselvessmall” involve iconicities (resemblances) tobody size – the larger the animal is, the louderthe signal – and state of agitation – the strongerthe emotion, the more beating. At the sametime, they are not iconic signs, since the similaritybetween the signal and the non-communicativebehavior is not perceived (at least not necessarily)by either the gesturer or the interpreter.Hence, while there is communication, there is norepresentation (in the sense of expressioncontentstructure, see Ikegami and Zlatev inpress) involved. That is why we can call suchphylogenetically ritualized signals proto-mimeticgestures.It is characteristic that when observed inhighly enculturated apes such as the signed languagetaught gorilla Koko, gestures such as“gimme” and “pound” were the first to develop,with the least effort on the part of the trainer. Aspointed out by Patterson (1980: 517):These were distinctive because they appearedwithout direct training at a time when thegreat majority of Koko’s signs were being acquiredby molding. Because other young gorillasreared in captive conditions who have notbeen exposed to sign language use these samegestures, they must be part of the gorilla’snatural repertoire.6.2 Dyadic mimetic gesturesIn the case of dyadic mimesis there is representationin the sense that there is both differentiationand correspondence between the gesture and its“meaning”, in this case the action that the gesturerwishes that the addressee will perform. Inthis respect, such gestures constitute signs, in thesense of the term defined at the beginning ofthis section. However, they have two major limitations:(a) their intended meaning is always adesired action rather than a statement, i.e. theyare imperative rather than declarative, and (b)they involve second-order mentality: a claim onthe attention and behavior of the addressee, butnot third-order mentality: there is no appeal onthe addressee to understand the gesturer’s ownmentality. In our terminology, they do not involvethe communicative sign function, unlikethe triadic gestures that we discuss in 6.3.6.2.1 Dyadic-mimetic indexicality: imperative pointingIn the human infant literature “reaching” is oftenused as a label for acts of intentional communication,rather than to attempts at pretension.Similarly to Leavens and Hopkins (1999),in their informative review of studies of pointingin children and apes, we find that even if thechild does not shape the hand and arm in a “canonicalpoint” (Butterworth 1998), given thatthere is evidence for gaze alternation between anobject and a third party, the act should be regardedas an instance of pointing. In humanchildren, such acts emerge spontaneously from 8months of age, and at least at first have an “imperative”function of attempting to make the addresseeperform some desired action. From this it followsthat imperative pointing requires secondordermentality, but not necessarily joint attentionand third-order mentality (see Section 5.2,Brinck 2003). Hence, imperative pointing can beclassified as a dyadic mimetic gesture. This conclusionis also supported by Tomasello’s (1999)observation that in this stage of development,children can master intentional pointing withoutunderstanding the pointing of others. Anotherkind of pointing that is perhaps even more obviouslydyadic mimetic that is also observed at thisstage is “pointing for oneself”, i.e. when thechild is using pointing gestures to direct its ownattention, rather than that of others (Bates,Camaioni and Volterra 1975).When it comes to apes, it was earlier thoughtthat apes do not point (Donald 1991) or at leastnot with the index finger (Povinelli and Davis1994). This has been recently shown to be untrue.Leavens and Hopkins (1999) summarizesome of the evidence showing that pointing hasbeen reported for 15 captive monkeys, 83 captiveapes, and even one wild ape, a bonobo whowas pointing at a group of humans while lookingback and forth between the humans and theother bonobos (Veà and Sabater-Pi 1998). 77 If it was not for the somewhat anecdotal characterof this piece of evidence, it would pose a problemfor our classification, since it appears to involvedeclarative pointing, which is otherwise observedonly in enculturated apes.

26Bodily mimesis as “the missing link” in human cognitive evolutionWhether or not feral apes really point, the evidencefrom captive apes (e.g. Call and Tomasello1994; Leavens, Hopkins and Bard 1996, Krauseand Fouts 1997; Leavens and Hopkins 1999)clearly shows that the criteria for imperativepointing are satisfied: attention getting, gazealternation and persistence until the goal is met.Leavens and Hopkins calls these audience effects,and observe that captive chimpanzees engagedin imperative pointing exhibit more audienceeffects, and above all more gaze alternation thanchildren at the corresponding stage of development.At the same time, the children were considerablymore prone to vocalize while pointing,showing the close association between bodilymimesis and speech in human beings.However, if we limit our attention to themanual gestures alone we can conclude that atleast when exposed to human contact, apesclearly display imperative pointing, requiringdyadic mimesis. Similarly, there is evidence thatat least enculturated apes engage in dyadic, noncommunicativepointing in examining objectsfor the purpose of directing their own attention(Bonvillian and Patterson 1999). Interestinglysuch non-communicative pointing emergedearlier than communicative pointing (of both theimperative and the declarative type) in the signlanguage taught gorilla Koko, similar to the casewith children (Bates, Camaioni and Volterra1975).6.2.2. Dyadic mimetic iconicity: ontogenetic ritualizationAs pointed out in 6.1, most animal signals arethe result of phylogenetic ritualization, and are inthis sense innate. But monkeys and especiallygreat apes use a good number of learned gesturesas well. Tomasello et al. (1994) distinguishedsix functional categories of gestures ininfant and juvenile chimpanzees involving nursing,walk-riding, grooming, eating, appeasing andplay/tickle. These were believed to have comeabout through a process called ontogenetic ritualizationin which the gestures were shaped by respondingto each other’s anticipations. Thishappens over the life of the individual ratherthan the species, hence the term given to thislearning process. Tomasello and Zuberbühler(2002: 295) describe ontogenetic ritualization asa sequence of 4 steps:(1) Individual A performs behavior X.(2) Individual B reacts consistently with behaviorY.(3) Subsequently B anticipates A’s performance ofX, on the basis of its initial step by performing Y.(4) Subsequently A anticipates B’s anticipationand produces the initial step in a ritualized form(waiting for a response) in order to ellicit Y. [originalemphasis]This way of describing the emergence of suchgestures sounds rather behaviorist, implying thatit is only a matter of (co-)reinforcement of behavioralpatterns. However, Tomasello et al.(1994) showed that most chimpanzee gesturesinvolve getting the other’s attention and that thegestures vary with the attentional state of therecipient: non-tactile, non-vocal gestures arepredominantly used when there is visual contactbetween the recipient and the gesturer. Furthermore,the animals clearly waited for a responseafter gesturing. The presence of such “audienceeffects” indicates that there was an intention onthe side of the gesturing chimpanzee to producean effect on the addressee. However, this intentionto communicate is not the same as a communicativeintention, which is an intention that thedesired effect will be produced by the addresseerecognizing the intention to produce the effect(Grice 1957; Sperber and Wilson 1995). As arguedin the Section 5, such forms of third-ordermentality appear to be beyond the capacity ofnon-enculturated apes.At the same time, to the extent that the gesturingape is differentiating between its (ontogenticallyritualized) gesture and the desiredeffect and being more or less aware of the correspondencerelation between them, it is justifiedto regard the two as standing in a representationalrelationship – even without involving acommunicative sign function. However, thisrepresentational relationship is fairly unstable.Tomasello et al. (1994) showed that chimpanzeeshad a limited arsenal of gestures that theyapplied highly “flexibly”: different gestures wereused for the same goal, and the same gesture wasoften used with different “meanings”. On theone hand, this speaks for these gestures beingintentional and different from species-specific

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 27signals. But on the other hand, it testifies to theirbeing non-conventional, and non-systematic.Similar findings have been replicated foryoung gorillas (Pika, Liebal and Tomasello 2003)and bonobos (Pika, Liebal and Tomasello 2005).However, in these more recent (and systematic)studies, a number of gestures where found to begroup-specific, and therefore “some form ofsocial learning” (Pika et al. 2003: 108), besidesontogenetic ritualization was considered to bethe mechanism behind their emergence. Furthermore,while Tomasello et al. (1994) did notobserve any social transmission of a specificbegging gesture learned by a chimpanzee toacquire rewards from human experimenters,Pika et al. (2005) summarize a number of differentstudies finding evidence for the social learningof gestures in great apes. It is therefore notpossible to exclude some degree of imitation inapes’ learning of gestures. Even so, that does notimply that such gestures, e.g. the “arm shake” ofthe gorillas of Apenheul zoo (Pika et al. 2003)are more than dyadic mimetic, since it is notclear that they involve the communicative signfunction: e.g. represent something for someone.Some dyadic mimetic gestures may also involveiconicity (similarity) between the spatiotemporalprofile of the gesture and the meaningof the gesture: the action to be performed by theaddressee. An example would be to gently rubdown someone’s back in the desired direction ofmovement instead of forcefully pushing theother down. A purely visual version would be tomake the gesture in the air in view of the recipient.Such gestures are usually labeled iconic (Tannerand Byrne 1996, 1999) and we will followthis practice. However, we still need to askwhether they function as iconic signs for both thesender and the receiver, i.e. the iconicity isgrasped as the ground for the representationalrelationship. What characterizes dyadic mimesis,as we stated above, is that this relationship is (a)unstable and (b) non-symmetrical: it is possiblethat only one of the communication parts perceivesthe iconicity as an iconic sign.In a study of captive bonobos by Savage-Rumbaugh et al. (1977) almost all gestures thatpreceded copulation appeared to be signs ofcopulatory positions. Since bonobos use manydifferent copulatory positions it is expected thatsome mutual agreement must be reached aboutwhich position to be used. Two animals couldsometimes “argue”, in the sense that one gesturedfor a ventro-ventral position while theother gestured for a dorso-ventral one. The likelihoodof unusual copulation postures increasedif the bout was preceded by gesturing. Thesegestures were grouped into three groups: (1)positioning motions: the recipient’s limbs aremanipulated by the initiator. The actual movementis done by the recipient but the direction isset by the initiator; (2) touch plus iconic handmotions: a touch on the part of the body thatshould be moved, followed by an indication withthe hand how this part should move; (3) iconichand motions: without touching the other individualthe gesturer shows what he wants theother to do; these are the most abstract ones.It is possible to interpret this progression asthe development of an understanding of iconicityand an emergence of iconic signs (Savage-Rumbaugh et al. 1977), but it should be notedthat the gestures from the first group were themost frequent in the bonobo group, and gesturesfrom the most abstract level, the thirdgroup, was the least frequent. In other studies ofbonobos (de Waal 1988; Pika et al. 2005) andgorillas (Pika et al. 2003) no gestures could bereadily interpreted as iconic in nature. So even ifthe evidence seems to grant apes some understandingof iconic signs, predominantly emergingthrough ontogenetic ritualization, this understandingdoes not seem to be reflective enoughto be systematically used in communication.6.3 Triadic mimetic gesturesAs pointed out repeatedly above, what is lackingin dyadic mimetic gestures is an understandingof the communicative sign function, which impliesthat the gesturer not only wishes the addresseeto perform some desired action, but todo this by appreciating the representational(sign) relation between his gesture and its referentialmeaning. As we show in this section apescan achieve such an understanding, but clearlyonly when they have been taught aspects of aconventional symbolic system, i.e. a language.This raises some doubts whether such gesturesare not actually post-mimetic.6.3.1 Triadic mimetic indexicality: declarative pointingThe third stage in the development of humanchildren’s pointing, mastered around 14 months,

28Bodily mimesis as “the missing link” in human cognitive evolutionis usually called declarative pointing (Bates,Camaioni and Volterra 1975; Brinck 2003). Thecrucial difference with respect to imperativepointing is that the child need not desire theobject pointed, but rather the act of achievingjoint attention with the addressee on the objectis the goal in itself. As argued in Section 5.3,joint attention involves understanding a simpleform of communicative intention, and is thustriadic mimetic. In this sense declarative pointingis also triadic mimetic.Brinck (2003) proposes that at an early stageof a child’s use, an additional purpose of declarativepointing is the evaluation of the indicatedobject, achieved by an exchange of emotionalinformation about it. For example, the childpoints to an object in order to obtain informationfrom the addressee whether the object isdangerous. The main benefit for the child of thiskind of exchange is that he/she can learn aboutobjects vicariously. This primary function presupposesthat the child can understand the emotionsof the addressee, and as we argued in Sections5.1 and 5.2, this is well within thecompetence of the child at this stage of development,and similarly for apes. Thus, it is possiblethat the evaluative side of declarative pointingwas the original cause of its proliferationamong the hominids (Brinck 2001).Enculturation has been shown to have dramaticeffects on the pointing of apes. While feralapes almost never point, as mentioned earlier,captive apes and monkeys readily learn to engagein imperative, though not declarative pointing.On the other hand, at least three of the documentedlanguage-taught apes – Koko, Chantekand Kanzi – consistently point declaratively, i.e.refer to objects and locations for sake of callingthese to the attention of someone else (Patterson1980; Miles 1990). In the mentioned review ofpointing in human children and apes, Leavensand Hopkins (1999), point out that languagetrainedapes were more likely to use a canonicalpoint with the index finger than children of theage 13-18 months. In some studies, it is evensuggested that language-taught apes combinesuch an act of declarative pointing and anothersign to make a predication, i.e. THIS is X(Greenfield and Savage-Rumbaugh 1990).However, the question arises if such acts arecases of triadic mimesis, rather than postmimeticskill. The close relation between declarativepointing and reference makes it likelythat the process of being taught language itself iswhat has simultaneously taught the significanceof declarative pointing to the apes. The followingepisode recorded by Savage-Rumbaugh andLevin (1994), involving not the famous, highlyenculturated bonobo Kanzi, but the somewhatmore traditionally trained Austin and Sherman,shows how enculturation, and specifically theteaching of communicative sign function, canaffect pointing:Upon spying the new food, Sherman reachedout for it and I suggested that he should askfor it. He quickly turned, scrutinized the keyboard,and deliberately, but with no fanfare,indicated one of the unassigned lexigrams asthe symbol for the food. Austin watched attentivelyand subsequently used the symbolSherman had selected to indicate that hewould like to try the new food also. This unexpectedskill encouraged us to believe thatSherman and Austin understood that a uniquecorrespondence existed between each fooditem and a specific symbol. Moreover, theysometimes pointed back and forth betweenthe symbol and the food item. They hadnamed new foods and apparently agreed oncoordinated use of the selected symbols. (ibid:79)We will return to this issue in the summary below.6.3.2 Triadic-mimetic iconicity: communicative intention?Possibly the clearest evidence for iconic triadicmimetic gestures in non-language-trained apescomes from a study of zoo-living gorillas, partiallyreared by human caregivers. Tanner andByrne (1996, 1999) studied a pair of westernlowland gorillas, Kubie and Zura, at the SanFrancisco Zoo, who seemed to have an unusuallyrich repertoire of gestures: audible, silent andtactile ones. The gestures were used almost exclusivelyin play sessions and sexual situations,and the choice of gesture differed according tothe attentional state of Zura, with audible andtactile ones used when she was not looking,which mirrors the findings for chimpanzees byTomasello et al. (1994).Some of Kubie’s gestures were “attentiongetters” and play invitations. His most successfulgesture was an armswing under his belly towardshis genitals. It had the highest frequency of accompanying“play faces” and resulting in physi-

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 29cal contact. It was only performed when thegorillas had visual contact. “Armswing under”was often preceded by a tap on Zura’s body.This is comparable with Savage-Rumbaugh etal.’s (1977) second group of iconic gestures:touch plus iconic hand motions. If Zura failed torespond, a gesture was often repeated, strengthenedor more gestures were made – this alsoresembling the pattern found in chimpanzees(Tomasello et al. 1994). Most of the gesturesended in contact between Kubie and Zurawithin 5 seconds, with Zura making most of thecontact, suggesting that she was (correctly) interpretingKubie’s gestures. In 66% of the caseswhen a tactile gesture was made Zura moved herbody in the direction indicated by Kubie. If agesture was coupled with a “play face” the likelihoodof physical contact was significantly higherthan for gesture alone. However, “head nod”,Kubie’s most frequent gesture, had a high degreeof play bouts without using “play faces”, whichshows that Zura did respond to gestures and notjust to the natural (pre-mimetic) gorilla signals.Still, the question remains if Zura really respondedto their (intended) meaning, comprehendingthe gestures as iconic signs. Associatingplayfulness (or sexual agendas) with particularbehaviors in another individual’s repertoire isnot sufficient to show that Zura understoodKubie’s gestures as communicative signs.Whether or not Kubie’s gestures had an iconicrelationship to their referents for him, they couldin each case remain indexical to Zura, i.e. bebehavioral correlates to moods. The iconicity ofsome of Kubie’s gestures could very well havebeen abstractions from more physical manipulationsshaped through ontogenetic ritualizations(Tomasello and Call 1997) and Zura’s appropriateresponse be derived from the context.However, over time Kubie used his gestureswith several females, and if he did not gothrough an identical abstraction phase with all ofthem, which of course cannot be ruled out completely,how could the receiver understand them?Tanner and Byrne (1996) suggest that comprehensionof motion depicted in gesture may bebiologically encoded for the great apes sincetheir adaptation to brachiation requires a threedimensionalunderstanding of space, and space isthe medium for gestures. 88 Brachiation also entails an ability that is unique toapes: that of being able to move the wrist, elbowIn summary, like the bonobo sexual gesturesdescribed in 6.2.2, the San Francisco gorilla gesturescould be argued to be at least dyadicmimetic,since they correspond to actions intendedto be performed by the recipient. Butthrough their iconicity the gestures also have apotential to be triadic-mimetic, under the conditionthat at least the recipient, and possibly alsothe sender understood their sign function. Triadicmimesis cannot be excluded as an alternativeto ontogenetic ritualization in the formationand use of such gestures because nonconventionalgestures could be understood notonly through the context but through the iconicityof the gestures themselves. If, and it is a big“if”, Kubie, Zura and the other females coulduse the iconic gestures effectively because ofsuch referential understanding, Kubie’s gestureswould be an example of triadic mimesis. Moreempirical research is necessary to decide thisissue.A final point should be made in relation tothe differences observed between feral and enculturatedapes pointed out earlier. The fact thatZura, the other female gorillas and Kubie all atsome point had been reared by humans (Tannerand Byrne 1996) might have implications for theconclusion we can draw about mimetic capacitiesin non-enculturated apes. Human beingsengage readily in triadic forms of communicationand apes with extensive exposure to humanlanguage show good comprehension of otherrepresentational mediums, like words and pictures(Savage-Rumbaugh, Shanker and Taylor1998). However, it is not yet known how muchexposure to human culture is needed to yield anunderstanding of triadic mimesis (if at all) inhand-reared zoo animals, and if this would beretained into adulthood.While the presence of iconic signs in captiveapes remains unsure, there is solid evidence fortheir use in enculturated apes. The sign-languagetaught gorilla Koko reportedly invented a numberof such signs herself, including signs referringto actions e.g. “blow”, “tickle”, “walk-myback”,“bite”, as well as objects: “bird”, “bracelet”,“stethoscope”. These gestures clearly involve areferential triangle involving the gesture, referentand the interpreter, and were sometimes evenused in the absence of the referent, showing oneof the crucial features of language: displacementand shoulder joints in all directions, making subtlegestures possible.

30Bodily mimesis as “the missing link” in human cognitive evolution(Hockett 1960), and even “metaphorically”, e.g.using “bird” as an insult.A problem for classifying them as triadicmimetic, however, is that these gestures werequite similar to the actual signed language signsthat she was taught. In accordance with the caveatstated earlier, it is difficult to exclude theinterpretation that they are actually postmimetic,i.e. induced by the language-teachingsituation itself.Some support for their mimetic-iconic natureis offered by the fact that Koko produced manyso-called gestural modulations with respect to size,manner, number, etc. – all having an iconic basis.For example, the signs for a “small bird” anda “big bird” would differ in size; those for “severalbirds” would involve repetition of the sign.Thus, gestural modulations are perhaps theclearest evidence of the presence of triadic mimesisin Koko’s productions, especially sincePatterson (1980: 539) emphasizes that they“were created by Koko herself”. Other testimonyfor the spontaneous production of triadicmimetic gestures in a language-trained ape canbe found in the spontaneous gestures of thebonobo Kanzi, who unlike Koko was not taughtsigned language but a combination of visuallexigrams (in production) and spoken English (incomprehension). Greenfield and Savage-Rumbaugh (1990, 1991) report on Kanzi’s spontaneoususe of pointing (see below), and at leastsome gestures involving iconicity such as“come”, “go” and “chase”.6.4 Post-mimetic gesturesMimesis was defined by Donald (1991) to be“non-linguistic”, and in Section 3 and in ourdefinition of bodily mimesis we clarified what wetake this to mean: lack of full conventionality(mutual knowledge) of the signs and systematicity,involving at least a degree of compositionality.To the extent that the gestures of apes displaythese properties, they can be regarded aspost-mimetic.Evidence from four of the most successfulprojects involving the teaching of language togreat apes – the chimpanzee (Pan troglodytes)Washoe (Fouts 1972, 1973), the gorillas Kokoand Michael (Patterson 1978, 1980) and thebonobos (Pan paniscus) Kanzi and Panbanisha(Savage-Rumbaugh and Lewin 1994; Savage-Rumbaugh et al. 1998) and the orangutan Chantek(Miles 1990) – has shown that many of theessential characteristics of language are withinthe grasp of our nearest non-human relatives. Asis the case with children a precondition for thesuccess of these projects has been a culturalenvironment rich with intersubjectivity and avariety of activities to stimulate communication(Miles 1990). Our conclusion is similar to that ofMiles (1999: 204), namely that all great apes“have the intelligence for a rudimentary, referential,generalizable, imitative, displaceable, symbolsystem”.The reason that this claim has been controversial(e.g. Terrace et al. 1981) is not foremostempirical, but rather a matter of a disagreementon what the “essential characteristics” of humanlanguage are. Turning to the data reported in the“ape language” literature we see convincing evidencethat apes can:• comprehend the referential (representational)function of spoken words, ASL signs andvisual lexigrams, and combinations of these;• use the sign-tokens in the absence of theirreferents, i.e. “displacement” (Hockett 1960);• acquire a considerable vocabulary ofwords/signs, according to some measurementsextending 600 signs, but even accordingto the most conservative criteria no lessthan 140 signs;• regard the acquired signs as conventionalnormative(consensual), to the point of correctingtheir teachers if the latter do not usethese appropriately;• understand novel combinations of spoken orsigned words;• produce novel combinations of signs;• use language for a number of different functions(speech acts), including labeling, answering,expressing emotion, arguing and insulting;• use language not only for communication,but for thinking (private speech).What has not yet been clearly demonstrated iswhether the spoken or signed utterances of apesconform to consistent principles of grammaticalorganization, though e.g. Greenfield and Savage-Rumbaugh (1990, 1991) describe at least twofairly consistent ordering rules in the two symbolcombinations of Kanzi, one of them involving apointing gesture (see below).It is thus unsurprising that those who have(over-)emphasized syntax as opposed to seman-

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 31tics as the defining characteristic of language(e.g. Bickerton 1990; Pinker 1994; Hauser,Chomsky and Fitch 2002) continue to deny “theproposition that language is no longer the exclusivedomain of man” (Patterson 1978: 95). Buteven if the language of apes such as Koko andKanzi is termed “proto-language” rather thantrue language, it is still qualitatively differentfrom mimesis, since it possesses at least rudimentsof conventionality and systematicity. Inthis sense, some, though not all, of the gesturesof the language-taught apes can be regarded aspost-mimetic.6.4.1 Post-mimetic indexicality: paralinguistic pointingAt least two forms of the pointing of languagetrained apes seem to go beyond mimesis, sincethe pointing gesture is integrated in the (proto-)linguistic system acquired. First of all are thecombinations of a pointing gesture referring toan object or an individual with another sign expressinga predicate applying to the individual,thus forming a predication. Furthermore, thebonobo Kanzi has been reported to form such“predications” – with imperative as well as declarativefunctions – in a fairly systematic manner(Greenfield and Savage-Rumbaugh 1990,1991).A related form post-mimetic indexicalityinvolves labeling, in which the pointing gesturedirected at an object, or a picture of one in abook, is accompanied with the production of(gestural) sign denoting the object. Such clearlydeclarative uses of pointing + label, some ofwhich are performed “privately”, i.e. not for thebenefit of a human interlocutor (Bodamer et al.1994) are perhaps the most convincing case forthe presence of post-mimetic, paralinguisticpointing in enculturated apes. Kanzi has beenshown to use pointing + lexigram, as well aslexigram + pointing combinations for the sakeof labeling quite frequently: altogether therewere 249 examples in his five-month corpus oftwo-element combinations in which Kanzi indicatedan entity gesturally and named it with alexigram (Greenfield and Savage-Rumbaugh1991).6.4.2 Post-mimetic iconic gestures: Signed languageHuman signed languages offer the clearest pictureof post-mimesis. Recent studies of thespontaneous emergence of Nicaraguan SignLanguage (NSL) during the past 25 years showthat signed languages have their origin in (triadic)mimesis, but quickly acquire the propertiesof conventionality and systematicity. Senghas,Kita and Özyürek (2004) compared the cospeechgestures of Nicaraguan speakers of Spanish,with the signing of three “cohorts”, or generations,of learners of NSL and could documentsome aspects of this transition in detail:The movements of the hands and body in thesign language are clearly derived from a gesturalsource. Nonetheless, the analyses reveal aqualitative difference between gesturing andsigning. In gesture, manner and path were integratedby expressing them simultaneouslyand holistically, the way they occur in the motion[event] itself. Despite this analogue, holisticnature of the gesturing that surroundedthem, the first cohort of children, who startedbuilding NSL in the late 1970s, evidently introducedthe possibility of dissecting outmanner and path and assembling them into asequence of elemental units. As second andthird cohorts learned the language in the mid1980s and 1990s, they rapidly made this segmented,sequenced construction the preferredmeans of expressing motion events. NSL thusquickly acquired the discrete, combinatorialnature that is hallmark of language.Given their mimetic-gestural origin signed languageshave a much greater degree of iconicitythan spoken languages and it has been proposedthat this plays a role in their faster acquisition by(deaf) children (Brown 1977). Recent studieshave questioned this, however, since only a minorityof the signs of signed language havetransparent iconic meanings, and in a study of 22children acquiring ASL it was shown that “of the44 different signs produced by the children beforethe age of 13 months, 36% were classifiedas iconic, 30% as metonymic, and 34% as arbitrary”(Bonvillian and Patterson 1999: 253). 9 In9 “Metonymic” signs are such that involve somedegree of iconicity between the sign and the referent,but ”the tie between the sign and its meaning isnot readily apparent – one would be unlikely toguess the meaning of a metonymic sign simplyseeing it produced” (Bonvillian and Patterson 1999:252). In the distinction introduced by Sonesson

32Bodily mimesis as “the missing link” in human cognitive evolutionthis study, the authors compared the rate andpattern of acquisition of ASL by children andthat of two gorillas, Koko and Michael. Despitecertain differences – the children’s acquisitionwas (unsurprisingly) faster – it was shown “thatthe similarities in early development across thespecies outweigh the differences” (ibid: 260).Thus, it can be concluded that gorillas, and byinference other great apes, not only can acquirethe basics of a post-mimetic symbolic systemsuch as ASL, but that they do this is a similarway. Interestingly, however, the gorillas seemedto rely somewhat more on the iconicity of thesigns in comparison with the children, so thatthe proportion of their first 46 signs was somewhatdifferent to the one reported above: 42%iconic, 32% metonymic and 26% arbitrary, whilefor the first 10 signs this difference was evenclearer: 60% iconic, 20% metonymic and 20%arbitrary. We interpret this as suggesting that theapes relied to a greater degree on their skills oftriadic mimesis than the children in their acquisitionof the sign language, which also would explainwhy the children quickly progressed beyondthe initial level of vocabulary acquisition tolearn the systematic character, i.e. the grammar,of the language, while the apes “stagnated” on asimple, “proto-language” level.6.5 SummaryIn this section we have reviewed some of theevidence on the production, and to some extentthe comprehension of gestures by great apes,and when relevant compared it to the acquisitionof gestures and sign languages by children. Applyingthe mimesis hierarchy to this evidence hasresulted in the classification summarized in Table5.Some of the natural signals of apes can beregarded as involving indexicality and iconicity,without these being perceived as signs by eitherthe sender or the receiver. On the other hand, ifan observer and, especially, a caregiver duringinfancy does see these as signs, this can providean impetus for making these gestures mimetic, asseems to be the case with the origin of imperativepointing. At least some ontogenetically ritualizedgestures appear to involve differentiationand correlation between the gesture and mean-(2001) they involve secondary (rather than primary)iconicity, which can be perceived first after knowingthe specific sign relationship.ing, and thus can be regarded as simple iconicsigns. Since these, however, can function withouthaving and interpreting communicative intentions,they are (at most) at the level of dyadicmimesis. Feral and even captive apes have notbeen convincingly shown to progress beyondthis level.Table 5. The mimesis hierarchy and ape gesturesLevel Indexical gestures Iconic gesturesProtomimesisDyadicmimesisTriadicmimesisSome ontogeneticallyritualized gesturesPostmimesisReachingPointing foroneselfImperativepointingDeclarativepointingParalinguisticpointingSome phylogenticallyritualized signalsIconic signs(Aspects of) signedlanguageEnculturation, and especially languageteachingchanges this and consequently bothdeclarative pointing and iconic gestures emerge,testified in individuals such as the gorilla Koko,the bonobo Kanzi and the orangutan Chantek.We interpret this as evidence for triadic mimesis,though with some hesitation, since it is likelythat being taught language was necessary for theapes to acquire the communicative sign function.Even so, triadic mimesis can be shown to bewithin the grasp of our nearest non-human relatives.In comparison, their skills in post-mimesisinvolving full conventionality and systematicity,i.e. language, are more limited. The conclusion,similar to that at the end of Section 5, is thattriadic mimesis is within the “zone or proximalevolution” of great apes, and by inference, toour common ancestor.7. Summary and conclusionsThe goal of this study has been to explore thepotentials of bodily mimesis to provide perhapsnot “the” missing link in explaining human cognitiveevolution, but at least one important missingpiece in the puzzle, and thus help unravel themystery of the origins of language. In Section 3we defined the concept of bodily mimesis andthe related mimesis hierarchy, thereby refining andmodifying the concept of mimesis proposed by

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 33Donald (1991). Our method has been to applythe mimesis hierarchy to evidence from primatology(and to some extent child development)in order to be able to compare ape skills in threedifferent domains: imitation, intersubjectivityand gesture and to determine whether any generalpatterns could be discerned. Table 6 belowsummarizes our results.Beginning with proto-mimesis, we could seethat there is abundant evidence from all threedomains that apes excel in this. There do notappear to be any (major) cross-species differenceswith humans in this respect, and we canassume with some confidence that the commonape-human ancestor had comparable skills inimitation, intersubjectivity and communicativegestures.On the next level, that of dyadic mimesis, webegin to discern some differences – both betweenthe levels, and between humans and apes.There is clear evidence for e.g. cognitive empathyand ontogenetically ritualized gestures inapes – and we suggested how these phenomenacan be explained in a dyadic-mimetic fashioninvolving both identification with and differentiationfrom the other. However, when it comesto imitation, apes appear to have difficulties inperforming detailed imitation on the action level,which is the type of imitation where bodily mimesisis most relevant. Human children in comparisonexcel in this and even “overmimic”. Thisdifference between the species does not appearto be due to a difference in intersubjectivity, i.e.understanding of others’ intentions, as wasoriginally suggested by Tomasello (1999) butrather in mimetic capacity per se. Thus, our resultsconfirm, at least in part, Donald’s proposalthat (bodily) mimesis constitutes the phenomenonin relation to which we should look forhuman cognitive specificity. At the same time,the results reviewed in Section 4 showed clearlythat apes have the basic capacity for dyadic miFinally, the capacity of apes for post-mimesis, relyingon the possession of a conventional symbolicsystem, is furthest away from the potentialof human beings. While enculturated apes suchas Kanzi may have mastered a proto-languageafter a good deal of exposure and effort, thisremains limited in grammatical and semanticcomplexity, approximately to the level of a twoyearold child. Thus, even post-mimesis is notcompletely beyond the ZPD of apes, though it isfurther in the periphery than triadic mimesis.Using the analogy suggested by Donald (2001),we can say that both triadic mimesis and postmimesisare within apes’ “zone of proximal evolution”,i.e the sphere in which evolutionaryadaptations are likely to occur, but since postmimesispresupposes triadic mimesis (even conceptually),the latter constitutes the most likelymissing link.Table 6. Comparing ape skills in the domains of imitation, intersubjectivity and communicative gestures along thelevels of the mimesis hierarchy: each level mentions example capacities, with “+” indicating positive evidence and “–” negative evidence. W = wild, C = captive, E = enculturated.DyadicmimesisTriadicmimesisLevel Imitation Intersubjectivity Communicative gestureProtomimesisNeonatal mirroringC +Mutual attentionEmpathyC +Phylogenetically ritualized gesturesW/C +PostmimesisAction level imitationC + (?)Imitative signsW –C –E + (?)Role-reversal imitation–Shared attentionCognitive empathyC +Joint attentionW –C –E +Belief understanding–Ontogentically ritualized gesturesImperative pointingC +Declarative pointingW –C + (?)E +Proto-languageE +

Jordan Zlatev, Tomas Persson and Peter Gärdenfors, Lund University Cognitive Studies, 121, (2005).What kind of factors brought this about inevolution?The evidence summarized and analyzed inthis article allows us to make two high-levelgeneralizations concerning the three domains ofbodily mimesis analyzed:(1) Ape skills of imitation are in comparisonwith other skills least developed along themimesis hierarchy;(2) Ape skills of gesture are in comparison withother skills most developed along the mimesishierarchy.Furthermore, we could discern a possible causalrelationship in the direction from gesture (signuse) toward intersubjectivity: the only apes whosucceed in tasks of joint attention, where thosewho where taught to use sign system throughenculturation. (In the case of imitation the relationis more of a logical sort: since triadic imitationis defined as the imitation of signs, it is impossibleto perform this without having firstlearned the latter.) Thus we can conclude bysuggesting the following two (partially independent)hypotheses of human cognitive evolution,corresponding to the two generalizations givenabove:(1´) The crucial adaptation that differentiatedearly hominids from apes involved detailedbodily imitation, allowing for the developmentof specific mimetic schemas which are prelinguisticrepresentational, intersubjectivestructures, emerging through imitation butsubsequently interiorized (see Piaget 1945;Zlatev 2005a; Zlatev, Persson and Gärdenforsin press).(2´) Since apes display some potential for spontaneouslydeveloping indexical and iconicgestures – e.g. in the rather unusual environmentof the San Francisco Zoo, see Tanner(2004) – it is reasonable to assume that thecommon ape-human ancestor possessed sucha potential as well. An “ecological niche” requiringa high degree of intentional communication(there are various scenarios forbringing this about; for a proposal regardingthe role of Oldowan material culture (Osvathand Gärdenfors 2005) would have spurredthe use of iconic and indexical gestures,which on their part would have both utilizedand further developed mimetic schematicrepresentations.In sum, our general proposal is that thesource of human cognitive specificity lies aboveall in two, mutually co-enforcing factors involvingbodily mimesis: (a) the capacity to form mimeticschemas: representations deriving from theimitation of public events and (b) the use ofmimetic schemas for intentional communication:triadic mimesis. The first of these factors is predominantlyrepresentational and the second,communicative, and it is likely that in evolutionthe two have co-evolved. We submit that thisproposal is consistent with the evidence involvingthe capacities and limitations of apes, andshows the explanatory potential of the conceptof bodily mimesis as “a missing link” in ourtheory of cognitive and linguistic evolution.ReferencesAhlgren, I. (2003) Teckenspråk. In: Sveriges officiellaminoritetsspråk. Finska, meänkieli, samiska, romani,jiddisch och teckenspråk. En kort presentation.Nordstedt.Anisfeld, M. (1996) Only tongue protrusion modelingis matched by neonates. Developmental Review, 16:149-161.Arbib, M. (2003) The evolving mirror system: Aneural basis for language readiness. In: LanguageEvolution, ed. M. Christiansen and S. Kirby, OxfordUniversity Press.- (in press) From monkey-like action recognition tohuman language: An evolutionary framework forneurolinguistics. Behavioral and Brain Sciences.Armstrong, D., Stokoe, W., and Wilcox, S. (1995)Gesture, and the Nature of Language. Cambridge:Cambridge University Press.Astington, J. W. and Jenkins, J.M. (1999) A longitudinalstudy of the relation between language andtheory-of-mind development. Developmental Psychology,35/5: 1311-1320.Badcock, C. (2000) Evolutionary Psychology. A CriticalIntroduction. Cambridge: Polity Press.Bard K. A., Myowa-Yamakoshi, M., Tomonaga, M.,Tanaka, M., Costall, A. and Matsuzawa, T. (inpress) Group differences in the mutual gaze ofchimpanzees. Developmental Psychology Vol. 41,2005.Bard, K. A. and Vauclair, J. (1984) The communicativecontext of object manipulation in ape andhuman adult-infant pairs. Journal of Human Evolution,13: 181-90.Baron-Cohen, S. (1995) Mindblindedness: An Essay onAutism and Theory of Mind. Cambridge, MA: MITPress.Bates, E., Camaioni, L. and Volterra, V. (1975) Performativesprior to speech. Merrill-Palmer Quarterly,21: 205-226.Lund University Cognitive Studies - LUCS 121 ISSN 1101 – 8453 © 2005 by the authors

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 35Bering, J. M., Bjorklund, D. F. and Ragan, P. (2000)Deferred imitation of object-related actions inhuman-reared juvenile chimpanzees and orangutans.Developmental Psychobiology, 36: 218-32.Bickerton, D. (1990) Language and Species. Chicago:University of Chicago Press.- (2003) Symbol and structure: A comprehensivestructure for language evolution. In: Christiansenand Kirby (2003), 77-93.Bjorklund D. F., Bering, J. M. and Ragan, P. (2000) Atwo-year longitudinal study of deferred imitationof object manipulation in a juvenile chimpanzee(Pan troglodytes) and orangutan (Pongo pygmaeus).Developmental Psychobiology, 37: 229-37.Bjorklund, D. F. and Bering, J. M. (2003) A note onthe development of deferred imitation in enculturatedjuvenile chimpanzees (Pan troglodytes). DevelopmentalReview, 23: 389-412.Bjorklund, D. F., Yunger, J. L., Being, J. M. and Ragan,P. (2002) The generalization of deferred imitationin enculturated chimpanzees (Pan troglodytes).Animal Cognition, 5: 49-58.Bloom, P. (2000) How Children Learn the Meaning ofWords. Cambridge, MA: MIT Press.Bodamer, M. D., Fouts, R. S., Fouts, D. H., andJensvold, K. B. (1994) Indicating acts duringcounting by a chimpanzee (Pan troglodytes). Journalof Comparative Psychology, 109: 47-51.Boesch, C. (2003) Is culture a golden barrier betweenhuman and chimpanzee? Evolutionary Anthropology,12: 82-91.Boesch, C. and Tomasello, M. (1998) Chimpanzeeand human cultures. Current Anthropology, 39/5:591-614.Bonvillan, J. D. and Patterson, F. G. (1993) Early signlanguage acquisition in children and gorillas: Vocabularycontent and sign iconicity. First Language,13: 315-38.- (1999) Early sign-language acquisition: Comparisonbetween children and gorillas. In: The Mentalitiesof Gorillas and Orangutans: Comparative Perspectives,eds, S.T. Parker, R.W. Mitchell, H. L.Miles, 240-264. Cambridge University Press.Brinck, I. (2001) Attention and the evolution of intentionalcommunication. Pragmatics and Cognition,9/2: 255-272.- (2003) The pragmatics of imperative and declarativepointing. Cognitive Science Quarterly, 3(4): 429-446.- (2004) Joint attention, triangulation and radicalinterpretation: A problem and its solution, Dialectica,58(2): 179-206.Brinck, I. and Gärdenfors, P. (2003) Co-operationand communication in apes and humans. Mindand Language, 18(5): 484-501.Brown, R. (1977) Why are signed languages easier tolearn than spoken languages? In: Proceedings of theNational Symposium on Sign Language Research andTraining, ed. W. Stokoe, 9-24. Silver Spring, MD:National Association of the Deaf.Butterworth, G. (1988) What is special about pointing?In: The Development of Sensory, Motor, and CognitiveCapacities in Early Infancy: From Perception toCognition, ed. F. Simion and G. Butterworth 171-190. Hove: Psychology Press.Byrne, R. W. and Stokes, E. J. (2002) Effects of manualdisability on feeding skills in gorillas andchimpanzees. International Journal of Primatology,23/3: 539-54.Byrne, R. W. (1995) The Thinking Ape: EvolutionaryOrigins of Intelligence. Oxford: Oxford UniversityPress.- (1998) Commentary on Boesch, C. andTomasello, M. Chimpanzee and human culture.Current Anthropology, 39: 604-5.- (1999) Imitation without intentionality. Usingstring parsing to copy the organization of behaviour.Animal Cognition, 2: 63-72.Byrne, R. W. and Russon, A. E. (1998) Learning byimitation: A hierarchical approach. Behavioral andBrain Sciences, 21: 667-21.Call, J. and Tomasello, M. (1994) Production andcomprehension of referential pointing byorangutans (Pongo pygmaeus). Journal of ComparativePsychology, 108: 307-317.- (1996) The effect of humans in the cognitivedevelopment of apes. In: Reaching into Thought:The Minds of Great Apes. ed. A. Russon, K. Bardand S. Parker. Cambridge: Cambridge UniversityPress.- (1998) Distinguishing intentional from accidentalactions in orangutans (Pongo pygmaeus), chimpanzees(Pan troglodytes), and human children (Homosapiens). Journal of Comparative Psychology, 112/2:192-206.- (1999) A nonverbal false belief task: The performanceof children and great apes. Child Development,70/2: 381-395.Call, J., Hare, B., Carpenter, M. and Tomasello,M. (2004) “Unwilling” versus “unable”: chimpanzees’understanding of human intentional action.Developmental Science, 7/4: 488-98.Cheney, D. and Seyfarth, R. (1990) How Monkeys Seethe World: Inside the Mind of Another Species. Chicago:University of Chicago Press.Christiansen, M. and Kirby, S. (2003) (eds.) LanguageEvolution. Oxford: Oxford University Press.Clark, H. (1996) Using Language. Cambridge: CambridgeUniversity Press.Corballis, M. C. (2002) From Hand to Mouth: The Originsof Language. Princeton, NJ: Princeton UniversityPress.Coussi-Kourbel, S. (1994) Learning to outwit a competitorin mangabeys (Cercocebus torquatus torquatus).Journal of Comparative Psychology, 108: 164-171.Custance, D. M., Whiten, A. and Bard, K. A. (1995)Can young chimpanzees (Pan troglodytes) imitatearbitrary actions? Hayes and Hayes (1952)revisited. Behaviour, 132/11-12: 837-59.

36Bodily mimesis as “the missing link” in human cognitive evolutionCustance, D. M., Whiten, A., Sambrook, T. andGaldikas, B. (2001) Testing for social learning inthe “artificial fruit” processing of wild bornorangutans (Pongo pygmaeus), Tanjung Puting, Indonesia.Animal Cognition, 4: 305-13.Custance, D. M., Whiten, A. and Fredman, T. (1999)Social learning of an artificial fruit task in capuchinmonkeys (Cebus apella). Journal of ComparativePsychology, 113/1: 13-23.de Waal, F. (1988) The communicative repertoire ofcaptive bonobos (Pan paniscus), compared to thatof chimpanzees. Behaviour, 106/3-4: 183-251.- (1996) Good Natured: The Origins of Right and Wrongin Humans and Other Animals. Harvard: HarvardUniversity Press.- (1998 [1982]) Chimpanzee Politics. Harper.de Waal, F. and Aureli, F. (1996) Consolation, Reconciliationand a Possible Cognitive Difference Between Macaquesand Chimpanzees. Cambridge: CambridgeUniversity Press.de Villiers, J., and Pyers, J. (1997) Complementingcognition: The relationship between languageand theory of mind. In Proceedings of the 21 st AnnualBoston University Conference on Language Development.Somerville, MA: Cascadillia Press.Deacon, T. (1997) The Symbolic Species: The Co-Evolution of Language and the Brain. New York:Norton.- (2003) Universal grammar and semiotic constraints.In Christiansen and Kirby (2003), 111-139.Dennett, D. (1996) Kinds of Minds. Towards an Understandingof Consciousness. London: Phoenix.Dessalles, J.-L. (2004) Language as an isolated niche.Presented at The Fifth Conference on the Evolution ofLanguage, Leipzig.Donald, M. (1991) Origins of the Modern Mind. ThreeStages in the Evolution of Culture and Cognition. Harvard:Harvard University Press.- (1998) Mimesis and the Executive Suite: Missinglinks in language evolution. In Approaches to theEvolution of Language, eds., J. Hurford, M. Studdert-Kennedyand C. Knight, 44-67. Cambridge:Cambridge University Press.- (2001) A Mind So Rare. The Evolution of HumanConsciousness. New York: Norton.Edelman, G. (1992) Bright Air, Brilliant Fire: On theMatter of the Mind. New York: Basic Books.Eldridge, N. and Gould. S. J. (1972) Punctuatedequilibria: An alternative to phyletic gradualism.In: Models in Paleobiology, ed., T. Schopf. SanFrancisco: Freeman.Fouts, R. S. (1972) Use of guidance in teaching signlanguage to a chimpanzee (Pan troglodytes). Journalof Comparative and Physiological Psychology, 80/3:515-22.- (1973) Acquisition and testing of gestural signs infour young chimpanzees. Science, 180: 978-980.Gallagher, S. (1995) Body schema and intentionality.In: The Body and the Self, ed. J. Bermúdez, N.Eilan, and A. Marcel. Cambridge, MA:MIT/Bradford Press.- (2005) How the Body Shapes the Mind. Oxford:Oxford University Press.- (in press) Phenomenological and experimentalcontributions to understanding embodied experience.In Ziemke, Zlatev and Frank (in press).Gallese, V., C. Keyners and Rizzolatti, G. (2004) Aunifying view of the basis of social cognition.Trends in Cognitive Sciences, 8/9: 396 – 403.Gallup, G. (1982) Self-awareness and the emergenceof mind in primates. American Journal of Primatology,2: 237-248.Gärdenfors, P. (2003) How Homo Became Sapiens: Onthe Evolution of Thinking. Oxford: Oxford UniversityPress.Gardner, R. A. and Gardner B.T. (1969) Teachingsign language to a chimpanzee. Science, 165: 664-672.Givón, T. (2002) Bio-Linguistics. The Santa BarbaraLectures. Amsterdan: Benjamins.Goldin-Meadow, S. (1999) The role of gesture incommunication and thinking. Trends in CognitiveSciences, 3/11: 419-429.Gómez, J. C. (1994) Mutual awareness in primatecommunication: a Gricean approach, In: Selfawarenessin Animals and Humans, ed. S. T. Parker,R. W. Mitchell and M. L. Boccia. Cambridge:Cambridge University Press.Gopnik, A. and Astington, J. W. (1988) Children’sunderstanding of representational change, and itsrelation to the understanding of false belief andthe appearance-reality distinction, Child Development,59: 26-37.Gopnik, A. and Graf, P. (1988) Knowing how youknow: young children’s ability to identify andremember the sources of their beliefs. Child Development,59: 1366-1371.Gopnik, A. and Meltzoff, A. N. (1997) Words,Thoughts, and Theories, Cambridge, MA: MITPress.Greenfield, P. M. and Savage-Rumbaugh. E. S. (1990)Grammatical combination in Pan pamiscus:Processes of learning and invention in the evolutionand development of language, In: “Language”and Intelligence in Monkeys and Apes, eds. S T.Parker and K. R. Gibson, 540-578. Cambridge:Cambridge University Press.Greenfield, P. M. and Savage-Rumbaugh, E. S. (1991)Imitation, grammatical development, and the inventionof protogrammar. In N. Krasnegor, D.Rumbaugh, M. Studdert-Kennedy, and R.Schiefelbusch (Eds.), Biological and behaviorial determinantsof language development (pp. 235-258).Hillsdale, NJ: Erlbaum.Greenfield, P. M., and Savage-Rumbaugh, E. S.(1993) Comparing communicative competence

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 37in child and chimp: The pragmatics of repetition.Journal of Child Language, 20, 1-26.Grice, P. (1957). Meaning. Philosophical Review, 66:377-88.Hale, C. M. and Tager-Flusberg, H. (2003) The influenceof language on theory of mind: A trainingstudy. Developmental Science 6/3: 346-359.Hare, B., Call, J., Agnetta, B. and Tomasello, M.(2000) Chimpanzees know what conspecifics doand do not see. Animal Behaviour, 59: 771-85.Hare, B., Call, J. and Tomasello, M. (2001) Do chimpanzeesknow what conspecifics know? AnimalBehaviour, 61: 139-51.Hauser, M. (1996) The Evolution of Communication.Cambridge, MA: MIT Press.Hauser, M., Chomsky, N. and Fitch, T. (2002) Thefaculty of language: What is it, who has it andhow did it evolve? Science, 298: 1559-1569.Heyes, C. (1998) Theory of mind in nonhuman primates.Behavioral and Brain Sciences, 21: 101-115.Hockett, C. F. (1960) The origin of speech. ScientificAmerican, 23: 89-96.Itkonen, E. (1978) Grammatical Theory and Metascience.Amsterdam: Benjamins.Ikegami, T. and Zlatev, J. (in press) From prerepresentationalcognition to language. InZiemke, Zlatev and Frank (in press).James, W. (1890) The Principles of Psychology. New York:Henry Holt.Johansson, S. (2005) Origins of Language. Constraints onHypotheses. Amsterdam: Benjamins.Kendon, A. (1981) Geography of gesture. Semiotica,37: 129-163.Kita, S. and Özyürek, A. (2003) What does crosslinguisticvariation in semantic coordination ofspeech and gesture reveal? Evidence for an interfacerepresentation of spatial thinking and speaking.Journal of Memory and Language, 48: 16-32.Knight, C. (2002) Language and revolutionary consciousness.In: The Transition to Language, ed. A.Wray, 138-160. Oxford: Oxford UniversityPress.Krause, M and Fouts, R. (1997) Chimpanzee (Pantroglodytes) pointing: Hand shapes, accuracy andthe role of gaze. Journal of Comparative Psychology,111 (4): 330-336.Laakso, A. (1993) Language equals mimesis plusspeech. Commentary to Donald: Origins of Mind.Behavioral and Brain Sciences, 14/4: 765-766.Landau, M. (1991) Narratives of Human Evolution. NewHaven: Yale University Press.Lewis, D. K. (1969) Convention: A Philosophical Study.Cambridge MA: Harvard University Press.Leavens, D. A., Hopkins, W. D. and Bard, K. (1996)Indexical and referential pointing in chimpanzees(Pan troglodytes). Journal of Comparative Psychology,110: 346-353.Leavens, D. A. and Hopkins, W. D. (1999) Thewhole-hand point: The structure and function ofpointing from a comparative perspective. Journalof Comparative Psychology, 113/4: 417-425.Lohmann, H. and Tomasello, M. (2003) The role oflanguage in the development of false belief understanding:A training study. Child Development74/4: 1130-1144.Liberman, A. M. and Mattingly, I. G. (1985) Themotor theory of speech perception revised. Cognition,21: 1-36.Mandler, J. (2004) The Foundations of Mind: Origins ofConceptual Thought. Oxford: Oxford UniversityPress.Marler, P. (1985) Representational vocal signals ofprimates. In: Experimental Behavioral Ecology andSociobiology, ed. B. Hölldobler and M. Lindauer.Stuttgart: Fischer Verlag.- (1991) Difference in behavioral development inclosely related species: Birdsong. In: The Developmentand Integration of Behavior, ed. P. Bateson, 41-70. Cambridge: Cambridge University Press.Matsuzawa, T., Biro, D., Humle, T. Inoue-Nakamura,N., Tanook, R. and Yamakoshi G. (2001) Emergenceof culture in wild chimpanzees: Educationby master-apprenticeship. In: Primate Origins ofHuman Cognition and Behavior, ed. T. Matzuzawa.Springer-Verlag.McNeill, D. (1992) Hand and Mind: What GesturesReveal about Thought. Chicago: University of ChicagoPress.Meltzoff, A. and Gopnik, A. (1993) The role of imitationin understanding other persons and developinga theory of mind. In: Understanding otherMinds: Perspectives from Autism, ed. S. Baron-Cohen, H. Tager-Flusberg, and D. J. Cohen. Oxford:Oxford University Press.Meltzoff, A. and Moore, M. K. (1977) Imitation offacial and manual gestures by human neonates.Science 198: 75-78.- (1983) Newborn infants imitate adult facial gestures.Child Development, 54: 702-709.- (1994) Imitation, memory, and the representationof persons. Infant Behavior and Development17: 83-99.Menzel, E. W. (1973) Chimpanzee spatial memoryorganization, Science 182: 943-945.- (1974) A group of chimpanzees in a 1-acre field:leadership and communication, In: Behavior ofNonhuman Primates, ed. A. M. Schrier and F.Stollnitz. Academic.Merleau-Ponty, M. (1962 [1945]) Phenomenology ofPerception. Routledge and Kegan Paul.- (2003) La Nature: Notes, Cours du College de France.Trans. Robert Vallier, Northwestern UniversityPress.Miles, H. L. (1990) The cognitive foundations forreference in a signing orangutan. In: “Language”and Intelligence in Monkeys and Apes, eds. S T.

38Bodily mimesis as “the missing link” in human cognitive evolutionParker and K. R. Gibson. Cambridge UniversityPress.- (1999) Symbolic communication with and bygreat apes. In: The Mentalities of Gorillas andOrangutans, ed. S. Taylor Parker, R. W. Mitchelland H. L. Miles. Cambridge University Press.Miles, H. L., Mitchell, R. W. and Harper, S. E. (1996)Simon says: The development of imitation in anenculturated orangutan. In: Reaching into Thought:The Minds of the Great Apes, ed. A. E. Russon, K.A. Bard. and S. T. Parker. Cambridge UniversityPress.Mitchell, P. (1997) Introduction to Theory of Mind: Children,Autism and Apes. Arnold.Mitchell, R. and Miles, H. L. (1993) Apes have mimeticculture. Commentary to Donald: Origins ofMind. Behavioral and Brain Sciences 14/4: 768.Myowa-Yamakoshi, M. (2001) Evolutionary foundationand development of imitation. In: PrimateOrigins of Human Cognition and Behavior, ed. T.Matsuzawa. Springer-Verlag.Myowa-Yamakoshi, M. and Matsuzawa, T. (1999)Factors influencing imitation of manipulatoryactions in chimpanzees (Pan troglodytes). Journalof Comparative Psychology, 113/2: 128-36.Myowa-Yamakoshi, M., Tomonaga, M., Tanaka, M.and Matsuzawa, T. (2004) Imitation in neonatalchimpanzees (Pan troglodytes). Developmental Science,7/4: 437-42.Nagell, K., Olguin, R. S. and Tomasello, M. (1993)Processes of social learning in the tool use ofchimpanzees (Pan troglodytes) and human children(Homo sapiens). Journal of Comparative Psychology,107/2: 174-86.Nelson, K. (1996) Language in Cognitive Development. TheEmergence of the Mediated Mind. Cambridge UniversityPress.Nelson, K. and Shaw, L. K. (2002) Developing asocially shared symbolic system. In: Language,Literacy and Cognitive Development, ed. J. Byrnesand E. Amseli, Lawrence Erlbaum.Osvath, M. and Gärdenfors, P. (2005) Oldowanculture and the evolution of anticipatory cognition.Lund University Cognitive Studies 122, Lund.Patterson, F. (1978) The gestures of a gorilla: Languageacquisition in another pongid. Brain andLanguage, 5: 72-97.- (1980) Innovative use of language in a gorilla: Acase study. In: Children’s Language, Vol 2, ed., K.Nelson 497-561. New York: Garnder Press.Perner, J. (1991) Understanding the Representational Mind.Cambridge, MA: MIT Press.Perner, J., Leekam, S. and Wimmer, H. (1987) Threeyear-old’sdifficulty with false belief: the case fora conceptual deficit, British Journal of DevelopmentalPsychology, 5: 125-137.Peterson, C. C. and Siegal, M. (1995) Deafness, conversationand the theory of mind. Journal of ChildPsychology and Psychiatry and Allied Disciplines, 36:459-474.Piaget, J. (1945) La Formation du Symbole Chez l'enfant.Delachaux et Niestlé.- (1953) The Origin of Intelligence in the Child. Routledge.- (1954) The Construction of Reality in the Child. BasicBooks.Pierce, C. S. (1931-35) The Collected Papers of CharlesSanders Peirce. Vols. 1-4. Cambridge, Mass.: HarvardUniversity Press.Pika, S., Liebal, K. and Tomasello, M. (2003) Gesturalcommunication in young gorillas (Gorilla gorilla):Gestural repertoir, learning, and use. AmericanJournal of Primatology, 60: 95-111.- (2005) Gestural communication in subadultbonobos (Pan paniscus): Repetoire and use. AmericanJournal of Primatology, 65: 39-61.Pinker, S. (1994) The Language Instinct. William Morrow.Povinelli, D. J. (2000) Folk Physics for Apes: The Chimpanzee’sTheory of How the World Works. OxfordUniversity Press.Povinelli, D. J. and Davis, D. R. (1994) Differencesbetween chimpanzees (Pan troglodytes) and humans(Homo sapiens) in the resting state of the indexfinger: Implications for pointing. Journal ofComparative Psychology, 108: 134-139.Premack, D. (1988) “Does the chimpanzee have atheory of mind?” revisited. In Machiavellian Intelligence.Social Expertise and the Evolution of Intellect inMonkeys, Apes and Humans, ed. R. W. Byrne andA. Whiten, 160-179. Oxford: Clarendon Press.Preston, S. and de Waal, F. (2002) Empathy: Its ultimateand proximal bases, Behavioral and Brain Sciences,25: 1-72.Reddy, V. (2001) Teasing, joking and mucking aboutin the first year. In: Natural Theories of Mind, ed.A. Whiten. Blackwell.- (2003) On being an object of attention: Implicationsof self-other-consciousness. Trends in CognitiveScience, 7/9: 397-402.- (2005) Before the “third element”: Understandingattention to self. In Naomi Klein (ed.) Joint Attention,85-109. Oxford: Oxford University Press.Rizzolatti, G., Fadiga, L., Gallese, V., and Fogassi, L.(1996) Premotor cortex and the recognition ofmotor actions. Cognitive Brain Research, 3, 131-141.Rizzolatti, G. and Arbib, M. (1998) Language withinour grasp. Trends in Neurosciences, 21: 188-194.Robinson, P. (1996) Consciousness, Rules and InstructedSecond Language Acquisition. Peter Lang.Russon, A. E, and Galdikas, B. M. F. (1993) Imitationin free-ranging rehabilitant orangutans (Pongopygmaeus). Journal of Comparative Psychology, 107/2 :147-61.Saussure, F. de (1916) Cours de Linguistique Générale[Course in General Linguistics]. Payot.

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 39Savage-Rumbaugh, S. E., Wilkerson, B. J. and Bakeman,R. (1977) Spontaneous gestural communicationamong conspecifics in the pygmy chimpanzee(Pan paniscus). In: Progress in Ape Research,ed. G. H. Bourne. Oxford: Academic.Savage-Rumbaugh, S. and Lewin. R. (1994) Kanzi: TheApe at the Brink of the Human Mind. New York:John Wiley.Savage-Rumbaugh, S., Shanker, S., and Taylor, T.(1998). Apes, Language and the Human Mind. Oxford:Oxford University Press.Searle, J. (1992) The Rediscovery of the Mind. MIT Press.Senghas, A., Kita, S. and Özyürek, A. (2004) Childrencreating core properties of language: Evidencefrom an emerging sign language in Nicaragua.Science 305: 1779-1782.Sinha, C. (1988) Language and Representation. A SocionaturalisticApproach to Human Development. HarversterPress.- (2003) The evolution of language: From signalsto symbols to system. In: Evolution of Communicationsystems: A Comparative Approach, ed. K. Ollerand U. Griebel. MIT Press.Sonesson, G. (1989) Pictorial Concepts. Lund UniversityPress.- (2001) From semiosis to ecology. On the theoryof iconicity and its consequences for the ontologyof the Lifeworld. VISIO 6/2-3: 85-110.- (in press) The “Symbolic species” revisited. Considerationson the semiotic turn in cognitive scienceand biology.Sperber, D. and Wilson, D. (1995) Relevance: Communicationand Cognition. Cambridge University Press.Stern, D. (2000) The Interpersonal World of the Infant: AView from Psychoanalysis and Developmental Psychology.Basic Books.Stoinski, T. S. and Whiten, A. (2003) Social learningby orangutans (Pongo abelii and Pongo pygmaeus).Journal of Comparative Psychology, 117/3: 272-82.Stoinski, T. S., Wrate, J. L., Ure, N. and Whiten,A. (2001) Imitative learning by captive westernlowland gorillas (Gorilla gorilla gorilla) in a simulatedfood-processing task. Journal of ComparativePsychology, 115/3: 272-81.Stokes, E. J. and Byrne, R. W. (2001) Cognitive capacitiesfor behavioural flexibility in wild chimpanzees(Pan troglodytes): the effect of snare injuryon complex manual food processing. AnimalCognition, 4: 11-28.Stokoe, W. (1960) Sign language structure: An outlineof the visual communication system of theAmerican deaf. Studies in Linguistics, Occasional Paper8, University of Buffalo, New York.- (2001). Language in Hand. Why Sign Came beforeSpeech. Washington, D.C.: Gallaudet UniversityPress.Tanner, J. E. and Byrne, R. W. (1996) Representationof action through iconic gesture in a captive lowlandgorilla. Current Anthropology, 37/1: 162-73.- (1999) The development of spontaneous gesturalcommunication in a group of zoo-living lowlandgorillas. In: The Mentalities of Gorillas and Orangutans– Comparative perspectives, ed. S. T. Parker, R.W. Mitchell and H. Lyn Miles. Cambridge UniversityPress.Tanner, J. E. (2004) Gestural phrases and gesturalexchanges by a pair of zoo-living lowland gorillas.Gesture, 4/1: 1-24.Terrace, H. S., Petitto, L. A., Sanders, R. J. and Bever,T. G. (1981) Science 211/4477: 87-88.Thompson, R. K. R., Boysen, S. T. and Oden, D. L.(1997) Language-naive chimpanzees (Pan troglodytes)judge relations between relations in a conceptualmatching-to-sample task. Journal of ExperimentalPsychology, 23/1: 31-43.Tomasello, M. (1992) First verbs: A Case Study of EarlyGrammatical Development. Cambridge UniversityPress.- (1993) It’s imitation, not mimesis. Commentaryto Donald: Origins of Mind. Behavioral and BrainSciences 14/4: 771-772.- (1996) Do apes ape? In: Social Learning in Animals:The Roots of Culture, ed. C. Heyes and B.Galef. Oxford: Academic.- (1999) The Cultural Origins of Human Cognition.Cambridge, MA: Harvard University Press.- (2003) Constructing a Language: A Usage-based Theoryof Language Acquisition. Cambridge, MA: HarvardUniversity Press.Tomasello, M. and Call, J. (1997): Primate Cognition.Oxford University Press.Tomasello, M., Call, J. and Gluckman, A. (1997)Comprehension of novel communicative signsby apes and human children. Child Development,68: 1067-1080.Tomasello, M., Call, J. and Hare, B. (2003) Chimpanzeesunderstand psychological states – the questionis which ones and to what extent. Trends inCognitive Sciences, 7/4: 153-156.Tomasello, M., Call, J., Nagell, K., Olguin, R.and Carpenter, M. (1994) The learning and useof gestural signals by young chimpanzees: Atrans-generational study. Primates, 35/2: 137-154.Tomasello, M., Carpenter, M., Call, J., Behne, T., andMoll, H. (in press) Understanding and sharing intentions:The origins of cultural cognition. Behavioraland Brain Sciences.Tomasello, M., Kruger, A. and Ratner, H. (1993)Cultural learning. Behavioral and Brain Sciences, 16,495-552.Tomasello M., Savage-Rumbaugh S. and Kruger,A. (1993) Imitative learning of actions on objectsby children, chimpanzees, and enculturatedchimpanzees. Child Development, 64: 1688-1795.Tomasello, M. and Zuberbühler, K. (2002) Primatevocal and gestural communication. In: The CognitiveAnimal: Empirical and Theoretical Perspectives onAnimal Cognition, ed. M. Bekoff, C. Allen and M.Burhgardt. MIT Press.

40Bodily mimesis as “the missing link” in human cognitive evolutionTrevarthen, C. (1992) An infant’s motives for speakingand thinking in the culture, In: The DialogicalAlternative, ed. H. Wold. Oslo: Scandinavian UniversityPress.Varela, F., Thompson, E. and Rosch, E. (1991) TheEmbodied Mind. Cognitive Science and Human Experience.Cambridge, MA: MIT Press.Veà, J. J. and Sabater-Pi, J. (1988) Spontaneous pointingbehaviour in the wild pygny chimpanzee (Panpaniscus). Folia Primatologica, 69: 289-290.von Hofsten, C. (1983). Catching skills in infancy.Journal of Experimental Psychology: Human Perceptionand Preference, 9: 75-85.Whiten, A. (1998) Imitation of the sequential structureof actions by chimpanzees (Pan troglodytes).Journal of Comparative Psychology, 112: 270-81.Whiten, A. and Byrne, R. W. (1988) Tactical deceptionin primates, Behavioral and Brain Sciences, 11:233-244.Whiten, A., Custance, D. M., Gomez, J.-C., Teixidor,P. and Bard, K. A. (1996) Imitative learning ofartificial fruit processing in children (Homosapiens) and chimpanzees (Pan troglodytes). Journal ofComparative Psychology, 110/1: 3-14.Whiten, A., Horner, V., Litchfield, C. A. and Marshall-Pescini,S. (2004) How do apes ape? Learningand Behaviour, 32/1: 36-52.Wilcox, S. (2004) Gesture and language: Crosslinguisticand historical data from signed languages.Gesture, 4/1: 43-73.Wimmer, H., Hogrefe, G.-J. and Perner, J. (1988)Children’s understanding of informational accessas a source of knowledge. Child Development, 59:386-396.Wittgenstein, L. (1953) Philosophical Investigations. BasilBlackwell.Woll, B. and Kyle, J. (2004) Sign language. Encylopediaof Language and Linguistics. Oxford: Elsevier.Woodruff, G. and Premack, D. (1979) Intentionalcommunication in the chimpanzee: the developmentof deception. Cognition, 7: 333-362.Vygotsky, L. S. (1962) Thought and Language. CambridgeMA: MIT Press.- (1978) Mind in Society. The Development of HigherPsychological Processes. Harvard University Press.Zlatev, J. (2002) Mimesis: The “missing link” betweensignals and symbols in phylogeny and ontogeny.In: Mimesis, Sign and the Evolution of Language, ed.A. Pajunen. University of Turku Press.- (2003) Meaning = Life (+ Culture). An outline ofa unified biocultural theory of meaning. Evolutionof Communication, 4/2: 253-296.- (2005a) What’s in a schema? Bodily mimesis andthe grounding of language. In: From Perception toMeaning: Image Schemas in Cognitive Linguistics, ed.B. Hampe and J. Grady. Mouton.- (2005b) The dependence of language on consciousness.Paper presented at the conferenceTowards a Science of Consciousness 2005, August 17-20, Copenhagen.Zlatev, J., Persson, T. and Gärdenfors, P. (in press)Triadic bodily mimesis is the difference! Commentaryto Tomasello et al. (in press).Zuberbühler, K. (2000) Interspecies semantic communicationin two forest primates. Proceedings ofthe Royal Society of London. Series B Biological Sciences267: 713-718.

Lund University Cognitive Studies ISSN 1101-84531991 1. Torbjörn Molin: Tractatus de Ratione Circumscripta -Rationality for Finite Agents, LUHFDA/HFKO-3001-SE2. Agneta Gulz: The Planning of Action as a Cognitive andBiological Phenomenon, LUHFDA/HFKO-1001-SE3. Christian Balkenius, Peter Gärdenfors: NonmonotonicInferences in Neural Networks, LUHFDA/HFKO-5001-SE,Published in Principles of Knowledge Repres-entation andReasoning: Proceedings of the Second International Conference,J.A. Allen, R. Fikes, and E. Sandewall, eds.(MorganKaufmann: San Mateo, CA), pp. 32-39, 1991. Revisedversion.4. Tom Andersson: Metaphorical Indices-Presenting Context inDiscourse through Metaphorical Expressions,LUHFDA/HFKO-5002-SE. Out of print.5. Peter Gärdenfors: The Emergence of Meaning,LUHFDA/HFKO-5003-SE, Published in Linguistics andPhilosophy, 16: 285-309, 1993.1992 6. Tom Andersson: Views on Nature in Metaphorical Discourse,LUHFDA/HFKO-5004-SE Published in Views of Nature,(FRN, Solna).7. Agneta Gulz: Conceptions of Anger and Grief in the Japanese,Swedish and American Cultures: The Role of Metaphor inConceptual Processes, LUHFDA/HFKO-5005-SE8. Tom Andersson: The Dynamics of Metaphor - ConflictingImages of Forest in Swedish Discourse, LUHFDA/HFKO-3002-SE9. Peter Gärdenfors: Three Levels of Inductive Inference,LUHFDA/HFKO-5006-SE, Published in Logic, Methodology,and Philosophy of Science IX, D. Prawitz, B. Skyrms and D.Westerståhl, eds., (Elsevier Science, Amsterdam) pp. 427-449,1994.10. Paul Hemeren: Frequency Norms for Actions in Swedish andAmerican Samples, LUHFDA/HFKO-5007-SE, Published as"Frequency, ordinal position, and semantic distance asmeasures of cross-cultural stability and hierarchies for actionverbs" in Acta Psychologica, 91, 1996.11. Peter Gärdenfors, Hans Rott: Belief Revision,LUHFDA/HFKO-5008-SE, Published in Handbook of Logicin Artificial Intelligence and Logic Programming, Vol. 4, D.M. Gabbay, C. J. Hogger and J. A. Robinson, eds., (OxfordUniversity Press, Oxford), 1995.12. Paul Davidsson: Concept Acquisition by Autonomous Agents:Cognitive Modeling versus the Engineering Approach,LUHFDA/HFKO-5009-SE13. Robert Pallbo: Neuronal Selectivity Without IntermediateCells, LUHFDA/HFKO-5010-SE14. Christian Balkenius: Neural Mechanisms for Self-organizationof Emergent Schemata, Dynamical Schema Processing, andSemantic Constraint Satisfaction, LUHFDA/HFKO-5011-SE15. Peter Gärdenfors: How Logic Emerges from the Dynamics ofInformation, LUHFDA/HFKO-5012-SE. Published in Logicand Information Flow, J. van Eijck and A.Visser, eds. (MITPress: Cambridge, MA), pp. 49-77, 1994.1993 16. Guy de Lucs: There are no Meaningless Sentences,LUHFDA/HFKO-7001-SE17. Kenneth Holmqvist: Implementing Cognitive Semantics:Image Schemata, Valence Accommodation and ValenceSuggestions for AI and Computational Linguistics,LUHFDA/HFKO-1002-SE18. Peter Gärdenfors, Christian Balkenius: Varför finns det ingariktiga robotar? LUHFDA/HFKO-7002-SE. Pub-lished inFramtider, Vol. 12, 1/93, pp. 11-14, (Institutet förFramtidsstudier, Stockholm).19. Tom Andersson: Real Forests in Our Minds: Metaphors andPrototypes in Discourse, LUHFDA/HFKO-5013-SE20. Tom Andersson: Metaphorical Reasoning and RealPrototypes: Empirical Studies of Metaphors and Prototypes ofForest, LUHFDA/HFKO-5014-SE21. Peter Gärdenfors: The Role of Expectations in Reasoning,LUHFDA/HFKO-5015-SE, Published in Knowledge;Representation and Reasoning Under Uncertainity, M.Masuch and L. Pólos, eds, (Springer-Verlag, Berlin), pp.1-16.22. Henrik Gedenryd: Active Perception and the "ResonanceMetaphor, LUHFDA/HFKO-5016-SE23. Henrik Gedenryd: Managing Complexity by SupportingKnowledge Growth in Design and Development Projects,LUHFDA/HFKO-5017-SE24. Robert Pallbo: Ontogenesis in Neural Networks,LUHFDA/HFKO-5018-SE1994 25. Peter Gärdenfors: Kategorisk perception i vetenskapen,LUHFDA/HFKO-7003-SE, Published in Tidskrift förVetenskapsstudier, 7:2, pp. 17-29.26. Peter Gärdenfors, Kenneth Holmqvist: Concept Formation inDimensional Spaces, LUHFDA/HFKO-5019-SE27. Tom Andersson: Conceptual Polemics: Dialectic Studies ofConcept Formation, LUHFDA/HFKO-1003-SE28. Kenneth Holmqvist: Conceptual Engineering I: Frommorphemes to valence relations, LUHFDA/HFKO-5020-SE29. Christian Balkenius: Natural Intelligence for AutonomousAgents, LUHFDA/HFKO-5021-SE30. Christian Balkenius: Biological Learning and ArtificialIntelligence, LUHFDA/HFKO-5022-SE31. Sofia Broström: The Role of Metaphor in CognitiveSemantics, LUHFDA/HFKO-3003-SE32. Lars Kopp: A Neural Network for Spatial Relations:Connecting Visual Scenes to Linguistic Descriptions,LUHFDA/HFKO-5023-SE33. Simon Winter: Förväntningar och kognitionsforskning,LUHFDA/HFKO-7004-SE34. Robert Pallbo: Mind as Evolution and Evolution as Such,LUHFDA/HFKO-5024-SE35. Simon Winter, Peter Gärdenfors: Linguistic Modality asExpressions of Social Power, LUHFDA/HFKO-5025-SE,Published in Nordic Journal of Linguistics, 18, 1995, pp. 137-166.1995 36. Erik Olsson: Can Information Theory Help us RemoveOutdated Beliefs? LUHFDA/HFKO-5026-SE37. Christian Balkenius: Natural Intelligence in ArtificialCreatures, LUHFDA/HFKO-1004-SE38. Peter Gärdenfors: Cued and Detached Representations inAnimal Cognition, LUHFDA/HFKO-5027-SE, Published inBehavioural Processes 36, 1996, pp. 263-273.39. Kenneth Holmqvist, Jaroslaw Pluciennik: Concept-ualisedDeviations from Expected Normalities: A SemanticComparison Between Lexical Items Ending in -ful and -less,LUHFDA/HFKO-5028-SE, Published in Nordic Journal ofLinguistics, 19, 1996, pp. 3-3340. Peter Gärdenfors: Meanings as Conceptual Structures,LUHFDA/HFKO-5029-SE, Published in Mindscapes:Philosophy, Science and the Mind, ed. by M. Carrier and P.Machamer, Pittsburgh University Press, Pittsburgh, 1997, pp.61-86.41. Peter Gärdenfors: Language and the Evolution of Cognition,LUHFDA/HFKO-5030-SE, Published in Penser l'esprit: Dessciences de la cognition à une philosophie cognitive, ed. by V.Rialle and D. Fisette, Presses Universitaires de Grenoble,Grenoble, 1996, pp. 151-172.42. David de Léon: The Limits of Thought and the Mind-BodyProblem, LUHFDA/HFKO-7005-SE43. Peter Gärdenfors: Human Communication: What Happens?LUHFDA/HFKO-7006-SE. in The Contrib-ution of Scienceand Technology to the Development of Human Society, ed byB. Reichert, ECSC-EC-EAEC, Brussels, abstract on p. 96

1996 44. Simon Winter: Anticipation and Violin Strings,LUHFDA/HFKO-5031-SE45. Mikael Johannesson: Obtaining Psychologically MotivatedSpaces with MDS, LUHFDA/HFKO-5032-SE46. Simon Winter: Det självklara behöver inte sägas,LUHFDA/HFKO-7007-SE47. Christian Balkenius, Lars Kopp: The XT-1 VisionArchitecture, LUHFDA/HFKO-5033-SE, Published inProceedings of the Symposium on Image Analysis, ed. by P.Linde and G. Sparr, Lund University, Lund, 1996, pp. 39-43.48. Simon Winter: Vad är pragmatik, egentligen? LUHFDA/HFKO-7008-SE49. Kenneth Holmqvist, Jana Holsánová: "Jag längtar efter svenskkvalité och tänkande...": Hur vi använder kategoriseringar isamtalet, LUHFDA/HFKO-5034-SE50. Kenneth Holmqvist, Jana Holsánová: Focus Movements andthe Internal Images of Spoken Discourse, LUHFDA/HFKO-5035-SE, Published as "Focus Movements duringVisualisation of Spoken Discourse," in DiscourseConstructions, ed. W. Liebert, G. Redeker and L. Waugh,Amsterdam/ Philadelphia: John Benjamins, 1997.51. Peter Gärdenfors: När tanken lyfter, LUHFDA/HFKO-7009-SE Published in Kungliga Vitterhets- Historie- ochAntikvitets-akademiens Årsbok 1996, Stockholm, pp. 106-117.1997 52. Christian Balkenius, Simon Winter: Explorations in SyntheticPragmatics, LUHFDA/HFKO-5036-SE, Published in “DoesRepresentation Need Reality?”, Proceedings from New Trendsin Cognitive Science -97, ed. A. Riegler and M. Peschl,Vienna, ASoCS Technical Report 97-01, Austrian Society ofCognitive Science, pp. 100-107.53. Peter Gärdenfors: Att tala med maskinerna, LUHFDA/HFKO-7010-SE, Abridged version publ-ished as "De kulturellakonsekvenserna av att tala med sin dator" in Moderna Tider,April 1997, pp. 40-42.54. Annika Wallin: Shaky Constructions, LUHFDA/ HFKO-5037-SE55. Mikael Johannesson: Modelling Asymmetric Similarity withProminence, LUHFDA/HFKO-5038-SE56. Christer Garbis: Spatial Representationand Haptic MentalRotation, LUHFDA/HFKO-5039-SE57. Robert Pallbo: Mind in Motion: The utilization of noise in thecognitive process, LUHFDA/HFKO-1005-SE58. David de Léon: The Qualities of Qualia, LUHFDA/ HFKO-5040-SE Communication and Cognition, 34 (1/2), pp. 121-138.59. Peter Gärdenfors: Media och berättande, LUHFDA/ HFKO-5041-SE pp. 315-336 published in Boken i tiden, SOU1997:141, Stockholm.60. David de Léon, Jana Holsánová: Revealing User Behaviour onthe World-Wide Web, LUHFDA/HFKO-5042-SE61. Jan Morén: Dynamic Action Sequences in ReinforcementLearning, LUHFDA/HFKO-5043-SE62. Christian Balkenius, Jan Morén: Computational Models ofClassical Conditioning: A Comparative Study LUHFDA/HFKO-5044-SE63. Ingar Brinck, Peter Gärdenfors: Representation and selfawarenessin intentional agents, LUHFDA/HFKO-5045-SE inSynthese 118, pp. 89-104.1998 64. Jens Månsson: Stereovision. A Model of Human Stereopsis,LUHFDA/HFKO-5046-SE65. Simon Winter: Dialogue Dynamics, Violin Strings, and thePragmatics-Semantics Continuum, LUHFDA/ HFKO-5047-SE66. Simon Winter: Evolution, Categorization, and ValuesLUHFDA/HFKO-5048-SE published in Evolution &Cognition, vol 4 no 2, 1998, Vienna University Press, Vienna.67. Annika Wallin: Choosing Problems, LUHFDA/HFKO-5049-SE68. Christian Balkenius: A Neural Network Model of ClassicalConditioning Part I: The Dynamics of Learning,LUHFDA/HFKO-5050-SE69. Simon Winter, Peter Gärdenfors: Evolving Social Constraintson Individual Conceptual Representations, LUHFDA/HFKO-5051-SE70. Kenneth Holmquist, Marianne Gullberg: Focus on Gesture,LUHFDA/HFKO-5052-SE71. Simon Winter: Expectations and Linguistic Meaning,LUHFDA/HFKO-1006-SE72. Bengt Carlsson: Evolutionary Models in Multi-AgentSystems, LUHFDA/HFKO-3004-SE73. Simon Winter: Introduktion till avhandlingen Expectationsand Linguistic Meaning, LUHFDA/ HFKO-7011-SE74. Nils Hulth, Peter Grenholm: A Distributed ClusteringAlgorithm, LUHFDA/HFKO-5053-SE75. Henrik Gedenryd: How Designers Work: Making Sense ofAuthentic Cognitive Activities, LUHFDA/ HFKO-1007-SE76. Peter Gärdenfors: Concept Combination: a Geometricalmodel, LUHFDA/HFKO-5054-SE pp. 129-146 in L. Cavedon,P. Blackburn,, N. Braisby and A. Shimojima (eds) Logiclanguage and Computation Vol 3, CSLI, Stanford, CA.77. Ingar Brinck, Peter Gärdenfors: Representation and Selfawarenessin Intentional Agents, LUHFDA/ HFKO-5055-SE78. Jana Holsánová: På Tal om Bilder, LUHFDA/HFKO-3005-SE1999 79. Jordan Zlatev: The Epigenesis of Meaning in Human Beings,and Possibly in Robots, LUHFDA/HFKO-5056-SE. Publishedin Minds and Machines Vol. 11 (2), pp. 155-195.80. Pierre Gander: Two Myths about Immersion in New StorytellingMedia, LUHFDA/HFKO-5057-SE81. Jens Månsson: Occluding Contours: A Computational Modelof Suppressive Mechanisms in Human Contour PerceptionLUHFDA/HFKO-5058-SE2000 82. Ingar Brinck: Attention and the Evolution of IntentionalCommunication. LUHFDA/HKFO-5059-SE2001 83. Jana Holsánová: Picture Viewing and Picture Description:Two Windows on the Mind. LUFDA/HFKO-1008-SE84. Philip Diderichsen: Visual Fixations, Attentional Detection,and Syntactic Perspective, LUHFDA/ HKFO-5060-SE85. Eds. Christian Balkenius et al: Proceedings of the FirstInternational Workshop on Epigenetic Robotics, ModelingCognitive Development in Robotic Systems, LUHFDA/HFKO-2002-SE86. Eds. Kai Oliver Arras et al: Proceedings of the FourthEuropean Workshop on Advanced Mobile Robots(Eurobot'01), LUHFDA/HFKO-2001-SE87. Mikael Johannesson: The problem of Combining Integral andSeperable Dimensions, LUHFDA/HKFO-5061-SE88. Ingar Brinck, Peter Gärdenfors: Co-operation and Communicationin Apes and Humans, LUHFDA/ HKFO-5062-SE89. Lars Kopp, Peter Gärdenfors: Attention as a MinimalCriterion of Intentinality in Robots, LUHFDA/HKFO-5063-SE2002 90. Mikael Johannesson: Geometric Models of Similarity.LUHFDA/HFKO-1009-SE91. Peter Gärdenfors: Cooperation and the Evolution of SymbolicCommunication, LUHFDA/HFKO-5064-SE92. Peter Gärdenfors: Att skapa gemensamma visioner: Omspråkets roll i människans utveckling, LUHFDA/HFKO-5065-SE93. Jan Morén: Emotion and Learning. A Computational Modelof the Amygdala LUHFDA/HFKO-1010-SE94. Eds. Christian Balkenius et al: Proceedings of the SecondInternational Workshop on Epigenetic Robotics, ModelingCognitive Development in Robotic Systems, LUHFDA/HFKO-2003-SE

95. Lars Hall, David de Léon & Petter Johansson: The Future ofSelf-Control: Distributed Motivation and Computer-MediatedExtrospection. LUHFDA/HFKO-5066-SE96. Petter Johansson: The New Contenders: On How to Model theMind. LUHFDA/HFKO-5067-SE97. Jens Månsson: The Uniqueness Constraint Revisited: ASymmetric Near-Far Inhibitory Mechanism ProducingOrdered Binocular Matches. LUHFDA/HFKO-5068-SE2003 98. Christian Balkenius, Anders J Johansson & Anna Balkenius:Color Constancy in Visual Scene Perception.LUHFDA/HFKO-5069-SE99. Annika Wallin: Explaining Everyday Problem Solving.LUHFDA/HFKO-1011-SE100. Lars Kopp: Natural Vision for Artificial Systems: ActiveVision and Thought. LUHFDA/HFKO-1012-SE101. Eds. Christian Balkenius et al: Proceedings of the ThirdInternational Workshop on Epigenetic Robotics, LUHFDA/HFKO-2004-SE102. Agneta Gulz: Thinking Styles and Socially Enriched LearningMaterial: Differential Effects on Motivation and MemoryPerformance. LUHFDA/HFKO-5070-SE103. David de Léon: The Cognitive Biographies of Things.LUHFDA/HFKO-5071-SE104. David de Léon: Actions, Artefacts and Cognition: AnEthnography of Cooking. LUHFDA/HFKO-5072-SE105. David de Léon: Artefactual intelligence: The devel-opmentand use of cognitively congenial artefacts. LUHFDA/HFKO-1013-SE.106. David Bengtsson: The Nature of Explanation in a Theory ofConsciousness. LUHFDA/HFKO-5073-SE107. Lars Hall & Petter Johansson: Introspection andExtrospection: Some Notes on the Contextual Nature of Self-Knowledge. LUHFDA/HFKO-5074-SE108. Lars Hall, & Petter Johansson: On the Role of Brain-Imagingin the Calibration of Mental Strategies. LUHFDA/HFKO-5075-SE109. Monica Bredefeldt Öhman: Livet som figur - Omsjälvbiografiskt minne. LUHFDA/HFKO-3006-SE110. Lars Hall och Petter Johansson: Neurofeedback and Metacognition.LUHFDA/HFKO-5076-SE111. Lars Hall och Petter Johansson: Self-Regulation in Education:A Ubiquitous Computing Perspective. LUHFDA/HFKO-5077112. Andreas Olsson: Emotion and Motivation in Learning:Current Research, Future Directions and PracticalImplications. LUHFDA/HFKO-5078-SE113. Lars Hall: Self-Knowledge/Self-Regulation/Self-Control: AUbiquitous Computing Perspective. LUHFDA/HFKO-1014-SE114. Anna Tebelius: Fonologiskt och semantiskt minne av bilderoch ord – En dyslexistudie. LUHFDA/HFKO-5079-SE2004 115. Åsa Harvard: Lek, lärande och motorik. Delrapport till ”TheLearning Brain, the Learning Individual, the Learningorganisation.” LUHFDA/HFKO-5080-SE116. Anders Sandberg: Computational Memory Models.LUHFDA/HFKO-5081-SE117. Eds Luc Berthouze et al.: Proceedings of the 4 th InternationalWorkshop on Epigenetic Robotics, LUHFDA/HFKO-2005-SE118. Marie Gustafsson: Making Agents Less Annoying: Towardsan Animated Agent that Responds to Social Cues.LUHFDA/HFKO-5082-SE119. Eds Yvonne Eriksson and Kenneth Holmqvist: Language andVisualisation.2005 120. Pierre Gander: Participating in a Story: Exploring audiencecognition. LUHFDA/HFKO-1015-SE121. Jordan Zlatev, Tomas Persson and Peter Gärdenfors: BodilyMimesis as the“Missing Link” in Human Cognitive Evolution.LUHFDA/HFKO-5083-SE122. Mathias Osvath and Peter Gärdenfors: Oldowan Culture andthe Evolution of Anticipatory Cognition. LUHFDA/HFKO-5084-SE123. Eds Luc Berthouze et al.: Proceedings of the 5 th InternationalWorkshop on Epigenetic Robotics, LUHFDA/HFKO-5070-SE124. Conference Abstracts: European Society for Philosophy andPsychology, LUHFDA/HFKO-8453-SE