Bodily mimesisas “the missing link” in human cognitive evolution

Zlatev, Jordan; Persson, Tomas; Gärdenfors, Peter

Published in: LUCS: Lund University Cognitive Studies

2005

Link to publication

Citation for published version (APA): Zlatev, J., Persson, T., & Gärdenfors, P. (2005). Bodily mimesisas “the missing link” in human cognitive evolution. LUCS: Lund University Cognitive Studies, 121. http://www.lucs.lu.se/LUCS/121/LUCS.121.pdf

Total number of authors: 3

General rights Unless other specific re-use rights are stated the following general rights apply: Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal

Read more about Creative commons licenses: https://creativecommons.org/licenses/ Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

LUND UNIVERSITY

PO Box 117 221 00 Lund +46 46-222 00 00 Jordan Zlatev, Tomas Persson and Peter Gärdenfors, Lund University Cognitive Studies, 121, (2005).

Bodily mimesis as “the missing link” in human cognitive evolution

Jordan Zlatev1, Tomas Persson2 and Peter Gärdenfors2

[email protected], [email protected], [email protected]

1Department of Linguistics, Lund University, 2Cognitive Science, Department of Philosophy, Lund University

1. Introduction definition permits the formulation of a mimesis hierarchy consisting of four evolutionary stages. It is fairly uncontroversial that there is a large While the first two stages are shown to be within gap between the communicative and cognitive the capacities of non-human (henceforth, systems of non-human animals and those of simply apes), and the fourth merges with lan- human beings. There is much less consensus, guage, the third stage, or what we call triadic however, on what the nature of this gap is, and mimesis constitutes a “missing link” that can help even less on how it was bridged in evolution. A us explain the emergence of human uniqueness common view is that it is that constitutes as well as the origins of language. the quintessential human feature, and if we could In the main part of the article – Sections 4 to only explain the origins of language, we would 6 – we proceed to apply the mimesis hierarchy also have the answer to the nature and source of to three cognitive-communicative domains human uniqueness (Christiansen and Kirby closely connected to bodily mimesis: imitation, 2003). Our view is in some respects contrary to intersubjectivity (“theory of mind”) and gesture. Our this. We agree that human language is qualita- main method will be to analyze evidence from tively different from , and (and to a lesser degree from child in Section 2 we outline what we find to be the development) and attempt to identify levels essential differences between the two. However, within these domains that correlate with or are precisely because the gap between animal com- even constituted by the levels of the mimesis munication and language is so large, it is ex- hierarchy. By investigating three socio-cognitive tremely difficult to see how one could “evolve” domains in parallel – imitation, intersubjectivity into the other. Therefore, despite much current and gesture – our intention is to suggest evi- enthusiasm and a variety of conflicting theories, dence for the most likely source of human we can at present only offer more or less com- uniqueness. To preempt our conclusions in Sec- pelling “narratives” (Landau 1991) or, somewhat tion 7, our analysis suggests that capacities such more scientifically, “constraints” (Johansson as the understanding of others’ intentions 2005) for a theory of language origins. (Tomasello 1999) or even the sharing of inten- In this article, we hope to add an important tions (Tomasello et al. in press) are not the cru- piece to the puzzle by arguing that there is a cial characteristic that differentiates human be- semiotic capacity that is more basic than lan- ings and apes. Rather we will suggest that it is guage that distinguishes us from other animals: the combination of spontaneous detailed imita- bodily mimesis. In Section 3 we define the concept, tion of actions leading to the formation of mi- following the lead of Donald (1991), but elabo- metic schemas, and the understanding of the semi- rating it in a way that makes it more precise and otic potential of gestures, or what we will call the less susceptible to criticism. Furthermore, our communicative sign function. We will suggest that the Lund University Cognitive Science-LUCS 121 ISSN 1101 – 8453 © 2005 by the authors

2 Bodily mimesis as “the missing link” in human cognitive evolution co-evolutionary cycle between imitation and lar circumstances and are produced more or less gesture (as well as intersubjectivity) paved the automatically, i.e. with very little volitional con- way for the evolution of language in an evolu- trol (Deacon 1997; Hauser 1996). In contrast, tionary process of human “self-enculturation” human utterances are largely independent from that parallels the Vygotskian notion of the inter- the physical context in which they are produced, nalization of social, interpsychological phenom- and usually refer to states and events that are not ena in ontogeny. currently present, thereby showing what Hockett (1960) called displacement, which also allows the speaker to inform the hearer of something he 2. From animal communication to human does not know. Furthermore, while not every language step in the construction of an utterance is under conscious control, the overall activity of human In comparing animal communication and human discourse would be impossible unless it was language it is not difficult to find striking differ- governed by a kind of “intermediate-term mem- ences on a number of cognitive dimensions: ory” that requires reflective consciousness First, the signals that constitute animal communi- (Donald 2001). cation are mostly predetermined by the genetic Looking at communication from the side of makeup of the organism. Even though it is clear the interpreter, the response to an animal signal that genes do not directly determine behavior is a behavioral pattern that is also relatively fixed, but set constraints on ontogenetic development, e.g. climbing up the tree in reaction to a specific and in this sense all development is epigenetic type of alarm call. In contrast, discourse inter- (Badcock 2000), it is still possible (and necessary) pretation is both individually flexible and collec- to distinguish the opposite poles of the endoge- tively negotiable (Vygotsky 1978; Sinha 2003). neous/exogenous continuum. In this respect, Moreover, even though animal signals can animal signals, be they the quacking of frogs or and do transmit information about the external the play-face of , can be regarded as environment to the interpreter of the signal (as by and large innate rather than learned. To the in the case of alarm calls) and such signals have extent that animal signals do involve learning – been called “referential” and even “symbolic” for example in fixing the group-specific “dialect” (Marler 1985), this is not the kind of referential of songbirds (Marler 1991) or determining the or representational relationship between expres- appropriate environmental trigger for the pro- sion and content that is involved in language. In duction of an innate alarm call in vervet mon- connection with context independence and voli- keys (Cheney and Seyfarth 1990) – this learning tional control, human reference can be intention- involves cognitively simpler (and better under- ally deceptive (“Wolf!”), while the calls of the stood) processes such as triggering, calibration vervet monkeys can at most be functionally so and inhibition (Hauser 1996). On the other (Whiten and Byrne 1988; Hauser 1996). One hand, neither the expression nor the meaning of way to summarize the difference is to say that linguistic signs is in any way genetically deter- human language is, in general, triadic – it is used mined, and the same may equally well apply to by one individual for explicitly representing an grammar (Tomasello 2003), though this is existing or imaginary state of affairs for another somewhat more controversial. Nevertheless, as a individual. On the other hand, animal signals are whole, it is clear that the learning of language is generally speaking dyadic: the animal responds to based on cultural processes of considerable a stimulus or cue in the environment, causing the cognitive complexity (Bloom 2000; Sinha 2003; emission of a particular signal directed to a con- Zlatev 2003), and arguably on reflective con- specific, without linking self, referent and ad- sciousness (Robinson 1996; Gärdenfors 2003; dressee into a referential triangle (Sinha 2003).1 Mandler 2004; Zlatev 2005b). Thus, despite the need for some qualifications, in general terms 1 The question of the referential status of animal animal signals are more or less innate, while signals may be debated, but it is significant that one language is learned, and culturally transmitted of those who most contributed to their rich interpre- across generations. tation as “symbols” is currently seeing them in a Another important difference compared to more modest light: “More recently, Marler’s ex- language is that animal signals are tied to particu- perimental approach and writings on the topic of signal meaning have moved away from high-level Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 3

Finally, the units of language stand in com- from a large number of empirical and theoretical plex internal semantic relations involving e.g. studies. complementarity (Deacon 1997) and grammati- cal relations involving hierarchical structure and Table 1. Six critical differences between animal recursivity (Hauser, Chomsky and Fitch 2002; communication and human language Givón 2002). Nothing of the kind has been Animal communi- Human language shown to exist in natural animal communication, cation though aspects of both semantic relations and grammatical patterns are not beyond the grasp 1. Degree of None (“innate”) Extensive of language-taught apes (Savage-Rumbaugh and learning or highly limited Lewin 1994). 2. Conscious None or highly High We may conclude that the six general differ- control limited ences between language and animal communica- 3. Contextual- Tied to a par- Flexible, rela- tion discussed in this section and summarized in ity ticular context tively independ- Table 1 define two very different types of semi- (stimulus setting) ent from specific otic systems. Even though one may argue that context for each of the individual features there may be 4. Interpreta- (Relatively) fixed Flexible, “nego- some forms of animal communication that tion response tiable” would appear to be “intermediary”, e.g. the Dyadic: 5. Communica- Triadic: - Cue (environ- flexible interpretations of alarm calls tive relations Speaker- ment, subject) by Diana monkeys (Zuberbühler 2000), no animal addressee- - Signal (subject, in the wild has anything approaching the socially trans- referent recipient) mitted, voluntarily controlled, contextually flexible, triadic semiotic system that is language. 6. Systematicity None, or very High (internal rela- limited As pointed out in the introduction, the diffi- tions and cult question is to account why and how such a combinatorics) distinctively “odd” (from a biological perspec- tive) communication system emerged in the first Animals and human beings differ not only in place. Proposals in the currently exploding litera- their means of communication, but in (appar- ture on language origins vary widely on what ently) non-communicative aspects of cognition they find to be the crucial feature of human such as planning, symbolic thought, “theory of cognition that could “bridge the gap” to lan- mind”, etc. In other words, there is a parallel guage: processing of recursivity (Hauser, Chom- between the “communicative gap” and the sky and Fitch 2002; Bickerton 2003), enhance- “cognitive gap”. What is the source of this paral- ments in social cognition (Tomasello 1999; lelism? When it comes to ontogeny, a possible Tomasello et al. in press), mechanisms support- explanation for the source of the “higher func- ing the learning and use of symbolic relations tions” of human beings is the classical proposal (Deacon 1997, 2003). In the present quest for laid out by Vygotsky (1978) that the latter can be the Holy Grail of explaining human uniqueness seen as resulting from the internalization of interper- that tends to focus on the sonal relations and representations, or as expressed by different researchers tend to offer their favourite Vygotsky in an often quoted passage: candidate for “the missing link” and do their best to support their hypothesis, often at the Every function in the child’s cultural devel- price of disregarding a good deal of evidence opment appears twice: first, on the social (see Johansson 2005). Our overall project simi- level, and later, on the individual level; first larly aims to resolve the puzzle of what makes between people (interpsychological), and then human beings cognitively and communicatively inside the child (intrapsychological). This ap- plies equally to voluntary attention, to logical special, but differs from the majority in (a) not memory, and to the formation of concepts. focusing on language but on a skill that we argue All the higher functions originate as actual re- is a prerequisite for language and (b) drawing lations between human individuals. (Vygotsky 1978: 57) cognitive interpretations and toward more low-level mechanisms linked to the particular stimulus fea- We wish to suggest that an evolutionary analog tures of the environment.” (Hauser 1996: 59) to this process, i.e. a process of hominid self- 4 Bodily mimesis as “the missing link” in human cognitive evolution enculturation in which more complex forms of thoughts; (d) autocuing: production is voluntary as communication created selection pressures for opposed to instinctive; and (e) generativity: the more complex forms of cognition, can help us “ability to “parse” one’s own motor actions into understand the close connections between the components and then recombine these compo- two “gaps”. Such a socio-cultural perspective on nents in various ways” (Donald 1991: 171). human cognition and its evolutionary origins It can be observed that features (a)-(e) define does not conflict with, but rather presupposes a system of communication that is intermediate the crucial underlying role of our general primate between animal communication and language, as cognition (Donald 1991; Tomasello 1999; Zlatev characterized in Section 2 and Table 1. Like the 2003). In this article we will adopt such a bio- latter it is learned, flexible and possibly triadic. cultural perspective as a general “working hy- At the same time, it lacks the following critical pothesis”, without thereby arguing against more features of human language: individual-based alternatives (e.g. Gärdenfors • Full conventionality: not just shared, but 2003). Ultimately, we will need a synthesis of the known to be shared and thus a part of mu- social and the individual aspects of human cog- tual knowledge (Itkonen 1978; Clark 1996); nition in order to make sense of the nature of our major theoretical concept that is both inter- • Arbitrariness: the semiotic ground for the and intra-personal, as we show below. expression-content relation requires neither similarity, nor contiguity (Peirce 1931-1935; Sonesson in press) 3. Bodily mimesis • Extensive systematicity of the internal rela- tions among signs (Deacon 1997; Zlatev The concept of mimesis goes back at least to Aris- 2003). toteles, but as far as human cognitive evolution is concerned, we take our departure from the Thus the hypothesis that mimesis plays the role mimesis hypothesis of human origins presented by of a “missing link” (in the words of Donald Donald (1991, 1998, 2001). This hypothesis himself) in human cognitive evolution is a priori states that a specific form of communication and attractive. Furthermore, it has been backed up cognition (and corresponding culture) mediated by evidence from archeology, neurobiology, between those of the common -ancestor and cognitive psychology and developmental psy- modern humans based on “the ability to pro- chology, such as the homology between “mirror duce conscious, self-initiated, representational neuron” systems in monkeys used for action acts that are intentional but not linguistic.” recognition, and structures for the control of (Donald 1991: 168) In brief, Donald proposes imitation, mentalizing and language in human that while ape culture is based on “episodic cog- beings, the presence of a “mimetic stage” in nition” and associational learning, early Homo – human ontogeny and the presence of mimetic most likely Homo erectus/ergaster considering the thought in patients with total aphasia (e.g. Don- relative jump in brain size and material culture ald 1991, 2001; Nelson 1996; Zlatev 2002, 2005; witnessed at this stage of hominid evolution – Corballis 2002). evolved a new form of cognition. What sup- Nevertheless, the mimesis hypothesis has ported this was the fact that the body could be been met with much resistance and criticism used volitionally to do what somebody else is (Mitchell and Miles 1993; Tomasello 1993; etc) doing (imitation), to represent external events that can be summarized as follows: On the one for the purpose of communication or thought hand, Donald’s theory underestimates the cogni- (mime, gesture) and to rehearse a given skill by tion of non-human primates with respect to matching performance to a goal. tool-making and cultural traditions (Boesch The most important features that Donald at- 2003) and their ability to understand others’ tributes to mimesis are: (a) reference: mimetic rep- intentions (Tomasello, Call and Hare 2003). On resentations stand for something (for someone); the other hand, features (a)-(e) attribute so much (b) intentionality: this referential relationship is representational complexity to mimesis, that this grasped both by the “mimer” and the inter- obviates the need for a second cognitive transi- preter; (c) communicativity: the relationship in (b) tion to language (Laakso 1993). If this criticism can be used for the purpose of conveying is correct, then mimesis can hardly “bridge the Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 5 gap” as required, since it, so to speak, both gives involves a mapping between proprioception, involv- too little to the apes and too much to Homo erec- ing kinesthetic sense and the input from skin tus, making the width of the gap too large to receptors and serving as our basic form of self- bridge. Finally, Donald’s theory does not address knowledge (e.g. Edelman 1992; Gallagher 1995, the question why mimesis evolved in the first 2005) and exteroperception (above all vision, but place, and why it later became superseded by also hearing and touch), this explains the sense language. in which bodily mimesis involves the body, in- Appart from these issues, Donald’s definition cluding specific organs such as the vocal tract – of mimesis is unclear with respect to at least the to the extent that they are involved in proprio- following points: Are the features (a)-(e) men- ceptive-exteroceptive mapping. Since tioned above necessary and jointly sufficient or production, and possibly even speech compre- are some more central than others? Is there a hension according to the “motor theory of priority for the communicative or the non- speech perception” (Liberman and Mattingly communicative forms of mimesis? In which way 1985) involve such a mapping, this implies that should one interpret the notoriously ambiguous even speech involves bodily mimesis, at the terms “reference” and “intentionality”? Fur- same time as it transcends it due to symbolicity. thermore, if mimesis possesses all the features of By focusing on full conventionality and sys- language except for phonetic realization as tematicity as distinguishing language from bodily claimed by Laakso (1993), it remains unclear mimesis, it becomes clear in which sense mime- what exactly makes it “not linguistic” in the sis is “not linguistic”. Most importantly, how- words of Donald (see the citation given at the ever, the five criteria in the definition of bodily beginning of this section). Hence, our first step mimesis presented above allow us to classify in elaborating the mimetic hypothesis of human human and ape socio-cognitive skills along a origins is to offer a more specific (re)definition scale consisting of at least four distinct stages. of the concept, which we refer to as bodily mimesis We refer to this as the mimesis hierarchy (see Table in order to distinguish it from the broader con- 2). cept with Aristotelian roots: Table 2. The mimesis hierarchy. Definitions of the four (Def) Bodily mimesis: A particular act of evolutionary stages (for terms in italics, see the defini- cognition or communication is an act of bod- tion of bodily mimesis in the text) and examples of ily mimesis if and only if: corresponding types of acts.

1. It involves a cross-modal mapping between proprioception (kinaesthetic experience) and Stage Definition Examples exteroception (normally dominated by vision), 1. Proto- A bodily act involving Facial expres- unless proprioception is compromised. mimesis Cross-modality with sions, bodily (Cross-modality) proprioception, but synchronization 2. It it realized by bodily motion that is, or can lacking Volition or be, under conscious control. (Volition) Representation (or 3. The body (part) and its motion correspond both) to – either iconically or indexically – to some action, object or event, but at the same time 2. Dy- An interpersonal or Shared attention, are differentiated from it by the subject. (Rep- adic intrapersonal bodily imperative mimesis act displaying Volition pointing, mirror resentation) and Representation, but self-recognition, 4. The subject intends for the act to stand for not Communicative sign do-as-I-do imita- some action, object or event for an addressee. function tion (Communicative sign function) 3. Tri- As (2) but also in- Joint attention, But it is not an act of bodily mimesis if: adic volving Communicative declarative 5. The act is fully conventional, i.e. a part of mimesis sign function pointing, pan- mutual knowledge, and breaks up tomime (semi)compositionally into meaningful sub- 4. Post- As (3), but also in- Signed language acts that relate systematically to each other mimesis volving Symbolicity and to other similar acts. (Symbolicity) This definition resolves some of the problems In the remainder of this article we shall apply the mentioned earlier. By stating that bodily mimesis proposed mimesis hierarchy to the evolution 6 Bodily mimesis as “the missing link” in human cognitive evolution

(and to a lesser degree, the ontogeny) of three have different degrees of the mimetic skills un- cognitive-communicative capabilities: imitation der discussion, and thus form much more of a (Section 4), intersubjectivity (Section 5), and gestures cline. With these caveats we procede with our (Section 6). By analyzing evidence from prima- analysis. tology, and to a smaller extent child develop- ment, we will identify levels within these do- mains that correspond to the levels of the 4. Mimesis and imitation mimesis hierarchy. In other words, we consider whether there are simple forms of these capaci- Chimpanzees, , and ties that are in essence proto-mimetic; whether can copy behavior to some extent and in seem- there are forms that are dyadic mimetic; whether ingly various ways. The literature on ape imita- there is furthermore evidence for communica- tion in the broad sense of the term (see Whiten tive intentions, and thus for triadic mimesis. et al. 2004 for a recent review) can for our pur- Finally we ask whether there are specific forms poses be grouped into the following main ex- of imitation, intersubjectivity and gesture that perimental paradigms: depend on the acquisition of a conventional  Neonatal mimicking symbolic system, i.e. language, and are thus post-  Instrumental tool-use mimetic.  Experiments with “artificial fruits” (puz- We will show that the progression proto- zle boxes) mimetic > dyadic-mimetic > triadic-mimetic > post-  “Do-as-I-do” tasks mimetic, instead of the dichotomy animal signal- - in respect to body movements (“Simon ing vs. human language outlined in Section 2, says”) and in particular by distinguishing between dy- - in manipulation of objects adic and triadic mimesis, has implications for a  Ethological methods (i.e. field observa- theory of the origins of human cognitive tions of food processing techniques or in- uniqueness. It may seem that such a theory direct evidence from local traditions) goesagainst the stream of much current theoriz- ing in evolutionary science since it invokes rela- Most obviously relevant for our discussion of tively discrete “stages” requiring qualitative tran- stages of bodily mimesis are those types of ob- sitions, while both the nature of evolutionary servational learning in which an observer copies processes involving many small mutations and something of a model’s bodily actions, may it be in a the available fossile evidence seem to speak in laboratory tool-using situation, natural foraging favor of a prolonged gradual process (Johansson situation or mimicking in a do-as-I-do game. 2005). However, we hasten to note that there is Other forms of observational or social learning, no contradiction between the stage-like theory like emulation, available in several versions (Byrne that we will pursue and the evolutionary evi- 1998; Custance, Whiten and Fredman 1999), dence. Firstly, it is completely possible to have response facilitation (see Byrne 1999), and stimulus- relatively discrete transitions, or “punctuated and local enhancement (see Tomasello 1996), can equilibria” (Eldredge and Gould 1972) within a inform observers about properties of the envi- basically gradual framework, i.e. without this ronment and objects, or the connection between implying any form of “saltations” or sudden certain manipulations and outcomes, but do not “macro-mutations” (Dessalles 2004). The likeli- involve copying bodily actions per se and thus it hood for fairly rapid and abrupt changes is even is not clear if they involve bodily mimesis. higher when biological and cultural evolution For our purposes, the collective term imitation interact, as has most likely been the case with pertains most clearly to the copying of bodily human cognitive evolution including language motions (the do-as-I-do experimental paradigm) (Deacon 1997). Secondly, our notion of a mime- as well as to imitation on the action level in object sis hierarchy, as other similar constructs (Den- manipulation. The notion of action level imita- nett 1996; Zlatev 2003), is meant to serve pri- tion was originally contrasted to that of program marily as a conceptual role in helping us relate and level (Byrne and Russon 1998), the latter being compare primate skills in somewhat different the hierarchical organization of a series of ac- domains: it is not an evolutionary ladder for tions and the former the execution of an indi- classifying different species, which may very well vidual component action. Note that imitation on Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 7 the program level can take place without imita- others as intentional agents (e.g. Tomasello tion on action level. The program level has been 1999). Since then both the field of primate imita- further divided by Whiten (1998) into being tion studies has expanded and the MPI Leipzig either of a sequential or a hierarchical kind. An group has modified their views on apes’ abilities example of sequence imitation from the artificial- to understand intentional agency (Tomasello et fruit experiments would be to manipulate obsta- al. in press, see Section 5 for more discussion). cle A before obstacle B, while an example of Donald’s definition of mimesis (presented in imitation on the action level would be poking vs. Section 3), comprising reference, intentionality, com- pulling to remove a plug. If the subject copies municativity, autocuing and generativity, is difficult to the manual movement of the model, it is re- apply to non-human primate imitation because garded as an instance of action-level imitation, observational learning through copying has been but if the ape instead copies the movement of studied in non-communicative tool using and the plug, it is categorized by Custance et al. puzzle box situations, with few exceptions such (1999) as object movement re-enactment. The latter is as the study by Tomasello et al. (1994) on chim- usually regarded as a form of emulation, where panzee gestures in which the authors did not the end state of an event is achieved (or copied) find any observational learning involved in func- but the subject invents his or her own way to tional gesture acquisition. However, using our achieve this end and the model is in principle definition (see Section 3), it is possible to apply superfluous. However, since object movement the concept of bodily mimesis to findings in the re-enactment is a special kind of emulation field of ape imitation. As we will show in this where at least the direction of the movement of section, these findings can be interpreted as the object is copied, in our analysis it would evidence that apes are capable of at least proto- qualify as a form of bodily mimesis. mimetic and dyadic mimetic imitation, and pos- There is also the hypothetical case where the sibly even triadic mimetic imitation when a com- subject only learns that the plug can be moved municative sign function is added. and then proceeds to bring about that effect in whatever way he or she prefers. To determine what is what, however, is a tricky business. In 4.1 Proto-mimetic imitation the latter case the subject might happen to bring about by chance the same movement of the plug In what we call proto-mimetic imitation there is as the model showed, and thus give the impres- (iconic) resemblance between the copy and the sion of object movement re-enactment. Fur- copied, which is probably based on a basic form thermore, in both a false and in a true case of of identification with the model (see Section object movement re-enactment, the subject can 5.1). For this to be classified as proto-mimetic also happen to perform the same manual actions rather than dyadic-mimetic, there should be no as the model by chance and give the impression clear evidence for volition or for differentiation of bodily imitation on the action level. At least between oneself and the model. three forms of observational learning can thus The prime candidate for proto-mimetic imi- yield the same expression: simple emulation, tation is neonatal facial mirroring. In the case of object movement re-enactment and action level human neonatal mirroring, the field is divided imitation. The experiments with artificial fruits, between those who prefer a minimalist interpre- some of which we will review below, are de- tation and those who argue for a more mentalis- signed to tease apart these copying strategies. tic one. For example, Anisfeld (1996) in a meta- The failures of great apes to fully copy be- analysis of experiments by Meltzoff and Moore havior in several early studies led Tomasello, and Heimann argues that the infant’s response is Kruger and Ratner (1993) to conclude that apes limited to one facial movement, tongue protru- emulate, but cannot imitate, since imitation in sion, and not several. If so, it would be a largely their view entails copying both of the goal and reflexive form of mirroring unlikely to require the method to bring about that goal. For volition on the part of the infant, and thus would Tomasello and his collaborators imitation thus be proto-mimetic. On the other hand Meltzoff has a “theory of mind” component by definition, and Moore (1977, 1983, 1994) have argued that and on the basis of the imitation data, the group the human infant not only displays volition (i.e. has argued that apes lack an understanding of choice on whether to imitate or not, and ability 8 Bodily mimesis as “the missing link” in human cognitive evolution to mirror more than tongue protrusion), but that When trying to implement the imitation of he/she differentiates between him/herself and the goals in an experimental setting the goal is often model. Meltzoff and Moore (1994) argue that to get hold of a food item, and these experi- there is evidence for deferred imitation in 6- ments have tended to either foster innovation or week-old infants who imitated the action when use of old reliable methods. This is not surpris- seeing the model again after a delay of 24 hours. ing. If you truly want to get hold of a food item, While differentiation in itself does not require and apes clearly do so, the most effective means the kind of displacement that is witnessed in one can come up with is used, and that is not deferred imitation (see Sonesson in press), we necessarily the one witnessed. agree with Piaget (1945) that the capacity for The well-known rake experiment of Nagell et deferred imitation rests on at least a simple form al. (1993) is an example of imitation studies us- of representational activity. Thus, if the Melt- ing an instrumental tool. If there is any learning zoff-Moore interpretation of neonatal imitation involved in the case of apes it is generally con- is correct, we must conclude that human chil- sidered to be in the form of emulation: the sub- dren perform dyadic imitation (see below) from jects learn something new about tools, their a very early age. movements or the environment, but not about In the case of apes, newborn chimpanzees the bodily actions the model is using to bring have been shown to have a period of innate about the intended goal. mirroring responsiveness to human facial stimuli Better imitation results have been obtained in (mouth opening and tongue protrusion) that is experiments with so called “artificial fruits”, i.e. present to 11 (Myowa-Yamakoshi 2001) or 9 boxes with obstacles that the subject has to work (Myowa-Yamakoshi et al. 2004) weeks of age. around in order to get to the food content. This shows that neonate chimpanzees have (at These experiments address the question: What least) the capacity for proto-mimesis with re- do apes spontaneously copy when they get an spect to faces. It is to our knowledge not yet opportunity to observe a model? Which strate- tested whether this capacity also involves differ- gies and mechanisms of social learning are in- entiation or volition. But different sources for volved? The artificial fruit experimental para- highly similar responses in human and ape new- digm has yielded mixed results, as strategies of borns seem unlikely. result emulation, object movement re- enactment, action-level imitation and sequence imitation have been recorded (Whiten et al. 4.2 Dyadic-mimetic imitation 2004). The results seem to vary over fruit used and species tested and/or age of subjects (Stoin- In what we call dyadic-mimetic imitation, the ski and Whiten 2003; Custance et al. 2001; Stoin- subject differentiates between oneself and the ski et al. 2001; Whiten et al. 1996). Action-level model and understands that a bodily posture or imitation implies that the subject can represent a motion can correspond to something else, like model’s manual movements and re-enact these an object or action. However, there is no at- from memory. Performing sequence imitation tempt to communicate this “something else” to means that the subjects are also able to parse a another individual by means of the bodily mo- string of component actions and later reassem- tion. ble these in the same fashion from memory, In a non-communicative setting, like in the which again show their ability to re-enact de- experiments on social learning studied with non- tached representations voluntarily. human primates the “something else” men- Apes have demonstrated some skills of ac- tioned above would be to understand the mean- tion-level and sequence imitation, but at the ing of a model’s actions: what the model is trying same time they experience some difficulties with to achieve. As mentioned above, imitating the them as evident by their lack of full copying. goal as well as the means to bring about that goal Imitating the use of familiar motor actions in is the hallmark of (true) imitation according to novel situations seem to come about more easily Tomasello, Kruger and Ratner (1993), and in- in apes than copying new motor actions alto- volves certain levels of intersubjectivity involv- gether (Myowa-Yamakoshi and Matsuzawa ing understanding of intentions. 1999). Especially for young individuals actions are more likely to be copied if they are close to Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 9 the ones already in the subjects’ repertoire Deferred imitation on objects, involving do- (Bjorklund et al. 2002). It is not yet settled as-I-do imitation, is perhaps the most promising whether apes’ relatively poor performance on evidence for dyadic mimetic capacities in apes. detailed imitation of bodily actions is due to In these experiments the subject is shown ac- motor-perceptual aspects or skills in intersubjec- tions on objects and is then either asked to “do tivity. However, since recent studies have shown the same thing” (if the the verbal command is that at least chimpanzees have a degree of un- learned) or (if not) the spontaneous handling of derstanding of the mentality of others (see below the object is recorded. The results show mainly and Section 5.2), the motor-perceptual explana- approximate imitation but also exact copying tion should not be underestimated. and a notable increase in ability with age (Bjork- Goal directedness is lacking in the most bod- lund and Bering 2003; Bjorklund et al. 2002; ily mimetic paradigm, the “do-as-I-do” experi- Bjorklund, Bering and Ragan 2000; Bering et al. ments, where part of making sure that the copy 2000). However, mother-reared chimpanzees is acquired through observation is to perform seem to do less well while enculturated apes can never-before seen, or at least unusual, move- outperform human children on certain tasks ments without any practical function. There is (Tomasello, Savage-Rumbaugh and Kruger no food item to get to at the end of the action 1993). The fact that the test is made after a delay sequence. The subject is verbally asked to do the heavily strengthens the need for representation and same as the model on cue. The do-as-I-do para- recall and execution from memory. Bjorklund digm is an example of studies that ask the ques- and Bering (2003) suggest that part of what tion: Can apes copy behavior? The answer is comprises the enculturation phenomenon is a positive. Both cross-fostered (or “enculturated”) greater conceptualization of human behavioral apes, e.g. the orangutang (Miles et al. programs. It is possible that non-human pri- 1996) and the chimpanzee (Custance et al. mates in wild populations have critically less 1995), and chimpanzees with more moderate exposure to long series of goal-directed behavior human upbringing (Custance et al. 1995) can than apes around humans to fully be able to pass these tests. Once they understand the follow through longer strings of actions. meaning of the “do the same” command, any Byrne (1999) makes a similar proposal, but faults in the copy is bound to be motor- stresses that it is general familiarity with the perceptual and not due to intersubjectivity is- component actions in human manipulation of sues. The do-as-I-do experiments prove that objects that gives enculturants the relevant ad- apes are capable of both volition and representation vantage. A third possibility is that deferred imita- (they represent the model’s body), which qualify tion on objects has been tested without any them as dyadic-mimetic imitators. However, other reward than what the object manipulation learning the concept of SAMENESS may influence itself can give, and that enculturated animals find the performance of apes profoundly. Thomp- human objects more interesting and rewarding son, Boysen and Oden (1997) found that lan- than do others. Infant apes seem to be more guage-naïve chimpanzees learned to judge rela- interested in objects when objects are handled by tions between relations (A is to A as B is to B, human caretakers interacting with the infant and not as C is to D) in a matching-to-sample than when the objects are handled by their con- task only if the subjects had previously learned a specific mothers, who typically do not engage token for the concept SAME. Ordinary matching- infants with objects (Bard and Vauclair 1984). to-sample training could not yield the same re- It has furthermore been found that the direc- sult. Experiments that hinge on the subject “get- tion of object manipulation, i.e. an object’s ting the point of the game”, so to speak, have movement in space, is more often copied than implications for ecological validity since the bodily actions performed on the object (Myowa- experimenter has to plant certain knowledge in Yamakoshi and Matsuzawa 1999). This again the subject that it might not have stumbled upon implies that apes copy the most salient rather naturally. This is especially problematic for evo- than detailed aspects of an action. It is also pos- lutionary reasoning around comparative experi- sible that since objects draw much attention it mental psychology, since it is not clear what might obscure the relevance of a model’s manual circumstances would have “planted” such actions, although this remains an untested hy- knowledge in the hominid line. pothesis to our knowledge. 10 Bodily mimesis as “the missing link” in human cognitive evolution

Whiten et al. (2004) conclude in their review Ranter 1993; Tomasello 1996, 1999) in assuming of the imitation literature that many experimen- that apes lack second-order intentionality (“I tal results on imitation can be interpreted as realize that you have intentions”), and that they object movement re-enactment (a form of emu- can therefore not imitate but only emulate, apes lation) rather than copies of bodily actions. But should necessarily lack mimetic abilities of a the line between the movement of an object and triadic kind. However, the most recent and the hand applying the force to the object in or- promising experiments on apes’ perspective- der to bring about that movement is not a sharp taking have shown that chimpanzees know the one. (Does one copy the fall of the hammer or importance of others’ visual attention, both in the fall of the hand holding it?) Thus, we see no communicative contexts (Tomasello et al. 1994; reason to treat both of these types of imitation – Krause and Fouts 1997; Pika, Liebal and action level imitation and object movement re- Tomasello 2003, 2005) e.g. use of attention- enactment – as being qualitatively different. This getting gestures and competitive contexts (Hare is captured in our analysis by stating that both et al. 2000; Hare, Call and Tomasello 2001). involve dyadic mimesis. Apes also seem to be able to distinguish inten- Finally, social learning of feeding techniques tional from accidental acts (Call and Tomasello in gorillas (Byrne and Russon 1998) and chim- 1998) and to judge an experimenter as either panzees (Stokes and Byrne 2001; Byrne and unwilling or unable to give them a treat (Call et Stokes 2002), and human chore re-enactment al. 2004). The conclusion is therefore that apes (e.g. washing clothes) in orangutans (Russon and do understand second-order mentality, at least in Galdikas 1993) are further behaviors that might the form of second-order attention and (possi- fall in the dyadic mimetic category, although in bly) second-order intentions. However, under- the latter case it is difficult to argue that the re- standing the communicative sign function, and enacted act represents the observed one, since thus triadic mimesis, requires third-order mental- the function of the act is not the same: An ity – as we will argue in Section 5 – and there is does not re-enact cleaning when yet no clear evidence that (non-enculturated) mopping a floor with dirty water. At the same apes are capable of this. time, as in the “Simon says” game, the orangutan Possible instances of triadic mimetic imita- can still be said to represent the bodily actions tion in non-human primates would be copying themselves, or if it is a case of object movement of referential signs in language-taught apes. re-enactment, the movements of objects. However, most of the learning of signs by apes that have been taught a signed language, such as the gorillas and (Bonvillan and 4.3 Triadic-mimetic imitation Patterson 1993, 1999), the chimpanzee (Fouts 1972) and the orangutan Chantek (Miles Triadic-mimetic imitation involves the full- 1990) takes place by getting the hands molded blown form of bodily mimesis, with the charac- into the signs and therefore does not qualify as teristics volition, representation and communicative sign either imitation or bodily mimesis. Still, Gardner function. Triadic mimesis introduces a communi- and Gardner (1969) also used the “do-as-I-do” cative context that is mostly lacking in imitation paradigm to teach new signs to Washoe with studies on apes since these studies seldom take some success. A more general problem, how- into account both parts of the learning dyad: ever, is that it is difficult to distinguish the imita- model and imitator. Teaching, in the form of tive learning of such signs from post-mimesis instruction, seems to be absent in ape interac- since these apes were taught a simplified form of tions (Boesch and Tomasello 1998; Matsuzawa (ASL), and that implies et al. 2001). that they were exposed to, and possibly even Above all, triadic bodily mimesis strengthens acquired at least a degree of symbolicity (involv- the call for understanding (others’) mentality, ing conventionality and systematicity). In a study which is perhaps the most problematic issue of the functions of the repetition of lexigrams by when it comes to relating bodily mimesis and language-trained chimpanzees and bonbobos imitation, mainly because the role of intentional- Greenfield and Savage-Rumbaugh (1993) report ity in imitation is already debated. If we were to at least some clear cases of the acquisition of follow Tomasello (Tomasello, Kruger and lexigrams through imitation. At the same time Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 11

Savage-Rumbaugh and Lewin (1994) emphasize by imitation, there may be doubt concerning the that only when the chimpanzees Sherman and degree of their conventionality and systematicity, Austin acquired the referential meaning of the e.g. deaf native signers have difficulties interpret- lexigrams, after a prolonged and laborious proc- ing apes’ signing (Pinker 1994). ess of training, could they match lexigrams to One form of imitation that is clearly symbolic samples. (and hence post-mimetic) is what Tomasello Perhaps the most convincing evidence for (1999) calls role-reversal imitation: imitation of sign triadic-mimetic imitation in apes involves the use in which the subject learns a new sign by invention of iconic signs by Koko (Patterson observing and imitating the signer, and at the 1980) and Chantek (Miles 1990). Such sign are same time shows understanding of the sign’s examples of representation (the ape represents meaning by “reversing” the perspective to fit his something in the environment) and communicative role in the interaction. A simple case would be to sign function (the ape addresses this to his or her learn the sequence “Here you are – Thank you!” caregivers). Since the imitation is creative, in- so that after a while even the child can offer an volving the resemblance (iconicity) between the object to the adult with words “Here you are” sign and the way the referent is typically used, it (or equivalent) and expect to hear “Thank you”. cannot be solely a product of being taught a A child acquiring a signed language performs symbolic code. The fact that Koko’s and Chan- likewise, reversing e.g. the direction of signs like tek’s sign inventions build on iconicity suggest GIVE, COME and the referent in deictic signs that they are cases of triadic-mimetic rather than like I and YOU, which involve pointing to the post-mimetic imitation. We return to these cases body of ego and the interlocutor, respectively. In in Section 6. one of the few well-documend cases of the spontaneous emergence of gestural communica- tion in apes, Tanner and Byrne (1996, 1999) and 4.4 Post-mimetic imitation Tanner (2004) show that at least one gorilla, the male Kubie, uses deictic gestures to refer to While speech may be regarded as in part post- himself, without his interlocutor, Zura imitating mimetic (see Section 3), the paradigm example these in a role reversal manner (see Section 6.3 of post-mimesis is signed language: it is based on for more discussion of these gestures). In the cross-modal mapping between proprioception case of language-taught apes, self-referential and vision, and possesses the other features of gestures are (typically) taught by molding, rather mimesis: volition, representation and than imitating (Patterson 1980). Hence we may communicative function. Furthermore, it is conclude that apes, including those who have symbolic, not in the Peircian sense of lacking any been taught language, do not display post- motivational relationship between the expression mimetic imitation in a non-ambiguous manner. and content poles of its signs – over 50% of the signs of ASL are judged to be iconic (Woll and Kyle 2004) – but in the sense of consisting of 4.5 Summary and conclusions conventional signs (presupposing third-order mentality, see Section 5) and systematicity (i.e. In this section we analyzed evidence from the internal relations between the signs). Other literature pertaining to apes’ imitation in terms instances of post-mimesis would be emblems like of the mimesis hierarchy defined in Section 3. the thumbs-up “OK” sign, or nodding instead Some of the various types of imitation discussed of saying “yes”, which fulfill the property of were classified according to our model as shown conventionality, and at least a limited form of in Table 3. The conclusion is that while apes are systematicity. clearly capable of proto-mimetic and even dyadic Accordingly, post-mimetic imitation would mimetic imitation, it is less clear if they are capa- consist of skills involved in the learning of such ble of triadic mimetic imitation (related to their signs. As pointed out above, some apes that difficulty in comprehending the communicative have been taught a signed language did acquire sign function). Finally, post-mimetic imitation is some of their signs in this manner, though the beyond their competence. majority of signs were acquired through molding of the hands. Even for those that were acquired 12 Bodily mimesis as “the missing link” in human cognitive evolution

In our framework, both sorts of limitations Table 3. The mimesis hierarchy and types of imita- can be argued to involve bodily mimesis: (a) with tion respect to forming mimetic representations (schemas), and (b) even more so of communicat- Level of mimesis Imitation skills ing these representations to another. In contrast, the “generative” human ability to parse series of Proto-mimesis Neonatal mirroring actions (regardless of understanding their func- tion or the model’s intentions) and to re-enact Dyadic mime- Do-as-I-do imitation these seems to be available to apes. In other sis words, and relevant for the debate on the origins Object movement re-enactment of language, it is not so much the syntax of imita- Action level imitation tion that is difficlt for apes, but its semantics. It is conceivable how ape copying ability, coupled Triadic mime- Learning sign use through imita- sis tion with a motivation to share one’s mental states (which can, of course, also mean attempting to Invention of iconic signs (rare impose one’s view on others), can give rise to cases) mimetic-communicatory interactions, in which Post-mimesis Role-reversal imitation (not at- levels of mimesis and intersubjectivity are closely tested) connected, as we will argue in the following section. This evidence supports some other conclusions as well. Although apes have been shown – in a number of different paradigms – to be able to 5. Mimesis and intersubjectivity copy actions and bodily movements through observation, a general trend is that they make We understand the notion of intersubjectivity proximate rather than exact matches, and some- broadly (and rather literally) as the sharing and times copy on a hierarchical or sequential level understanding of others’ mentality. The term “mental- rather than on the more basic action level where ity” is taken here to involve not only beliefs and the actual bodily matching takes place. This is in other forms of conceptual knowledge, but all stark contrast to some human children who tend forms of consciousness, including emotions, to “over-mimic” on the action level on the same desires, attentional foci and intentions (see artificial-fruit tasks as apes (Whiten et al. 1996) Gärdenfors 2003). Not all of these need to be or copy a way to use a tool that is ineffective for understood conceptually, but may involve pre- the goal (Nagell et al. 1993). conceptual, non-reflective understanding. Apes also tend to focus on and re-enact the While the close connection between bodily movement of objects rather than body move- mimesis and imitation discussed above, and ments. This, of course, can be argued to be the gesture (to be analysed in Section 6) is fairly logical strategy when dealing with the mechanics straightforward, this may not be the case with of objects. Approximations in matching have respect to intersubjectivity. The reason for this implications for our volition and representation re- lies, we believe, in the dominant cognitivist view quirements since they suggest that apes do not of mentality as consisting entirely (or at least possess skills that are specialized enough to predominantly) of propositional representations. match their movements against a template, In this view, to understand another’s mind which is a prerequisite for mimetic signing. This would imply having beliefs concerning their has sometimes been interpreted as emulation beliefs. This conception can easily be framed as caused by lack of theory of intentions (e.g. having some sort of “theory of mind” and, in- Tomasello, Kruger and Ratner 1993), but can deed, this has been the dominant conception in also, and probably more likely, be ascribed to the field (e.g. Baron-Cohen 1995). motor-perceptual difficulties per se. Also, as A great deal of the debate about the cognitive pointed out earlier, there is growing evidence abilities of apes has concerned whether or not they that apes are capable of understanding others’ have a “theory of mind” (Woodruff and mental states. Premack 1979; Byrne 1995; Gómez 1994; Heyes 1998; Tomasello 1999). Also in the discussion of Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 13 human cognitive development, the question when that is perceiving another animate organism.” and how children develop one has been a central (Gallagher in press). focus of content (Perner et al. 1987; Gopnik and This perspective on intersubjectivity was fur- Astington 1988; Gopnik and Meltzoff 1997; ther developed by Merleau-Ponty (1962), in Mitchell 1997). particular through the notion of the “corporeal But regarding intersubjectivity as a “theory” schema” which serves as “a normal means of involving (primarily) beliefs has had some nega- knowing other bodies” (Merleau-Ponty 2003: tive implications for understanding this funda- 218). More recently, Gallagher (1995, 2005, in mental suite of socio-cognitive capacities: (a) the press) has elaborated the distinction between question is typically posed as involving a dis- body schema and body image where the first is pre- crete, all-or-nothing property: either the child or conscious and serves as a precondition and animal “has a theory of mind” or doesn’t; (b) the backdrop for intentionality (cf. Searle’s (1992) role of one’s own body and actions – and those of notion of the Background), while the latter is “a the other – is minimized, and (c) “theorizing” (sometimes conscious) system of perceptions, about the minds of others (or even oneself) is attitudes, beliefs and dispositions pertaining to considered prior to acting in consort with them, one’s own body” (Gallagher 2005: 37). Empirical giving intersubjectivity a rather static, detached neuroscience has recently given support for the quality. role of the body in understanding others through Recently, however, these implications have mechanisms involving “mirror neurons” (Rizzo- been seriously questioned on the basis of both latti et al. 1996; Gallese, Keysers and Rizzolati new empirical evidence from child development 2004; Arbib in press), that mediate between the and primatology and older philosophical, above perception of another and the subject’s own all phenomenological, insights. With respect to proprioception and action. (a) it has become clear that “the generic term Within this “embodied” and action-oriented ‘theory of mind’ actually covers a wide range of perspective on the understanding of other processes of social cognition” (Tomasello, Call minds, unlike in the intellectualist cognitivist and Hare 2003: 239). Consequently one must one, bodily mimesis becomes clearly relevant. distinguish between these different processes in With respect to distinguishing different levels of order to understand the cognitive capabilities of intersubjectivity, we can apply the mimesis hier- animals, as well as children of different ages. For archy defined in Section 3 to yield a possible example, Gärdenfors (2003) splits the so-called evolutionary and developmental hierarchy of “theory of mind” into six levels, where the mid- intersubjectivitity. dle four clearly correspond to different socio- It should be noted that this way of “splitting cognitive capabilities: understanding others’ the theory of mind” is different from that pro- emotions, attention, intentions and beliefs, re- posed by Gärdenfors (2003) since the different spectively. Such a division can be complemented capabilities distinguished by the Gärdenfors with the hypothesis that the different capabilities (understanding emotion, attention, intention and correspond to different phylogenetic and onto- belief) are in a sense orthogonal to the mimesis genetic stages in evolution and development, and hierarchy. For example, the understanding of indeed Gärdenfors (2003) makes this proposal. others’ emotions occurs on each one of the lev- The criticism concerning (b) and (c) was de- els of the mimesis hierarchy, with increasing veloped first within the phenomenology of cognitive complexity. However, if we consider Husserl some one hundred years ago and tar- what may be regarded as a typical manifestation of geted the intellectualist perspective on the “un- the respective capability (emotion, attention and derstanding of other minds” (the usual term for intention), there turns out to be considerable intersubjectivity in philosophy) as a matter of overlap between the levels of intersubjectivity inference or analogy from the knowledge of defined in this article and those of Gärdenfors one’s own mind. As summarized by a modern (2003): Emotion can be shared on a proto- interpreter: “For Husserl, understanding another mimetic level, sharing attention involves (at person is not a matter of intellectual inference, least) dyadic mimesis, understanding communica- but a matter of sensory activations that are uni- tive intentions is centrally implicated in triadic fied in or by the animate organism or lived body mimesis (but “simple” intentions are understood dyadically), and understanding beliefs appears to 14 Bodily mimesis as “the missing link” in human cognitive evolution require explicit symbolic skills, which can be with experience and is accessible to conscious- argued to be post-mimetic since they depend on ness. language. Furthermore, the forms of intersubjectivity A central question in relating the proposed described here do not require a conceptual dif- mimesis hierarchy and intersubjectivity is ferentiation between self and other, which is whether the respective mimesis level serves as a necessary for establishing a correspondence precondition and a causal factor for the development relation between them, i.e. what we refer in our of corresponding skills of intersubjectivity. Or is definition as representation. This is not to say that it rather that independently reached insights into the young infant lives in a completely undiffer- the mind of others makes increasingly complex entiated world in which there is no awareness of forms of mimesis possible? Our general Vy- self whatsoever, as in classical accounts of infant gotskyan approach suggests the first scenario: cognition (e.g. James 1890). Nevertheless, even a bodily mimesis is a fundamentally interpersonal modern developmental psychologist who em- activity that excercises (in ontogeny) and pro- phasizes the role of awareness of others’ atten- vides selection pressures (in phylogeny) for de- tion and the presence of affective self- veloping more refined skills in mind reading. consciousness in the first months of life points Thus, our predominant take is that mimesis out that “older infants reveal a greater focus on drives intersubjectivity rather than the other way the self and the younger ones reveal a more im- around. At the same time we acknowledge that mersed, less detached focus on the other” (Reddy the “driving” metaphor is not completely ade- 2003: 401). This “more immersed, less de- quate since the causality runs in both ways and is tached” quality of the earliest forms of intersub- possibly best described in terms of co-evolution. jectivity supports our classification of them as We will return to this issue in the summary at proto-mimetic. the end of this section. Can this analysis be extended to the (early) interpersonal relations among apes? As pointed out in Section 4.3 “neonatal mirroring” has also 5.1 Proto-mimetic intersubjectivity been observed in apes (though this has so far only be attested with human, rather than ape Given our definition of bodily mimesis we can faces as stimuli). Since this is typically attributed regard some of the most basic forms of in- to a form of identification with the person imi- tersubjectivity as “proto-mimetic” to the extent tated, serving as a basis for intersubjectivity that they (a) consist of forms of interpersonal when children are concerned (Meltzoff and interaction that involve cross-modal mapping Gopnik 1993; Gallagher 2005), it can be viewed between proprioception and the (visual) percep- as evidence that at least apes, and possibly other tion of others, (b) do not fulfill either one of the mammals too, have such basic proto-mimetic characteristics volition and representation. This can intersubjectivity. The function of such “mirror- be made more precise using the distinction made ing” can be related to what is possibly the most by Gallagher (1995, 2005) between body schema basic form of intersubjectivity, both ontogeneti- and body image. The first is characterized as “a cally and phylogenetically: the ability to share system of sensory-motor processes that con- emotions, or empathy (Einfühlung). As a proto- stantly regulate posture and movement – proc- mimetic, non-representational capacity, this is esses that function without reflective awareness testified in early infancy and sometimes referred or the necessity of perceptual monitoring” (Gal- to as interaffectivity (Stern 2001). The well-known lagher 2005: 37-38). The second is, as pointed experiments described by Trevarthen (1992) out, “a (sometimes conscious) system of percep- show that parent-infant interactions in the first tions, attitudes, beliefs and dispositions pertain- few months take the form of a reciprocal rhyth- ing to one’s own body” (Gallagher 2005: 37). mic “dance”, and that frustration follows if this Given these definitions, we can state that acts of “attunement” is disrupted (notice the musical proto-mimesis involve (above all) the body schema, metaphors). The suggestion is that emotions which is largely innate (in the sense of being such as joy and suffering are perceived directly, present at birth) and pre-conscious, rather than possibly involving mirror-neuron structures the body image, which is gradually constructed similar to those involved in action recognition and imitation, rather than involving inferences to Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 15 underlying states (Gallese, Keysers and Rizzolati awareness of attention, but also as evidence of 2004). awareness of attention” (Reddy 2005: 95). Preston and de Waal (2002) have argued per- Until recently it has not been clear whether suasively that as a basic mechanism involving the such awareness of another’s attention exists in linkage of perception and action, a basic form of the interaction between infant apes and their empathy is available to most if not all mammal mothers, but in a recent study, Bard et al. (in species. Defining empathy as “any process press) report that the rates of mutual gaze be- where the attended perception of the object’s tween infants and their mothers are virtually the state generates a state in the subject that is more same in 3-month old human children and 3- applicable to the object’s state or situation than month old chimpanzees; 18-20 and 17 times per to the subject’s own prior state or situation” hour, respectively (though humans tend to en- (ibid: 4), they see a clear evolutionary motivation gage in longer bouts of mutual gaze). Further- for its emergence in the ability to recognize and more, the authors noticed a “cultural” difference understand the behavior of con-specifics. It is an between the apes at Primate Research Institue, open question how much of such matching be- Japan and those at Yerkes National Primate tween the visually perceived body of the other Research Center, USA, with the ones in Japan and the proprioceptively perceived body of one- engaging in mutual gaze at much higher rates (22 self is domain-general – and thus can be ex- vs. 12 times per hour), while the ape mothers in pected to be general across species – and how the USA cradled their infants more often (71% vs. much is specialized in the form of species- 40% of the total time). Intriguingly, a similar specific communicative signals such as facial inverse corrlelation between visual and tactile expressions. It is characteristic that such signals, contact has been observed in human societies, such as the famous “play-face” expression of with traditional cultures favouring touch and great apes (an evolutionary precursor to the Western ones gaze: “With reduced physical con- human smile) typically carry emotional rather tact found in Western societies, mutual engage- than referential meaning. ment shifts to the visual system, arguably a more A second socio-cognitive skill that relates to evolutionarily derived pattern” (Bard et al. in intersubjectivity and appears to have a proto- press). Since apes do not seem to differ from mimetic origin, at least in human children, is humans in the capacity to perform this shift (and attention. Reddy (2001, 2003, 2005) has argued possibly even transmit it culturally to their de- that prior to awareness of the other’s attention scendents), this confirms the conclusions from to an external object and much prior to joint neonatal imitation that there is no major differ- attention appearing around 12 months (see be- ence between the species on the proto-mimetic low), children “show an awareness of others as level. attending beings, as well as an awareness of self as an object of others’ attention” (Reddy 2003: 357), displayed in phenomena such as eye- 5.2 Dyadic mimetic intersubjectivity contact, intense smiling, coyness, ‘calling’ vocali- zations, showing-off etc. Since awareness of self In our previous discussion, we proposed that (at this stage) is largely proprioceptive, while proto-mimetic intersubjectivity, without a clear awareness of the attention of others (in seeing differentiation between self and other is based children) is mostly based on vision, this satisfies on the (mostly) unattended body schema and our first criterion for bodily mimesis. Reddy’s similarly unattended mechanisms for “body claim is that such dyadic (though not dyadic- copying”. On the other hand, what we here refer mimetic) interactions underpin later developments to as dyadic mimetic intersubjectivity is based on the in intersubjectivity. Evidence for this is the ob- conscious control of the movements of one’s servation that autistic children show difficulties body and attention to their correspondence to even with such simple interpersonal engage- the body of another, whereby one can imagine ments, “suggesting that whatever is going on in what the other experiences on the basis of one’s dyadic attentional engagements may indeed be own experiences in similar circumstances. In the critical, not just as a source of information and terminology of Gallagher (1995, 2005), we pro- experience about attentional behavior, or as a pose that the role that bodily mimesis proper scaffold for the subsequent development of plays for the development of intersubjectivity 16 Bodily mimesis as “the missing link” in human cognitive evolution implies not the body schema but the body image. This supports the interpretation that cogni- While the body schema and the body image tive empathy involves a more sophisticated rep- normally interact, Gallagher (2005) shows how resentational capacity than what is necessary for in certain pathologies they can be disassociated. simple empathy. Since dyadic mimesis involves The distinction between the dyadic form of both the ability to identify with the other, and at mimetic intersubjectivity, described here, and the the same time to differentiate between self and triadic one described in the following sub-section other, our (Vygotskyan) hypothesis is that it is is that while the dyadic form involves the prop- dyadic mimesis, implicated in e.g. imitation (Sec- erties of volition and representation (see Section tion 4) that scaffolds the development of such 3), e.g. in the case of recognizing oneself in a representational capacity. mirror or in shared attention, there is no under- Since dyadic mimesis allows to “place oneself standing of communicative sign function. Thus there in the shoes of others”, it also gives the oppor- is no basis for intentional communication on the tunity to understand what someone else is attend- basis of shared representations. As we shall see ing to. Such “second-order attention” is well here and in the discussion of ape gestures in testified among great apes (Hare et al. 2000). Section 6, it is the latter that appears to be most When two individuals become aware that the difficult for non-human primates. Below, we other is attending to the same object, what re- briefly describe how understanding others’ emo- sults is shared attention. This comes a good deal tions, attention and intentions can be seen as towards the construction of a “consensual real- intimately related to dyadic mimesis. ity” that can be communicated about, but does Whereas (simple) empathy is proto-mimetic, not quite reach it. To make a given object X fully what Preston and de Waal (2002) call cognitive intersubjective between you and me, I would empathy requires a differentiation between subject need not only to “see that you see X”, (second- and object where “the subject is thought to use order attention, see Figure 1a), but also “to see perspective-taking processes to imagine or pro- that you see that I see X” (third-order attention, ject into the place of the object” (ibid: 18). Evi- see Figure 1b) and vice versa – which is our dence that this is not an isolated phenomenon, interpretation of what it means to engage in joint but shows a more advanced level of intentional- attention. ity is the fact that cognitive empathy “appears to emerge developmentally and phylogenetically with other ‘markers of mind’ … including per- spective taking …, mirror self-recognition …, deception, and tool-use.” (ibid: 18). Research concerning cognitive empathy in apes has fo- cused on their consolation behavior, which is well attested in at least chimpanzees, but has not been found in monkey species (de Waal and Aureli 1996) or any other mammalian species. Consolation is cognitively more complex than simple empathy since the consoling individual not only feels that somebody else experiences a Figure 1a. Shared attention: Second-order attention: “I negative emotion, but also intends to help relieve see that you see X” (and vice versa) this, implying an ability to imagine the more positive emotional state.2

2 An interesting question is what goes on in the mind of an ape that is being consoled: If one could identify cognitive processes of the form “I can feel that you feel that I feel bad”, this would be equiva- lent to “third order emotion”. However, given the lack of methodology for testing the existence of such a process, we feel complelled to apply Oc- cam’s razor and ascribe only second order emotions of the form “I feel that you feel”. Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 17

enculturated and language-naïve apes are capable of this.3 However, it has not been shown that (non- enculturated) apes are capable of understanding another’s mental states about their own mental states, which would involve, again, third-order mentality.4 As suggested earlier, joint attention also requires third-order mentality, and is simi- larly something that apes seem to find difficult. In natural settings, some cases of deception may be interpreted in a way to involve third- order mentality, but do not require this. For Figure 1b. Joint attention: Third-order attention: “I see example de Waal (1998 [1982]) describes the that you see that I see X” (and vice versa) chimpanzee Yeroen who has had a fight with Full joint attention thus involves third-order Nikkie, and continues to fake a limp only when mentality and (possibly because of this, see be- in the presence of Nikkie, apparently in an act of low) appears to be beyond the cognitive capaci- wishing to provoke Nikkie’s empathy: “I wish to ties of apes (Tomasello 1999). It also goes be- make you see that I hurt”. The more parsimoni- yond dyadic mimesis, so we will leave this ous explanation, however, is that Yeroen has capacity for the time being, but return to it in the learned from previous experience that he is not following subsection, since there is a close con- bothered by Nikkie when he is hurt: “He may nection between communicative intention and have learned from incidents in the past in which the emergence of joint attention. he is seriously wounded that his rival was less Concerning the understanding of another’s hard on him during periods when he was (of intentions, it has been for sometime a predomi- necessity) limping” (ibid: 35-36), so he mimes the nant opinion that apes cannot do this, and in- appropriate behavior. Here we have a clear cor- deed this was one of the central claims of respondence between dyadic mimesis and sec- Tomasello (1999). Recently, however, there has ond-order intersubjectivity. Notice that Yeroen’s been mounting evidence that (at least) limping was not a form of intentional communi- “[c]himpanzees understand psychological states cation – he did not wish Nikkie to understand – the question is which ones and to what ex- that he was faking a limp – if he did, that would tent”, which is the title of Tomasello, Call and be a clear case of triadic mimesis. Hare (2003). In the experiments behind this Another possible instance of third-order claim a subordinate and a dominant chimpanzee mentality is that reported by Woodruff and competed for food placed on the subordinate’s Premack (1979) in which chimpanzees learned to side of two barriers, so that in some cases only suppress glances and body orientation toward the subordinate, but not the dominant could see the place where food was hidden, when a trainer the food. The subordinates could also monitor who would not share the food was present. the visual access of their dominant competitor. However, since it took a long number of trials The results showed that the subordinates prefer- entially retrieved the food that dominants could 3 While it still not conclusively shown that apes are not see, suggesting that chimpanzees understand capable of such mental “projection”, and it is con- what con-specifics have and have not seen and ceivable how the evidence can be explained in a even have some awareness of the dominant’s goal more behavioristic manner involving learning gen- (i.e. “a mental representation of a desired state”, eralizations over other individuals’ behavior in Tomasello et al. in press) of obtaining the food. relation to food, we contend that the mental expla- nation is (a) ultimately more parsimonious and (b) This does not have to be done on the basis of consistent with other mental skills in apes, as well explicit reasoning or inference, but possibly as in human children in roughly comparable stages through the “projection” of one’s own motiva- of development. tional state in a similar situation onto the other 4 Note that when we in this section link triadic mi- (“I would get the food if I were in his place!”), in mesis and third-order mentality, we do not imply a dyadic mimetic fashion. Apparently, even non- that an agent capable of triadic mimesis can handle all forms of third-order mentality, in particular those which involve beliefs. 18 Bodily mimesis as “the missing link” in human cognitive evolution for the apes to learn this behavior, a simpler “Machiavellian intelligence” (Byrne and Whiten explanation involving conditioning is possible. 1988), this puts the focus on cooperation rather The conclusion that we can draw from these than competition (see also Brinck and Gärden- various examples is that wild apes as well as fors 2003; Tomasello et al. in press). A predic- those who are exposed to different degrees of tion from this hypothesis is that enculturated human contact (captive, nursery-raised and labo- apes – and these have all been taught at least ratory-trained apes), but are not raised in a some communication through signs – will de- “something like a human cultural environment” velop higher-level skills of intersubjectivity. and thus enculturated (Call and Tomasello 1996) There is some support for this prediction. In can indeed understand second-order mentality. summarizing some 200 studies of the role of However, such apes do not seem able to under- human influence, Call and Tomasello (1996) stand third-order mentality – neither in the do- conclude that “[t]he sociocognitive domains in mains of emotion, attention nor intention – in which humans seem to have the highest effect which their own mental state needs to be either on apes are intentional communication and so- intentionally communicated – in collaboration, cial learning” (ibid: 391). or hidden – in competition. This corresponds In Section 6, we will focus more specifically well with the capacity of apes for dyadic mime- on the understanding and use of communicative sis, but their relative difficulty with triadic mime- gestures by apes. Here and in the following sub- sis, which we discuss further below. section, we will mostly review the evidence that sign use in general, and the acquisition of (as- pects of) language in particular is the major fac- 5.3 Triadic mimetic intersubjectivity tor behind the development of higher forms of intersubjectivity that may deserve the label “the- In the case of triadic mimesis, there is by defini- ory of mind”. tion not only an understanding of representation As pointed out earlier, wild apes do not seem itself, but an understanding that representations to be capable of engaging in full, third order can be used communicatively, or what we refer joint attention. Furthermore, as Tomasello to as communicative sign function. This implies an (1999: 21) points out, wild apes (1) do not point understanding that the sign has the same meaning to objects; (2) do not hold up objects to show for the addressee as for the sender. This involves at them to others; (3) do not take someone along to a least second-order mentality, which was shown place to show them something; (4) do not ac- above to correlate with dyadic mimesis. But tively offer something to someone; and (5) do not “having the same meaning” is a reflexive notion intentionally teach other individuals new behav- and this implies at least some degree of third- iors. Tomasello’s original account of these ab- order mentality. Consider the simple example of sences was based on the claim that apes are un- what knowing the meaning of the word cat im- able to understand other’s intentions. Given the plies: more recent evidence, this explanation is no a) I know that cat means “a small furry ani- longer tenable, and indeed Tomasello et al. (in mal that meows”. press), suggest instead that the crucial difference b) I assume you know that cat means “a between apes and humans involves the motiva- small furry animal that meows”. tion to participate in joint collaborative engage- c) I assume that you know that I know that ments, and the lack of this motivation prevents cat means “a small furry animal that me- apes from constructing “dialogical cognitive ows”. representations”. While it is possible for intentional communi- Our explanation is similar but more specific: cation to begin without a full realization of (c), it we believe that non-enculturated apes fail to is practically inevitable that discursive experience develop the communicative sign function, and (including failures in communication) will pro- related to that, the ability to engage in third-order mote the development of third-order mentality. mentality. The motivational difference between Thus, it is possible that it was sign use itself that apes and humans appealed to by Tomasello et al. was the major driving force behind the devel- (in press) cannot be the full explanation since opment of intersubjectivity in hominid evolu- enculturated apes such as Koko, Kanzi and tion. Unlike competing hypotheses related to Chantek manage at least (1) and apparently Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 19 communicative skills (2-5) above as well (Miles par with all other types of discriminative cues 1999), even if in restricted forms. This seems to that have to be laboriously learned over re- imply that the human cultural environments5 of peated experiences. The children, in contrast, the enculturated apes have taught them the ba- treated each communicative attempt as an ex- sics of intentional, sign-mediated communica- pression of the adult’s intention to direct their attention in ways relevant to the current situa- tion, and thereby (the roots of) third-order men- tion. [our emphasis] tality. How this could occur can be seen again with In other words, while the children clearly under- respect to joint attention, which can be seen as stood the communicative intentions of the ex- emerging from second-order attention combined perimenter, the apes did not. This interpretation with the recognition of another’s intention con- is supported by a similar experiment designed to cerning my attention: “I see that you see X” test “false beliefs” (Call and Tomasello 1999), in (second-order attention, Figure 1a), and fur- which the enculturated and language-taught thermore “I realize that you want me to look at orangutan Chantek clearly performed differently X” (see Brinck 2001). In other words, joint at- from all the other apes in understanding a hu- tention can be brought forth by understanding a man communicator’s signals. Even though this simple form of communicative intention, com- was not the goal of the experiment, and Chantek bined with already exisiting second-order atten- did not score better than the other apes in the tion. Thus, communicating the intention to false beliefs task, his much better performance jointly attend may be said to involve the simplest could be explained by considering that he under- kind of triadic mimesis: whatever kind of behav- stood the signals as communicative signs (in this ior that is used to convey that intention – some case indexes), rather than as “discriminative form of index (see the example in the next para- cues”. graph) – can be said to stand for that intention for Finally, we should mention the case of cap- both sender and interpreter. tive apes living in a zoo and thus involved with Without enculturation, experiments indicate at least some degree of interaction with human that apes do not understand communicative culture. In their study of spontaneous gestural intentions. A rather typical example is an ex- communication in a group of gorillas in the San periment by Tomasello, Call and Gluckman Francisco zoo, Tanner and Byrne (1996, 1999), (1997), where the authors in different ways indi- found a wealth of gestures used by several cated for both chimpanzees and two- to three- members of the group, in particular by the adult year-old children which out of three containers male Kubie. We return to this is Section 6, but it contained a reward: by pointing to the correct suffices here to note that the gestures seemed to container; by placing a marker on top of the be used in a communicative way so that: correct container; and holding up a replica of the [w]hether the receiving partner was a human correct container. Tomasello (1999: 102) sum- or another ape, the signaling ape made sure marizes the results of the experiment as follows: that visual contact was established (except for Children already knew about pointing, but tactile close gestures), and seemed to under- they did not know about using markers and stand both the other’s potential actions and what the replicas as communicative signs. They never- partner might, in turn, understand from his (the sig- theless used these novel signs very effectively naler’s) performance of gestures. (Tanner and Byrne to find the reward. In contrast, no ape was 1999: 231, [our emphasis]) able to do this for any of the communicative signs We can conclude that a triadic from of mimetic that they did not know before the experiment. intersubjectivity, involving understanding not One explanation of these results is that the only con-specifics’ intentions, but their commu- apes were not able to understand that the human be- nicative intentions, and consequently a degree of ings had intentions toward their own attentional states. The apes thus treated the communicative at- third-order mentality, appears to be not completely tempts of the human as discriminative cues on beyond the cognitive potential of apes. To realize this potential, they need an environment that is rich in opportunities for developing the communica- 5 Consisting of what Wittgenstein (1953) called the tive sign function, i.e. a particular form of encul- “forms of life” that provide the necessary context turation. Thus triadic mimesis may be said to be for the emergence and functioning of intentional within apes’ “Zone of Proximal Development” communication and language. 20 Bodily mimesis as “the missing link” in human cognitive evolution

(ZPD) – the notion introduced by Vygotsky Mutual knowledge is often stated as a form of (1978) to refer to skills that children could ac- “third-order belief”, but this is potentially mis- quire with the help of adults, but not alone. leading since it is not necessary that the under- If “enculturation” provides the ZPD for pre- standing on all three orders of mentality is ex- sent-day apes, it is reasonable to suppose that it plicit enough to be a matter of “belief”, i.e. a did the same for some particularly social group propositional representation that is actively held of hominids through a form of “self- to be true. Consider again the three orders of domestication” giving rise to a bootstrapping knowing the conventional meaning of cat men- spiral of sign use and intersubjectivity. In the tioned earlier: I assume that you know that I terms of Donald (2001), triadic mimesis must know that cat means “a small furry animal that have been within the common ancestor’s “zone meows”. The highest order, my assumption that of proximal evolution”. you know that I know, is not properly speaking a belief for the 4 year old child, since it is taken for granted, without pondering on whether it is true 5.4 Post-mimetic intersubjectivity or not. At the same time, children at this age become (a) capable of understanding that others What differentiates post-mimetic, or symbolic, lack knowledge or have “false beliefs” (e.g., cognition from mimesis is the use of fully conven- Perner 1991; Wimmer et al. 1988; Gopnik and tional signs, interrelated within a system (Deacon Graf 1988; Mitchell 1997), implying metarepresen- 1997; Zlatev 2003). The most obvious example tational capacity, and (b) at least somewhat fluent of post-mimesis, involving all the previous fea- in their first language. It appears that these two tures but also symbolicity is a conventional, insti- developments are closely connected, and that tutionalized signed language such as ASL (Sto- acquiring a language, spoken or signed, is a ma- koe 1960) or Swedish Sign Language (Ahlgren jor causal factor for developing a fully-fledged 2003). What is the relation between acquiring “theory of mind”. Despite successes in under- such a system and intersubjectivity? standing other’s attention and even intentions, A convention (Lewis 1969; Clark 1996) or a apes, including enculturated ones, have so far norm (Itkonen 1978) exists as a form of mutual consistently failed so-called “false belief tests” (or common) knowledge among the members of the (Premack 1988; Call and Tomasello 1996, 1999; group that share the convention. A common Povinelli 2000). They have also failed in acquir- explication of mutual knowledge is that it con- ing a full human language – despite successes in sists of third-order knowledge: “I know that you acquiring certain aspects of it. Three different know that I know X”. Translated to the capacity sides to language (use) combine to promote to understand the minds of others, common metarepresentational capacity. knowledge amounts not only to my understand- First, as mentioned above, language is a con- ing the mentality of others, but also to my un- ventional symbolic system, and as such its mas- derstanding that others (can) understand my tery implies third-order knowledge, which would mentality. This has been expressed by Itkonen carry with it training in the understanding of (1978: 96) as follows: others’ beliefs. Second, two specific (universal) [P]erson A cannot know that he is doing or features of human are (a) mental thinking X, and thus cannot do or think X, predicates such as “think”, “believe”, “know”… unless he is able to know what it is for some and (b) sentential complement constructions other person B to do and think X. The same such as “say that…”. If one can meaningfully is true, in turn of B’s knowledge of his own formulate sentences such as “I think that you actions with respect to A’s (possible) actions think that X”, then one should be able to think and thoughts. Hence it can be shown on the corresponding thought. Third, as pointed out purely conceptual grounds that A must be by Tomasello (1999), not just the logical struc- able, in principle, to identify B’s various men- ture of language, but its use in discourse would tal states or processes, and vice versa. This promote the understanding of others as “mental means that mental states and processes exist agents”: There are at least “three kinds of dis- only at the level of common knowledge i.e. of a course, each of which requires [children] to take common ability to identify them, wherever they occur. the perspective of another person in a way that goes beyond the perspective-taking inherent in Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 21 comprehending individual linguistic symbols and periment the enculturated and language-taught constructions.” (ibid: 173): disagreements, re- orangutan Chantek performed differently from pairs/explanations and meta-discourse. the other apes, displaying a better understanding Empirical evidence for the existence of a of the experimenter’s signals as communicative strong connection from language to the under- signs. A prediction from our analysis would be standing of beliefs has been accumulating during that if Chantek, or any of the other “langauge recent years and involves e.g. the following: apes” that have been the subject of so much controversy, could progress in their language • Deaf children who are not exposed to sign development to involve more mental terms, as language at an early age pass false belief tasks well as grammatical constructions involving significantly later than those who are (Peter- sentence complementation, they would also be son and Siegal 1995); able to pass false belief tasks. • There are correlations between parent use of A counter-example to this Vygotskyan hy- mental predicates in their child-directed pothesis of interpersonal communication pre- speech and the children’s performance in ceding intrapersonal cognition would be to show false belief tasks (de Villiers and Pyers 1997); that non-enculturated apes are capable of under- • Longitudinal studies indicate that language standing beliefs (or at least third-order mental- development predicts false belief task per- ity), even outside of communicative settings. A formance, but not vice versa (Astington and candidate domain is again deception, but as Jenkins 1999); mentioned in Section 5.2, the kind of examples • Training in (non-mental predicate) sentential involving deception, usually provided in support complement constructions significantly im- for theory-of-mind skills (e.g. Yeroen’s fake proves performance on false belief tasks limp) can be explained by simpler skills. A fur- (Hale and Tager-Flusberg 2003; Lohmann ther candidate is counter-deception. In a study per- and Tomasello 2003). formed with chimpanzees by Menzel (1973, • Training in perspectival discourse alone 1974), only the female Belle was allowed to see (without sentential complements or mental where food was hidden within a large enclosed predicates) contributes to performance on area, and subsequently she led the group to the false belief tasks (Lohmann and Tomasello food, and all were able to share it. When the 2003). experiment was repeated, however, the male

Rock began to take all the food for himself. Our conclusion is therefore that the understand- After this, Belle did not reveal where the food ing of (false) beliefs is a form of post-mimetic was hidden if Rock was nearby. She sat further intersubjectivity, in the sense that is based on and further from the place where the food was language, either spoken or signed. Notice that hidden, and it was only when Rock was looking this does not contradict analyses by e.g. Bloom in another direction that she went to the hiding (2000) that the acquisition of language presup- place. Rock, however, countered this by pretend- poses “theory of mind” skills, since we have ing to go away and then turning just as Belle was argued that the latter, including joint attention about to reveal where the food was. Byrne and communicative intention are triadic mimetic (1995) interprets this as implying that Rock has phenomena, which also in our analysis are pre- an understanding that Belle is trying to deceive requisites for language. him, i.e. third-order mentality. However, In one of the relevant studies, Call and Tomasello and Call (1997) note that the experi- Tomasello (1999) used a non-verbal false belief ment with all its steps lasted several months so task with chimpanzees and orangutans as well as Belle and Rock may have successively learned with human children. The main results were that how to predict the behaviour of the other with- the children's performance on verbal and non- out understanding deceptive intentions. This verbal false belief tasks was highly correlated, hypothesis is supported by a later experiment supporting the hypothesis of a possible causal with mangabey monkeys performed by Coussi- connection. At the same time, no ape could pass Korbel (1994), who replicated Menzel’s study, the nonverbal false belief task even though they but performed a detailed day-to-day analysis. succeeded in all of the control trials indicating One of the informed mangabeys, step by step in mastery of the general task demands. However, as mentioned in Section 5.3, in this latter ex- 22 Bodily mimesis as “the missing link” in human cognitive evolution four days, learned the strategy of leading her object) by recognizing my intention that you do main competitor away from the food. this” from the sender’s perspective and “I un- derstand that you want me to do X” from the recepient’s. 5.5 Summary This is not something that comes easily to apes in natural conditions, but through encul- In this section we have tried to show that there turation and especially through extensive sign is a close connection between different levels of use, some understanding of communicative in- intersubjectivity and the different levels of the tentions seems to be within the reach of apes’ mimesis hierarchy, as summarized in Table 4 “Zone of Proximal Development”, even though below. in may be in its periphery. Evidence for this is the relative mastery of joint attention by encul- Table 4. The mimesis hierarchy and intersubjectivity turants, and as argued this can be seen to origi- nate in (dyadic mimetic) second-order attention Level Intersubjectivity skills Type of mentality combined with the understanding of the other’s Proto- (simple) empathy 1st order: lack of intention that I attend. Finally, post-mimesis, mimesis complete differentia- mutual attention with its features of conventionality and sys- tion between self tematicity necessarily brings with it sufficient and other understanding of third-order mental states, and Dyadic Cognitive empathy 2nd order: under- (at least) second-order beliefs, e.g. “I think that mimesis standing the other Shared attention you know (or don’t)”. The latter have been through “projection” shown in numerous false belief tasks, which are Understanding (identification, but regularly failed by apes. On the other hand, there other’s intention differentiation) (in competitive is mounting evidence that the acquisition of contexts) language, with its structural properties and dis- cursive use significantly contributes to higher Triadic Joint attention 3rd order (attention mimesis and intentions) forms of “theory of mind” development. Communicative In summary, in this section we have argued intentions that bodily mimesis, in its proto, dyadic and Post- (False) belief un- 3rd order (beliefs) triadic forms, is a major factor in the develop- mimesis derstanding ment of intersubjectivity. Sign use itself was suggested to be a driving force in the develop- Proto-mimesis is crucially implicated in mutual ment of an understanding of third-order mental- attention and the awareness of others’ feelings, ity, and the performance of enculturants shows through a species-general capacity for empathy that this achievement is within the reach of apes. that has possibly been further developed in the Therefore one can conclude that (mimetic) sign “ultra-social” species Homo sapiens. Dyadic mi- use was possibly within the “zone of proximal mesis leads to the ability to map between one’s evolution” (Donald 2001) of the common ape- own body and that of others, in a more de- human ancestor, considerably more so than tached, differentiated way, and in this way un- fully-fledged language, characterised by full con- derstand others’ emotions, i.e. cognitive empa- ventionality and systematicity. thy, shared attention and even intentions Finally, while we have focused on the mime- through a conscious process of “projection”: sis-to-intersubjectivity direction of causality, this what would I see/feel/wish if I were you. Unlike does not mean that there is no “feedback” link earlier claims to the contrary, newer evidence and thus we assume a co-evolutionary scenario. and analyses show that apes do not have much It was implicit in the discussion that achieving difficulties with this level and that they have the different levels of intersubjectivity is necessary capacity for second-order mentality. for moving to still higher levels on the mimesis The crucial step in the evolution of intersub- hierarchy, e.g. joint attention for the acquisition jectivity involves triadic mimesis, implying hav- of language. It is also possible that in the present ing and understanding others’ communicative analysis we have overemphasized the Vy- intentions, which requires third-order mentality: gotskyan principle of the “interpsychological “I want you to do X (e.g. share attention on an preceding the intrapsychological” (see Section 1), Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 23 but this has come rather naturally by defining the omy), but none, to our knowledge, that has developmental levels or stages in terms of bodily explicitly taken an evolutionary perspective – mimesis, with its originally interpersonal charac- apart from the analysis of the origins of pointing, ter. see below. Our goal in this section is to apply the mimesis hierarchy to gesture, and in showing how different types of gestures correspond to 6. Mimesis and gesture the different stages of the mimesis hierarchy, suggest such an evolutionary taxonomy. We will Of the the three socio-cognitive domains we also show how the four stages of the hierarchy discuss in this article, gesture is the one most cross-cut with two basic semiotic distinctions: obviously related to bodily mimesis: at least indexicality and iconicity, which differ with respect some gestures posses all the properties empha- to the semiotic ground (Peirce 1931-1935; Sonesson sized by Donald (1991): reference, intentionality, 1989) for the relationship between the “sign communicativity, autocuing and generativity (see vehicle” (the expression) and the meaning of the Section 3). However, not all bodily acts that are gesture: called “gestures” have these properties. Fur- • Indexical gestures, where the ground is one thermore, in the currently highly active field of of spatio-temporal contiguity, the prototypi- gesture studies (e.g. Kendon 1981; McNeil 1992; cal case being pointing. Goldin-Meadow 1999; Kita and Özyürek 2003), • Iconic gestures, where the ground is of studying among other things the spontaneous similarity, with the prototype pantomime. co-speech gesturing of human children and adults, it is often pointed out that gestures and Figure 2 illustrates the two types schematically: language interact in complex ways. Thus, not all what is essential for indexical gestures, as with all gestures are easily subsumed under the label indexical signs, is that expression and content “non-linguistic” or “non-verbal”. In such a case, are closely related in time-space. An iconic ges- it becomes questionable whether they are truly ture (sign), on the other hand, requires similarity mimetic rather than post-mimetic. (often of a quite schematic type) but on the Perhaps in the most general sense gesture has other hand it can be removed in time-space from been defined as “a functional unit, an equiva- their meanings. lence class of coordinated movements that achieve some end” (Armstrong, Stockoe and Wilcox 1995: 46; Wilcox 2004: 44), which en- compasses communicative as well as non- communicative body movements as long as they posses the property of goal-directedness. This is much too general for our purposes and we will consider as gestures only goal-directed communicative body movements, i.e. such that require interpretation from an audience for achieving the gesturer’s goal. This in- cludes a variety of acts such as attention getting, pointing and miming. At the same time, our Figure 2. Indexical signs, based on contiguity vs. iconic definition excludes bodily signals such as laugh- signs, based on similarity. In both case expression and ter, which is not goal-directed, or acts that content need to be differentiated and perceived as standing in a representational relationship by an in- achieve their ends through “mechanical causal- terpreter for there to be a (genuine) sign. Without ity” rather than interpretation, e.g. pulling some- this, there are only indexicalities and iconicities. one, as opposed to beckoning him. Notice that while we focus on communication, we do not In agreement with Sonesson (1989, in press), we require all gestures to be intentionally communica- distinguish between iconicity and indexicality as tive: what is essential is that the bodily motion is a type of semiotic ground, and the sign function interpreted by an observer as revealing some- (Piaget 1945). The latter presupposes an inter- thing about the state of mind of the gesturer. preter (signified by the “happy face” in Figure 2) There are numerous classifications of ges- who differentiates between the expression and tures (see Kendon 1981 for a well-known taxon- the content and is furthermore aware of the 24 Bodily mimesis as “the missing link” in human cognitive evolution representational relationship between them: Furthermore, ape mothers are less “intervention- Only in this case do we have iconic and indexical ist” and less likely to oblige their infants, even signs. In the case of gestures, this means that only when the latter do reach out to distal objects if the gesturing individual is aware of the expres- (Bard and Vauclair 1984). In general they inter- sion-content structure of his gesture are the fere only when they deem that there is a risk of gestures intentionally used as signs. If not, the their offspring getting in harm’s way. Such dif- gestures are only a matter of indexicality or ico- ferences may be the major reason why feral apes nicity for the gesturer; however, they may still be point much less (if at all) than human beings, seen as signs by an interpreter, as is the case with since they are quite capable of acquiring pointing the first types of gestures discussed below. gestures when in contact with humans (see be- low).

6.1 Proto-mimetic gestures 6.1.2 Proto-mimetic iconicity: phylogenetic ritualization We have defined proto-mimesis as involving a cross-modal mapping between exteroperception Phylogenetic ritualization is often proposed as and proprioception, but lacking (clear) differen- the basic mechanism for the emergence of ani- tiation and or/volition. From this perspective mal signals, as explained by e.g. Knight (2002: the spontaneous reaching behavior of the child 147): observed soon after birth (von Hofsten 1983) Over evolutionary time, certain aspects of and various forms of phylogenetic ritualization non-communicative behaviour assume a sig- may be classed as proto-mimetic (for the ges- nalling function, becoming correspondingly turer), involving indexicality and iconicity, re- specialized through a process known as ‘ritu- spectively, but not the sign function. alization’. Signal evolution begins when others read some aspect of normal behaviour as sig- nificant. If the subject of such surveillance can 6.1.1 Proto-mimetic indexicality: reaching benefit from having its mind read or its be- haviour anticipated, then over evolutionary In the first 6 months of life, the human child time natural selection will accentuate the cues, coordinates its hand movements and reaches reducing any ambiguity. By definition, such phylogenetic ritualization entails special elabo- towards interesting objects, including towards ration and added costs. such which it desires but cannot grasp because they are too far away. At the same time there is Gestures such as gorilla chest beating (and other no gaze alternation between the object and a forms of display) and the “submissive gesture” third party, and no attempt to get the other’s of apes, in which they crouch and “make them- attention – and possibly help – in obtaining the selves small” involve iconicities (resemblances) to object. Clearly this is not a communicative ges- body size – the larger the animal is, the louder ture from the point of view of the child. How- the signal – and state of agitation – the stronger ever, at least in typical Western caregiver-child the emotion, the more beating. At the same interactions at this period of development, the time, they are not iconic signs, since the similarity adult often interprets such reaching as commu- between the signal and the non-communicative nicative and offers the desired object. This ap- behaviour is not perceived (at least not necessar- pears to have an effect on the child’s interpreta- ily) by either the gesturer or the interpreter. tion of its own action through a process of Hence, while there is communication, there is no internalization of the interpersonal to intraper- representation (in the sense of expression- sonal (Vygotsky 1978, see Section 1). Eventually, content structure, see Ikegami and Zlatev in this transforms reaching into (proto-) imperative press) involved. That is why we can call such pointing, which, as discussed below, is a dyadic phylogenetically ritualized signals proto-mimetic mimetic gesture. gestures. Ape infants reach out to objects that attract It is characteristic that when observed in their attention similarly to human children, but highly enculturalted apes such as the signed lan- less so to distal objects due to the fact that they guage taught gorilla Koko, gestures such as become capable of locomotion much sooner. “gimme” and “pound” were the first to develop, Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 25 with the least effort on the part of the trainer. As Hence, imperative pointing can be classified as a pointed out by Patterson (1980: 517): dyadic mimetic gesture. This conclusion is also These were distinctive because they appeared supported by Tomasello’s (1999) observation without direct training at a time when the that in this stage of development, children can great majority of Koko’s signs were being ac- master intentional pointing without understand- quired by molding. Because other young goril- ing the pointing of others. Another kind of las reared in captive conditions who have not pointing that is perhaps even more obviously been exposed to sign language use these same dyadic mimetic that is also observed at this stage gestures, they must be part of the gorilla’s is “pointing for oneself”, i.e. when the child is natural repertoire. using pointing gestures to direct its own atten- tion, rather than that of others (Bates, Camaioni 6.2 Dyadic mimetic gestures and Volterra 1975). When it comes to apes, it was earlier thought In the case of dyadic mimesis there is representa- that apes do not point (Donald 1991) or at least tion in the sense that there is both differentiation not with the index finger (Povinelli and Davis and correspondence between the gesture and its 1994). This has been recently shown to be un- “meaning”, in this case the action that the ges- true. Leavens and Hopkins (1999) summarize turer wishes that the addressee will perform. In some of the evidence showing that pointing has this respect, such gestures constitute signs, in the been reported for 15 captive monkeys, 83 cap- sense of the term defined at the beginning of tive apes, and even one feral ape, a who this section. However, they have two major limi- was pointing at a group of humans while looking tations: (a) their intended meaning is always a back and forth between the humans and the 6 desired action rather than a statement, i.e. they other bonobos (Veà and Sabater-Pi 1998). are imperative rather than declarative, and (b) Whether or not feral apes really point, the evi- they involve second-order mentality: a claim on dence from captive apes (e.g. Call and Tomasello the attention and behavior of the addresse, but 1994; Leavens, Hopkins and Bard 1996, Krause not third-order mentality: there is no appeal on and Fouts 1997; Leavens and Hopkins 1999) the addresse to understand the gesturer’s own clearly shows that the criteria for imperative mentality. In our terminology, they do not in- pointing are satisfied: attention getting, gaze volve the communicative sign function, unlike alternation and persistence until the goal is met. the triadic gestures that we discuss in 6.3. Leavens and Hopkins calls these audience effects, and observe that captive chimpanzees engaged in imperative pointing exhibit more audience 6.2.1 Dyadic-mimetic indexicality: imperative pointing effects, and above all more gaze alternation than children at the corresponding stage of develop- In the human infant literature “reaching” is of- ment. At the same time, the children were con- ten used as a label for acts of intentional com- siderably more prone to vocalize while pointing, munication, rather than to attempts att prehe- showing the close association between bodily sion. Similarly to Leavens and Hopkins (1999), mimesis and speech in human beings. in their informative review of studies of pointing However, if we limit our attention to the in children and apes, we find that even if the manual gestures alone we can conclude that at child does not shape the hand and arm in a “ca- least when exposed to human contact, apes nonical point” (Butterworth 1998), given that clearly display imperative pointing, requiring there is evidence for gaze alternation between an dyadic mimesis. Simarly, there is evidence that at object and a third party, the act should be re- least enculturated apes engage in dyadic, non- garded as an instance of pointing. In human communicative pointing in examining objects children, such acts emerge sponatenously from 8 for the purpose of directing their own attention months of age, and at least at first have an “im- perative” function of attempting to make the ad- 6 If it was not for the somewhat anecdotical charac- dressee perform some desired action. From this it fol- ter of this piece of evidence, it would pose a prob- lows that imperative pointing requires second- lem for our classification, since it appears to in- order mentality, but not necessarily joint atten- volve declarative pointing, which is otherwise tion and third-order mentality (see Section 5.2). observed only in enculturated apes.

26 Bodily mimesis as “the missing link” in human cognitive evolution

(Bonvillian and Patterson 1999). Interestingly an effect on the addressee. However, this inten- such non-communicatieve pointing emerged tion to communicate is not the same as a commu- earlier than communicative pointing (of both the nicative intention, which is an intention that the imperative and the declarative type) in the sign desired effect will be produced by the addressee language taught gorilla Koko, similar to the case recognizing the intention to produce the effect with children (Bates, Camaioni and Volterra (Grice 1957; Sperber and Wilson 1995). As ar- 1975). gued in the Section 5, such forms of third-order mentality appear to be beyond the capacity of non-enculturated apes. 6.2.2. Dyadic mimetic iconicity: ontogenetic ritualization At the same time, to the extent that the ges- turing ape is differentiating between its (onto- As pointed out in 6.1, most animal signals are gentically ritualized) gesture and the desired the result of phylogenetic ritualization, and are in effect and being more or less aware of the corre- this sense innate. But monkeys and especially spondence relation between them, it is justified great apes use a good number of learned ges- to regard the two as standing in a representa- tures as well. Tomasello et al. (1994) distin- tional relationship – even without involving a guished six functional categories of gestures in communicative sign function. However, this infant and juvenile chimpanzees involving nurs- representational relationship is fairly unstable. ing, walk-riding, grooming, eating, appeasing and Tomasello et al. (1994) showed that chimpanzes play/tickle. These were believed to have come had a limited arsenal of gestures that they ap- about through a process called ontogenetic ritualiza- plied highly “flexibly”: different gestures were tion in which the gestures were shaped by re- used for the same goal, and the same gesture was sponding to each other’s anticipations. This often used with different “meanings”. On the happens over the life of the individual rather one hand, this speaks for these gestures being than the species, hence the term given to this intentional and different from species-specific learning process. Tomasello and Zuberbühler signals. But on the other hand, it testifies to their (2002: 295) describe ontogenetic ritualization as being non-conventional, and non-systematic. a sequence of 4 steps: Similar findings have been replicated for young gorillas (Pika, Liebal and Tomasello 2003) (1) Individual A performs behavior X. and bonobos (Pika, Liebal and Tomasello 2005). (2) Individual B reacts consistently with behavior However, in these more resent (and systematic) Y. studies, a number of gestures where found to be (3) Subsequently B anticipates A’s performance of group-specific, and therefore “some form of X, on the basis of its initial step by performing Y. social learning” (Pika et al. 2003: 108), besides (4) Subsequently A anticipates B’s anticipation ontogenetic ritualization was considered to be and produces the initial step in a ritualized form the mechanism behind their emergence. Fur- (waiting for a response) in order to illicit Y. [original thermore, while Tomasello et al. (1994) did not emphasis] observe any social transmission of a specific begging gesture learned by a chimpanzee to This way of describing the emergence of such acquire rewards from human experimenters, gestures sounds rather behavioristic, implying Pika et al. (2005) summarize a number of differ- that it is only a matter of (co-)reinforcement of ent studies finding evidence for the social learn- behavioral patterns. However, Tomasello et al. ing of gestures in great apes. It is therefore not (1994) showed that most chimpanzee gestures possible to exclude some degree of imitation in involve getting the other’s attention and that the apes’ learning of gestures. Even so, that does not gestures vary with the attentional state of the imply that such gestures, e.g. the “arm shake” of recipient: non-tactile, non-vocal gestures are the gorillas of Apenheul zoo (Pika et al. 2003) predominantly used when there is visual contact are more than dyadic mimetic, since it is not between the recipient and the gesturer. Further- clear that they involve the communicative sign more, the animals clearly waited for a response function: e.g. represent something for someone. after gesturing. The presence of such “audience Some dyadic mimetic gestures may also in- effects” indicates that there was an intention on volve iconicity (similarity) between the spatio- the side of the gesturing chimpanzee to produce temporal profile of the gesture and the meaning Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 27 of the gesture: the action to be performed by the through ontogenetic ritualization, this under- addressee. An example would be to gently rub standing does not seem to be reflective enough down someone’s back in the desired direction of to be systematically used in communication. movement instead of forcefully pushing the other down. A purely visual version would be to 6.3 Triadic mimetic gestures make the gesture in the air in view of the recipi- ent. Such gestures are usually labeled iconic (Tan- As pointed out repreatedly above, what is lack- ner and Byrne 1996, 1999) and we will follow ing in dyadic mimetic gestures is an understand- this practice. However, we still need to ask ing of the communicative sign function, which whether they function as iconic signs for both the implies that the gesturer not only wishes the sender and the receiver, i.e. the iconicity is addressee to perform some desired action, but to grasped as the ground for the representational do this by appreciating the representational relationship. What characterises dyadic mimesis, (sign) relation between his gesture and its refer- as we stated above, is that this relationship is (a) ential meaning. As we show in this section apes unstable and (b) non-symmetrical: it is possible can achieve such an understanding, but clearly that only one of the communication parts per- only when they have been taught aspects of a ceives the iconicity as an iconic sign. conventional symbolic system, i.e. a language. In a study of captive bonobos by Savage- This raises some doubts whether such gestures Rumbaugh et al. (1977) almost all gestures that are not actually post-mimetic. preceded copulation appeared to be signs of copulatory positions. Since bonobos use many different copulatory positions it is expected that 6.3.1 Triadic mimetic indexicality: declarative pointing some mutual agreement must be reached about which position to be used. Two animals could The third stage in the development of human sometimes “argue”, in the sense that one ges- children’s pointing, mastered around 14 months, tured for a ventro-ventral position while the is usually called declarative pointing (Bates, other gestured for a dorso-ventral one. The like- Camaioni and Voltera 1975; Brinck 2004). The lihood of unusual copulation postures increased crucial difference with respect to imperative if the bout was preceded by gesturing. These pointing is that the child need not desire the gestures were grouped into three groups: (1) object pointed, but rather the act of achieving positioning motions: the recipient’s limbs are joint attention with the addressee on the object manipulated by the initiator. The actual move- is the goal in itself. As argued in Section 5.3, ment is done by the recipient but the direction is joint attention involves understanding a simple set by the initiator; (2) touch plus iconic hand form of communicative intention, and is thus motions: a touch on the part of the body that triadic mimetic. In this sense declarative pointing should be moved, followed by an indication with is also triadic mimetic. the hand how this part should move; (3) iconic Brinck (2004) proposes that at an early stage hand motions: without touching the other indi- of a child’s use, an additional purpose of declara- vidual the gesturer shows what he wants the tive pointing is the evaluation of the indicated other to do; these are the most abstract ones. object, achieved by an exchange of emotional It is possible to interpret this progression as information about it. For example, the child the development of an understanding of iconic- points to an object in order to obtain informa- ity and an emergence of iconic signs (Savage- tion from the addressee whether the object is Rumbaugh et al. 1977), but it should be noted dangerous. The main benefit for the child of this that the gestures from the first group were the kind of exchange is that he/she can learn about most frequent in the bonobo group, and ges- objects vicariously. This primary function pre- tures from the most abstract level, the third supposes that the child can understand the emo- group, was the least frequent. In other studies of tions of the addressee, and as we argued in Sec- bonobos (de Waal 1988; Pika et al. 2005) and tions 5.1 and 5.2, this is well within the gorillas (Pika et al. 2003) no gestures could be competence of the child at this stage of devel- readily interpreted as iconic in nature. So even if opment, and similarly for apes. Thus, it is possi- the evidence seems to grant apes some under- ble that the evaluative side of declarative point- standing of iconic signs, predominantly emerging 28 Bodily mimesis as “the missing link” in human cognitive evolution ing was the original cause of its proliferation We will return to this issue in the summary be- among the hominids. low. Enculturation has been shown to have dra- matic effects on the pointing of apes. While feral apes almost never point, as mentioned earlier, 6.3.2 Triadic-mimetic iconicity: communicative intention? captive apes and monkeys readily learn to engage in imperative, though not declarative pointing. Possibly the clearest evidence for iconic triadic On the other hand, at least three of the docu- mimetic gestures in non-language-trained apes mented language-taught apes – Koko, Chantek comes from a study of zoo-living gorillas, par- and Kanzi – consistently point declaratively, i.e. tially reared by human caregivers. Tanner and refer to objects and locations for sake of calling Byrne (1996, 1999) studied a pair of western these to the attention of someone else (Patterson lowland gorillas, Kubie and Zura, at the San 1980; Miles 1990). In the mentioned review of Francisco Zoo, who seemed to have an unusu- pointing in human children and apes, Leavins ally rich repertoire of gestures: audible, silent and and Hopkins (1999), point out that language- tactile ones. The gestures were used almost ex- trained apes were more likely to use a canonical clusively in play sessions and sexual situations, point with the index finger than children of the and the choice of gesture differed according to age 13-18 months. In some studies, it is even the attentional state of Zura, with audible and suggested that language-taught apes combine tactile ones used when she was not looking, such an act of declarative pointing and another which mirrors the findings for chimpanzees by sign to make a predication, i.e. THIS is X Tomasello et al. (1994). (Greenfield and Savage-Rumbaugh 1990). Some of Kubie’s gestures were “attention However, the question arises if such acts are getters” and play invitations. His most successful cases of triadic mimesis, rather than post- gesture was an armswing under his belly towards mimetic skill. The close relation between de- his genitals. It had the highest frequency of ac- clarative pointing and reference makes it likely companying “play faces” and resulting in physi- that the process of being taught language itself is cal contact. It was only performed when the what has simultaneously taught the significance gorillas had visual contact. “Armswing under” of declarative pointing to the apes. The follow- was often preceded by a tap on Zura’s body. ing episode recorded by Savage-Rumbaugh and This is comparable with Savage-Rumbaugh et Levin (1994), involving not the famous, highly al.’s (1977) second group of iconic gestures: enculturated bonobo Kanzi, but the somewhat touch plus iconic hand motions. If Zura failed to more traditionally trained Austin and Sherman, respond, a gesture was often repeated, strength- shows how enculturation, and specifically the ened or more gestures were made – this also teaching of communicative sign function, can resembling the pattern found in chimpanzees affect pointing: (Tomasello et al. 1994). Most of the gestures Upon spying the new food, Sherman reached ended in contact between Kubie and Zura out for it and I suggested that he should ask within 5 seconds, with Zura making most of the for it. He quickly turned, scrutinized the key- contact, suggesting that she was (correctly) in- board, and deliberately, but with no fanfare, terpreting Kubie’s gestures. In 66% of the cases indicated one of the unassigned lexigrams as when a tactile gesture was made Zura moved her the symbol for the food. Austin watched at- body in the direction indicated by Kubie. If a tentively and subsequently used the symbol gesture was coupled with a “play face” the likeli- Sherman had selected to indicate that he hood of physical contact was significantly higher would like to try the new food also. This un- than for gesture alone. However, “head nod”, expected skill encouraged us to believe that Kubie’s most frequent gesture, had a high degree Sherman and Austin understood that a unique of play bouts without using “play faces”, which correspondence existed between each food shows that Zura did respond to gestures and not item and a specific symbol. Moreover, they sometimes pointed back and forth between just to the natural (pre-mimetic) gorilla signals. the symbol and the food item. They had Still, the question remains if Zura really re- named new foods and apparently agreed on sponded to their (intended) meaning, compre- coordinated use of the selected symbols. (ibid: hending the gestures as iconic signs. Associating 79) playfulness (or sexual agendas) with particular Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 29 behaviors in another individual’s repertoire is and Byrne 1996) might have implications for the not sufficient to show that Zura understood conclusion we can draw about mimetic capaci- Kubies gestures as communicative signs. ties in non-enculturated apes. Human beings Whether or not Kubie’s gestures had an iconic engage readily in triadic forms of communica- relationship to their referents for him, they could tion and apes with extensive exposure to human in each case remain indexical to Zura, i.e. be language show good comprehension of other behavioral correlates to moods. The iconicity of representational mediums, like words and pic- some of Kubie’s gestures could very well have tures (Savage-Rumbough, Shanker and Taylor been abstractions from more physical manipula- 1998). However, it is not yet known how much tions shaped through ontogenetic ritualisations exposure to human culture is needed to yield an (Tomasello and Call 1997) and Zura’s appropri- understanding of triadc mimesis (if at all) in ate response be derived from the context. hand-reared zoo animals, and if this would be However, over time Kubie used his gestures retained into adulthood. with several females, and if he did not go While the presence of iconic signs in captive through an identical abstraction phase with all of apes remains unsure, there is solid evidence for them, which of course cannot be ruled out com- their use in enculturated apes. The sign-language pletely, how could the receiver understand them? taught gorilla Koko reportedly invented a num- Tanner and Byrne (1996) suggest that compre- ber of such signs herself, including signs refer- hension of motion depicted in gesture may be ring to actions e.g. “blow”, “tickle”, “walk-my- biologically encoded for the great apes since back”, “bite”, as well as objects: “bird”, “bracelet”, their adaptation to brachiation requires a three- “stethoscope”. These gestures clearly involve a dimensional understanding of space, and space is referential triangle involving the gesture, referent the medium for gestures.7 and the interpreter, and were sometimes even In summary, like the bonobo sexual gestures used in the absence of the referent, showing one described in 6.2.2, the San Francisco gorilla ges- of the crucial features of language: displacement tures could be argued to be at least dyadic- (Hockett 1960), and even “metaphorically”, e.g. mimetic, since they correspond to actions in- using “bird” as an insult. tended to be performed by the recipient. But A problem for classifying them as triadic through their iconicity the gestures also have a mimetic, however, is that these gestures were potential to be triadic-mimetic, under the condi- quite similar to the actual signed language signs tion that at least the recipient, and possibly also that she was taught. In accordance with the ca- the sender understood their sign function. Tri- veat stated earlier, it is difficult to exclude the adic mimesis cannot be excluded as an alterna- interpretation that they are actually post- tive to ontogenetic ritualization in the formation mimetic, i.e. induced by the language-teaching and use of such gestures because non- situation itself. conventional gestures could be understood not Some support for their mimetic-iconic nature only through the context but through the iconic- is offered by the fact that Koko produced many ity of the gestures themselves. If, and it is a big so-called gestural modulations with respect to size, “if”, Kubie, Zura and the other females could manner, number, etc. – all having an iconic ba- use the iconic gestures effectively because of sis. For example, the signs for a “small bird” and such referential understanding, Kubie’s gestures a “big bird” would differ in size; those for “sev- would be an example of triadic mimesis. More eral birds” would involve repetition of the sign. empirical research is necessary to decide this Thus, gestural modulations are perhaps the issue. clearest evidence of the presence of triadic mi- A final point should be made in relation to mesis in Koko’s productions, especially since the differences observed between feral and en- Patterson (1980: 539) emphasizes that they culturaed apes pointed out earlier. The fact that “were created by Koko herself”. Other testi- Zura, the other female gorillas and Kubie all at mony for the spontaneous production of triadic some point had been reared by humans (Tanner mimetic gestures in a language-trained ape can be found in the spontaneous gestures of the 7 Brachiation also entails an ability that is unique to bonobo Kanzi, who unlike Koko was not taught apes: that of being able to move the wrist, elbow signed language but a combination of visual and shoulder joints in all directions, making subtle lexigrams (in production) and spoken English (in gestures possible. 30 Bodily mimesis as “the missing link” in human cognitive evolution comprehension). Greenfield and Savage- ing to the most conservative criteria no less Rumbaugh (1990, 1991) report on Kanzi’s spon- than 140 signs; taneous use of pointing (see below), and at least • regard the acquired signs as conventional- some gestures involving iconicity such as normative (consensual), to the point of cor- “come”, “go” and “chase”. recting their teachers if the latter do not use these appropriately; 6.4 Post-mimetic gestures • understand novel combinations of spoken or signed words; Mimesis was defined by Donald (1991) to be • produce novel combinations of signs; “non-linguistic”, and in Section 3 and in our • use language for a number of different func- definition of bodily mimesis we clarified what we tions (speech acts), including labeling, an- take this to mean: lack of full conventionality swering, expressing emotion, arguing and in- (mutual knowledge) of the signs and systematic- sulting; ity, involving at least a degree of compositional- • use language not only for communication, ity. To the extent that the gestures of apes dis- but for thinking (private speech). play these properties, they can be regarded as post-mimetic. What has not yet been clearly demonstrated is Evidence from four of the most successful whether the spoken or signed utterances of apes projects involving the teaching of language to conform to consistent principles of grammatical great apes – the chimpanzee (Pan troglodytes) organization, though e.g. Greenfield and Savage- Washoe (Fouts 1972, 1973), the gorillas Koko Rumbaugh (1990, 1991) describe at least two and Michael (Patterson 1978, 1980) and the fairly consistent ordering rules in the two symbol bonobos (Pan paniscus) Kanzi and combinations of Kanzi, one of them involving a (Savage-Rumbaugh and Lewin 1994; Savage- pointing gesture (see below). Rumbaugh et al. 1998) and the orangutan Chan- It is thus unsurprising that those who have tek (Miles 1990) – has shown that many of the (over-)emphasized syntax as opposed to seman- essential characteristics of language are within tics as the defining characteristic of language the grasp of our nearest non-human relatives. As (e.g. Bickerton 1990; Pinker 1994; Hauser, is the case with children a precondition for the Chomsky and Fitch 2002) continue to deny “the success of these projects has been a cultural proposition that language is no longer the exclu- environment rich with intersubjectivity and a sive domain of man” (Patterson 1978: 95). But variety of activities to stimulate communication even if the language of apes such as Koko and (Miles 1990). Our conclusion is similar to that of Kanzi is termed “proto-language” rather than Miles (1999: 204), namely that all great apes true language, it is still qualitatively different “have the intelligence for a rudimentary, referen- from mimesis, since it possesses at least rudi- tial, generalizable, imitative, displaceable, symbol ments of conventionality and systematicity. In system”. this sense, some, though not all, of the gestures The reason that this claim has been contro- of the language-taught apes can be regarded as versial (e.g. Terrace et al. 1981) is not foremost post-mimetic. empirical, but rather a matter of a disagreement on what the “essential characteristics” of human language are. Turning to the data reported in the 6.4.1 Post-mimetic indexicality: paralinguistic pointing “ape language” literature we see convincing evi- dence that apes can: At least two forms of the pointing of language trained apes seem to go beyond mimesis, since • comprehend the referential (representational) the pointing gesture is integrated in the (proto-) function of spoken words, ASL signs and linguistic system acquired. First of all are the visual lexigrams, and combinations of these; combinations of a pointing gesture referring to • use the sign-tokens in the absence of their an object or an individual with another sign ex- referents, i.e. “displacement” (Hockett 1960); pressing a predicate applying to the individual, • acquire a considerable vocabulary of thus forming a predication. Furthermore, the words/signs, according to some measure- bonobo Kanzi has been reported to form such ments extending 600 signs, but even accord- “predications” – with imperative as well as de- Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 31 clarative functions – in a fairly systematic man- 1980s and 1990s, they rapidly made this seg- ner Greenfield and Savage-Rumbaugh (1990, mented, sequenced construction the preferred 1991). means of expressing motion events. NSL thus quickly acquired the discrete, combinatorial A related form post-mimetic indexicality in- nature that is hallmark of language. volves labeling, in which the pointing gesture Given their mimetic-gestural origin signed lan- directed at an object, or a picture of one in a guages have a much greater degree of iconicity book, is accompanied with the the production of than spoken languages and it has been proposed (gestural) sign denoting the object. Such clearly that this plays a role in their faster acquisition by declarative uses of pointing + label, some of (deaf) children (Brown 1977). Recent studies which are performed “privately”, i.e. not for the have questioned this, however, since only a mi- benefit of a human interlocutor (Bodamer et al. nority of the signs of signed language have 1994) are perhaps the most convincing case for transparent iconic meanings, and in a study of 22 the presence of post-mimetic, paralinguistic children acquiring ASL it was shown that “of the pointing in enculturated apes. Kanzi has been 44 different signs produced by the children be- shown to use pointing + lexigram, as well as fore the age of 13 months, 36% were classified lexigram + pointing combinations for the sake as iconic, 30% as metonymic, and 34% as arbi- of labeling quite frequently: altogether there trary” (Bonvillian and Patterson 1999: 253).8 In were 249 examples in his five-month corpus of this study, the authors compared the rate and two-element combinations in which Kanzi indi- pattern of acquisition of ASL by children and cated an entity gesturally and named it with a that of two gorillas, Koko and Michael. Despite lexigram (Greenfield and Savage-Rumbaugh certain differences – the children’s acquisition 1991). was (unsurprisingly) faster – it was shown that “that the similarities in early development across the species outweigh the differences” (ibid: 260). 6.4.2 Post-mimetic iconic gestures: Signed language Thus, it can be concluded that gorillas, and by inference other great apes, not only can acquire Human signed languages offer the clearest pic- the basics of a post-mimetic symbolic system ture of post-mimesis. Recent studies of the such as ASL, but that they do this is a similar spontaneous emergence of Nicaraguan Sign way. Interestingly, however, the gorillas seemed Language (NSL) during the past 25 years show to rely somewhat more on the iconicity of the that signed languages have their origin in (tri- signs in comparison with the children, so that adic) mimesis, but quickly acquire the properties the proportion of their first 46 signs was some- of conventionality and systematicity. Senghas, what different to the one reported above: 42% Kita and Özyürek (2004) compared the co- iconic, 32% metonymic and 26% arbitrary, while speech gestures of Nicaraguan speakers of Span- for the first 10 signs this difference was even ish, with the signing of three “cohorts”, or gen- clearer: 60% iconic, 20% metonymic and 20% erations, of learners of NSL and could document arbitrary. We interpret this as suggesting that the some aspects of this transition in detail: apes relied to a greater degree on their skills of The movements of the hands and body in the triadic mimesis than the children in their acquisi- sign language are clearly derived from a ges- tion of the sign language, which also would ex- tural source. Nonetheless, the analyses reveal a plain why the children quickly progressed be- qualitative difference between gesturing and yond the initial level of vocabulary acquisition to signing. In gesture, manner and path were in- tegrated by expressing them simultaneously and holistically, the way they occur in the mo- 8 “Metonymic” signs are such that involve some tion [event] itself. Despite this analogue, holis- degree of iconicity between the sign and the refer- tic nature of the gesturing that surrounded ent, but ”the tie between the sign and its meaning is them, the first cohort of children, who started not readily apparent – one would be unlikely to building NSL in the late 1970s, evidently in- guess the meaning of a metonymic sign simply troduced the possibility of dissecting out seeing it produced” (Bonvillian and Patterson 1999: manner and path and assembling them into a 252). In the distinction introduced by Sonesson sequence of elemental units. As second and (2001) they involve secondary (rather than primary) third cohorts learned the language in the mid iconicity, which can be perceived first after know- ing the specific sign relationship. 32 Bodily mimesis as “the missing link” in human cognitive evolution learn the systematic character, i.e. the grammar, the bonobo Kanzi and the orangutang Chantek. of the language, while the apes “stagnated” on a We interpret this as evidence for triadic mimesis, simple, “proto-language” level. though with some hesitation, since it is likely that being taught language was necessary for the 6.5 Summary apes to acquire the communicative sign function. Even so, triadic mimesis can be shown to be In this section we have reviewed some of the within the grasp of our nearest non-human rela- evidence on the production, and to some extent tives. In comparison, their skills in post-mimesis the comprehension of gestures by great apes, involving full conventionality and systematicity, and when relevant compared it to the acquisition i.e. language, are more limited. The conclusion, of gestures and sign languages by children. Ap- similar to that at the end of Section 5, is that plying the mimesis hierarchy to this evidence has triadic mimesis is within the “zone or proximal resulted in the classifiction summarized in Table evolution” of great apes, and by inference, to 5. our common ancestor.

Table 5. The mimesis hierarchy and ape gestures 7. Summary and conclusions

The goal of this study has been to explore the Level Indexical gestures Iconic gestures potentials of bodily mimesis to provide perhaps Proto- Reaching Some phylogentically not “the” missing link in explaining human cog- mimesis ritualized signals nitive evolution, but at least one important miss- ing piece in the puzzle, and thus help unravel the Dyadic Pointing for Some ontogeneti- mystery of the origins of language. In Section 3 mimesis oneself cally ritualized ges- tures we defined the concept of bodily mimesis and Imperative the related mimesis hierarchy, thereby refining and pointing modifying the concept of mimesis proposed by Triadic Declarative Iconic signs Donald (1991). Our method has been to aply the mimesis pointing mimesis hierarchy to evidence from primatology Post- Paralinguistic (Aspects of) signed (and to some extent child development) in order mimesis pointing language to be able to compare ape skills in three different domains: imitation, intersubjectivity and gesture Some of the natural signals of apes can be re- and to determine whether any general patterns garded as involving indexicality and iconicity, could be discerned. Table 6 below summarizes without these being perceived as signs by either our results. the sender or the receiver. On the other hand, if Beginning with proto-mimesis, we could see an observer and, especially, a caregiver during that there is abundant evidence from all three infancy does see these as signs, this can provide domains that apes excel in this. There do not an impetus for making these gestures mimetic, as appear to be any (major) cross-species differ- seems to be the case with the origin of impera- ences with humans in this respect, and we can tive pointing. At least some ontogenetically ritu- assume with some confidence that the common alized gestures appear to involve differentiation ape-human ancestor had comparable skills in and correlation between the gesture and mean- imitation, intersubjectivity and communicative ing, and thus can be regarded as simple iconic gestures. signs. Since these, however, can function with- On the next level, that of dyadic mimesis, we out having and interpreting communicative in- begin to discern some differences – both be- tentions, they are (at most) at the level of dyadic tween the levels, and between humans and apes. mimesis. Feral and even captive apes have not There is clear evidence for e.g. cognitive empa- been convincingly shown to progress beyond thy and ontogenetically ritualized gestures in this level. apes – and we suggested how these phenomena Enculturation, and especially language- can be explained in a dyadic-mimetic fashion teaching changes this and consequently both involving both identification with and differen- declarative pointing and iconic gestures emerge, tiation from the other. However, when it comes testified in individuals such as the gorilla Koko, to imitation, apes appear to have difficulties in Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 33 performing detailed imitation on the action level, hierarchy. As for the evidence concerning non- which is the type of imitation where bodily mi- enculturated apes, it is much less certain that mesis is most relevant. Human children in com- they reach the level of triadic mimesis, but best parison excel in this and even “overmimic”. This with respect to iconic gesture (e.g. Tanner and difference between the species does not appear Byrne 1996, 1999). We can generalize and sug- to be due to a difference in intersubjectivity, gest that apes appear to have difficulties with i.e. understanding of others’ intentions, as was third-order mentality, involving both attention and originally suggested by Tomasello (1999) but intention. However, as shown by the studies of rather in mimetic capacity per se. Thus, our re- enculturated apes, through the practice of com- sults confirm, at least in part, Donald’s proposal municative sign use, apes have the potential to that (bodily) mimesis constitutes the phenome- understand communicative intentions, thereby non in relation to which we should look for showing that even triadic mimesis is within apes’ human cognitive specificity. At the same time, “zone of proximal development”. the results reviewed in Section 4 showed clearly Finally, the capacity of apes for post-mimesis, rely- that apes have the basic capacity for dyadic mi- ing on the possession of a conventional sym- mesis, and are thus beyond the limitations im- bolic system, is furthest away from the potential posed by “episodic cognition” (Donald 1991). of human beings. While enculturated apes such as Kanzi may have mastered a proto-language In contrast, when we come to the capacities of after a good deal of exposure and effort, this apes to perform triadic mimesis we find predomi- remains limited in grammatical and semantic complexity, approximately to the level of a two- nantly negative evidence in all three domains. year old child. Thus, even post-mimesis is not Enculturated apes such as Kanzi, Chantek and completely beyond the ZPD of apes, though it is Koko are capable of this with respect to gesture further in the periphery than triadic mimesis. (e.g. declarative pointing) and intersubjectivity Using the analogy suggested by Donald (2001), (e.g. joint attention), and possibly also, but less we can say that both triadic mimesis and post- clearly with respect to the learning and/or inven- mimesis are within apes’ “zone of proximal tion of signs through imitation, which also here appears to be “lagging behind” on the mimesis

Table 6. Comparing ape skills in the domains of imitation, intersubjectivity and communicative gestures along the levels of the mimesis hierarchy: each level mentions example capacities, with “+” indicating positive evidence and “– ” negative evidence. W = wild, C = captive, E = enculturated.

Level Imitation Intersubjectivity Communicative gesture Proto- Neonatal mirroring Mutual attention Phylogenically ritualized gestures mimesis Empathy C + W/C + C + Dyadic Action level imitation Shared attention Ontogentically ritualized gestures mimesis Cognitive empathy Imperative pointing C + (?) C + C + Triadic Imitatititave signs Joint attention Declarative pointing mimesis W – W – W – C – C – C + (?) E + (?) E + E + Post- Role-reversal imitation Belief understanding Proto-language mimesis – – E + Jordan Zlatev, Tomas Persson and Peter Gärdenfors, Lund University Cognitive Studies, 121, (2005). evolution”, i.e. the sphere in which evolutionary In sum, our general proposal is that the adaptations are likely to accur, but since post- source of human cognitive specificity lies above mimesis presupposes triadic mimesis (even con- all in two, mutually co-enforcing factors involv- ceptually), the latter constitutes the most likely ing bodily mimesis: (a) the capacity to form mi- missing link. metic schemas: representations deriving from the What kind of factors brought this about in imitation of public events and (b) the use of evolution? The evidence summarized and ana- mimetic schemas for intentional communication: lyzed in this article allow us to make two high- triadic mimesis. The first of these factors is pre- level generalizations concerning the three do- dominantly representational and the second, mains of bodily mimesis analysed: communicative, and it is likely that in evolution (1) Ape skills of imitation are comparatively the two have co-evolved. We submit that this least developed along the mimesis hierarchy; proposal is consistent with the evidence involv- (2) Ape skills of gesture are comparatively most ing the capacities and limitations of apes, and developed along the mimesis hierarchy. shows the explanatory potential of the concept of bodily mimesis as “a missing link” in our Furthermore, we could discern a possible causal theoretization of cognitive and linguistic evolu- relationship in the direction from gesture (sign tion. use) toward intersubjectivity: the only apes who succeed in tasks of joint attention, where those References who where taught to use sign system through enculturation. (In the case of imitation the rela- Ahlgren, I. (2003) Teckenspråk. In: Sveriges officiella tion is more of a logical sort: since triadic imita- minoritetsspråk. Finska, meänkieli, samiska, romani, tion is defined as the imitation of signs, it is im- jiddisch och teckenspråk. En kort presentation. possible to perform this without having first Nordstedt. learned the latter.) Thus we can conclude by Anisfeld, M. (1996) Only tounge protrusion modeling is matched by neonates. Developmental Review, 16: suggesting the following two (partially independ- 149-161. ent) hypotheses of human cognitive evolution, Arbib, M. (2003) The evolving mirror system: A corresponding to the two generalizations given neural basis for language readiness. In: Language above: Evolution, ed. M. Christiansen and S. Kirby, Ox- (1´) The crucial adaptation that differentiated ford University Press. early hominds from apes involved detailed bodily - (in press) From monkey-like action recognition to imitation, allowing for the development of spe- human language: An evolutionary framework for cific mimetic schemas which are pre-linguistic rep- neurolinguistics. Behavioral and Brain Sciences. resentational, intersubjective structures, emerg- Armstrong, D., Stokoe, W., and Wilcox, S. (1995) ing through imitation but subsequently Gesture, and the Nature of Language. Cambidge: interiorized (see Piaget 1945; Zlatev 2005a; Cambridge University Press. Zlatev, Persson and Gärdenfors in press). Astington, J. W. and Jenkins, J.M. (1999) A longitu- dinal study of the relation between language and (2´) Since apes display some potential for theory-of-mind development. Developmental Psy- spontaneously developing indexical and iconic chology, 35/5: 1311-1320. gestures – e.g. in the rather unusual environment Badcock, C. (2000) Evolutionary Psychology. A Critical of the San Francisco Zoo, see Tanner (2004) – it Introduction. Cambridge: Polity Press. is reasonable to assume that the common ape- Bard K. A., Myowa-Yamakoshi, M., Tomonaga, M., human ancestor possessed such a potential as Tanaka, M., Costall, A. and Matsuzawa, T. (in well. An “ecological niche” requiring a high de- press) Group differences in the mutual gaze of gree of intentional communication (there are chimpamzees. Developmental Psychology Vol. 41, 2005. various scenarios for bringing this about; for a Bard, K. A. and Vauclair, J. (1984) The communica- proposal regarding the role of material tive context of object manipulation in ape and culture see Osvath and Gärdenfors 2005) would human adult-infant pairs. Journal of Human Evolu- have spurred the use of iconic and indexical tion, 13: 181-90. gestures, which on their part would have both Baron-Cohen, S. (1995) Mindblindedness: An Essay on utilized and futher developed mimetic schematic Autism and Theory of Mind. Cambridge, MA: MIT representations. Press.

Lund University Cognitive Science-LUCS 121 ISSN 1101 – 8453 © 2005 by the authors

Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 35

Bates, E., Camaioni, L. and Volterra, V. (1975) Per- Butterworth, G. (1988) What is special about point- formatives prior to speech. Merrill-Palmer Quar- ing? In: The Development of Sensory, Motor, and Cog- terly, 21: 205-226. nitive Capacities in Early Infancy: From Perception to Bering, J. M., Bjorklund, D. F. and Ragan, P. (2000) Cognition, ed. F. Simion and G. Butterworth 171- Deferred imitation of object-related actions in 190. Hove: Psychology Press. human-reared juvenile chimpanzees and orangu- Byrne, R. W. and Stokes, E. J. (2002) Effects of man- tans. Developmental Psychobiology, 36: 218-32. ual disability on feeding skills in gorillas and Bickerton, D. (1990) Language and Species. Chicago: chimpanzees. International Journal of Primatology, University of Chicago Press. 23/3: 539-54. - (2003) Symbol and structure: A comprehensive Byrne, R. W. (1995) The Thinking Ape: Evolutionary structure for language evolution. In: Christiansen Origins of Intelligence. Oxford: Oxford University and Kirby (2003), 77-93. Press. Bjorklund D. F., Bering, J. M. and Ragan, P. (2000) A - (1998) Commentary on Boesch, C. and two-year longitudinal study of deferred imitation Tomasello, M. Chimpanzee and human culture. of object manipulation in a juvenile chimpanzee Current Anthropology, 39: 604-5. (Pan troglodytes) and orangutan (Pongo pygmaeus). - (1999) Imitation without intentionality. Using Developmental Psychobiology, 37: 229-37. string parsing to copy the organization of be- Bjorklund, D. F. and Bering, J. M. (2003) A note on haviour. , 2: 63-72. the development of deferred imitation in encul- Byrne, R. W. and Russon, A. E. (1998) Learning by turated juvenile chimpanzees (Pan troglodytes). De- imitation: A hierarchical approach. Behavioral and velopmental Review, 23: 389-412. Brain Sciences, 21: 667-21. Bjorklund, D. F., Yunger, J. L., Being, J. M. and Ra- Call, J. and Tomasello, M. (1994) Production and gan, P. (2002) The generalization of deferred imi- comprehension of referential pointing by tation in enculturated chimpanzees (Pan troglo- orangutans (Pongo pygmaeus). Journal of Comparative dytes). Animal Cognition, 5: 49-58. Psychology, 108: 307-317. Bloom, P. (2000) How Children Learn the Meaning of - (1996) The effect of humans in the cognitive Words. Cambridge, MA: MIT Press. development of apes. In: Reaching into Thought: Bodamer, M. D., Fouts, R. S., Fouts, D. H., and The Minds of Great Apes. ed. A. Russon, K. Bard and S. Parker. Cambridge: Cambridge University Jensvold, K. B. (1994) Indicating acts during Press. counting by a chimpanzee (Pan troglodytes). Journal - (1998) Distinguishing intentional from accidental of Comparative Pscyhology, 109: 47-51. actions in orangutans (Pongo pygmaeus), chimpan- Boesch, C. (2003) Is culture a golden barrier between zees (Pan troglodytes), and human children (Homo human and chimpanzee? Evolutionary Anthropol- sapiens). Journal of Comparative Psychology, 112/2: ogy, 12: 82-91. 192-206. Boesch, C. and Tomasello, M. (1998) Chimpanzee - (1999) A nonverbal false belief task: The per- and human cultures. Current Anthropology, 39/5: formace of children and great apes. Child Devel- 591-614. opment, 70/2: 381-395. Bonvillan, J. D. and Patterson, F. G. (1993) Early sign Call, J., Hare, B., Carpenter, M. and Tomasello, language acquisition in children and gorillas: Vo- M. (2004) “Unwilling” versus “unable”: chim- cabulary content and sign iconicity. First Lan- panzees’ understanding of human intentional ac- guage, 13: 315-38. tion. Developmental Science, 7/4: 488-98. - (1999) Early sign-language acquisition: Compari- Cheney, D. and Seyfarth, R. (1990) How Monkeys See son between children and gorillas. In: The Men- the World: Inside the Mind of Another Species. Chi- talities of Gorillas and Orangutans: Comparative Per- cago: University of Chicago Press. spectives, eds, S.T. Parker, R.W.Mitchell, H.L. Christiansen, M. and Kirby, S. (2003) (eds.) Language Miles, 240-264. Cambridge University Press. Evolution. Oxford: Oxford University Press. Brinck, I. (2001) Attention and the evolution of in- Clark, H. (1996) Using Language. Cambridge: Cam- tentional communication. Pragmatics and Cognition, 9/2: 255-272. bridge University Press. - (2004) Joint attention, triangulation and radical Corballis, M. C. (2002) From Hand to Mouth: The Ori- interpretation: A problem and its solution, Dialec- gins of Language. Princeton, NJ: Princeton Univer- tica, 58(2): 179-206. sity Press. Brinck, I. and Gärdenfors, P. (2003) Co-operation Coussi-Kourbel, S. (1994) Learning to outwit a com- and communication in apes and humans. Mind petitor in mangabeys (Cercocebus torquatus tor- and Language, 18(5): 484-501. quatus). Journal of Comparative Psychology, 108: 164- Brown, R. (1977) Why are signed languages easier to 171. learn than spoken languages? In: Proceedings of the Custance, D. M., Whiten, A. and Bard, K. A. (1995) National Symposium on Sign Language Research and Can young chimpanzees (Pan troglodytes) imi- Training, ed. W. Stokoe, 9-24. Silver Spring, MD: tate arbitrary actions? Hayes and Hayes (1952) National Association of the Deaf. revisited. Behaviour, 132/11-12: 837-59. 36 Bodily mimesis as “the missing link” in human cognitive evolution

Custance, D. M., Whiten, A., Sambrook, T. and Eilan, and A. Marcel. Cambridge, MA: Galdikas, B. (2001) Testing for social learning in MIT/Bradford Press. the “artificial fruit” processing of wildborn - (2005) How the Body Shapes the Mind. Oxford: orangutans (Pongo pygmaeus), Tanjung Puting, In- Oxford University Press. donesia. Animal Cognition, 4: 305-13. - (in press) Phenomenological and experimental Custance, D. M., Whiten, A. and Fredman, T. (1999) contributions to understanding embodied expe- Social learning of an artificial fruit task in capu- rience. In Ziemke, Zlatev and Frank (in press). chin monkeys (Cebus apella). Journal of Comparative Psychology, 113/1: 13-23. Gallese, V., C. Keyners and Rizzolatti, G. (2004) A de Waal, F. (1988) The communicative repertoire of unifying view of the basis of social cognition. captive bonobos (Pan paniscus), compared to that Trends in Cognitive Sciences, 8/9: 396 – 403. of chimpanzees. Behaviour, 106/3-4: 183-251. Gallup, G. (1982) Self-awareness and the emergence - (1996) Good Natured: The Origins of Right and Wrong of mind in primates. American Journal of Primatol- in Humans and Other Animals. Harvard: Harvard ogy, 2: 237-248. University Press. Gardner, R. A. and Gardner B.T. (1969) Teaching - (1998 [1982]) Chimpanzee Politics. Harper. sign language to a chimpanzee. Science, 165: 664- de Waal, F. and Aureli, F. (1996) Consolation, Reconcilia- 672. tion and a Possible Cognitive Difference Between Ma- Givón, T. (2002) Bio-Linguistics. The Santa Barbara caques and Chimpanzees. Cambridge: Cambridge Lectures. Amsterdan: Benjamins. University Press. Goldin-Meadow, S. (1999) The role of gesture in de Villiers, J., and Pyers, J. (1997) Complementing communication and thinking. Trends in Cogni- cognition: The relationship between language tive Sciences, 3/11: 419-429. and theory of mind. In Proceedings of the 21st An- Gómez, J. C. (1994) Mutual awareness in primate nual Boston University Conference on Language Devel- opment. Somerville, MA: Cascadillia Press. communication: a Gricean approach, In: Self- awareness in Animals and Humans, ed. S. T. Parker, Deacon, T. (1997) The Symbolic Species: The Co- R. W. Mitchell and M. L. Boccia. Cambridge: Evolution of Language and the Brain. New York: Cambridge University Press. Norton. Gopnik, A. and Astington, J. W. (1988) Children’s - (2003) Universal grammar and semiotic con- understanding of representational change, and its straints. In Christiansen and Kirby (2003), 111- relation to the understanding of false belief and 139. the appearance-reality distinction, Child Develop- Dennett, D. (1996) Kinds of Minds. Towards an Under- ment, 59: 26-37. standing of Consciousness. London: Phoenix. Gopnik, A. and Graf, P. (1988) Knowing how you Dessalles, J.-L. (2004) Language as an isolated niche. know: young children’s ability to identify and Presented at The Fifth Conference on the Evolution of remember the sources of their beliefs. Child De- Language, Leipzig. velopment, 59: 1366-1371. Donald, M. (1991) Origins of the Modern Mind. Three Gopnik, A. and Meltzoff, A. N. (1997) Words, Stages in the Evolution of Culture and Cognition. Har- Thoughts, and Theories, Cambridge, MA: MIT vard: Harvard University Press. Press. - (1998) Mimesis and the Executive Suite: Missing Greenfield, P. M. and Savage-Rumbaugh. E. S. (1990) links in language evolution. In Approaches to the Grammatical combination in Pan pamiscus: Evolution of Language, eds., J. Hurford, M. Stud- Processes of learning and invention in the evolu- dert-Kennedy and C. Knight, 44-67. Cambridge: tion and development of language, In: “Language” Cambridge University Press. and Intelligence in Monkeys and Apes, eds. S T. - (2001) A Mind So Rare. The Evolution of Human Parker and K. R. Gibson, 540-578. Cambridge: Consciousness. New York: Norton. Cambridge University Press. Edelman, G. (1992) Bright Air, Brilliant Fire: On the Greenfield, P. M. and Savage-Rumbaugh, E. S. (1991) Matter of the Mind. New York: Basic Books. Imitation, grammatical development, and the in- Eldridge, N. and Gould. S. J. (1972) Puctuated vention of protogrammar. In N. Krasnegor, D. equilibria: An alternative to phyletic gradualism. Rumbaugh, M. Studdert-Kennedy, and R. In: Models in Paleobiology, ed., T. Schopf. San Fran- Schiefelbusch (Eds.), Biological and behaviorial de- sisco: Freeman. Fouts, R. S. (1972) Use of guidance in teaching sign terminants of language development (pp. 235-258). language to a chimpanzee (Pan troglodytes). Journal Hillsdale, NJ: Erlbaum. of Comparative and Physiological Psychology, 80/3: Greenfield, P. M., and Savage-Rumbaugh, E. S. 515-22. (1993) Comparing communicative competence - (1973) Acquisition and testing of gestural signs in in child and chimp: The pragmatics of repetition. four young chimpanzees. Science, 180: 978-980. Journal of Child Language, 20, 1-26. Gallagher, S. (1995) Body schema and intentionality. Grice, P. (1957). Meaning. Philosophical Review, 66: In: The Body and the Self, ed. J. Bermúdez, N. 377-88. Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 37

Gärdenfors, P. (2003) How Homo Became Sapiens: On pointing from a comparative perspective. Journal the Evolution of Thinking. Oxford: Oxford Univer- of Comparative Psychology, 113/4: 417-425. sity Press. Lohmann, H. and Tomasello, M. (2003) The role of Hale, C. M. and Tager-Flusberg, H. (2003) The influ- language in the development of false belief un- ence of language on theory of mind: A training derstanding: A training study. Child Development study. Developmental Science 6/3: 346-359. 74/4: 1130-1144. Hare, B., Call, J., Agnetta, B. and Tomasello, M. Liberman, A. M. and Mattingly, I. G. (1985) The (2000) Chimpanzees know what conspecifics do motor theory of speech perception revised. Cog- and do not see. Animal Behaviour, 59: 771-85. nition, 21: 1-36. Hare, B., Call, J. and Tomasello, M. (2001) Do chim- Mandler, J. (2004) The Foundations of Mind: Origins of panzees know what conspecifics know? Animal Conceptual Thought. Oxford: Oxford University Behaviour, 61: 139-51. Press. Hauser, M. (1996) The Evolution of Communication. Marler, P. (1985) Representational vocal signals of Cambridge, MA: MIT Press. Hauser, M., Chomsky, N. and Fitch, T. (2002) The primates. In: Experimental Behavioral Ecology and faculty of language: What is it, who has it and Sociobiology, ed. B. Hölldobler and M. Lindauer. how did it evolve? Science, 298: 1559-1569. Stuttgart: Fischer Verlag. Heyes, C. (1998) Theory of mind in nonhuman pri- - (1991) Difference in behavioral development in mates. Behavioral and Brain Sciences, 21: 101-115. closely related species: Birdsong. In: The Develop- Hockett, C. F. (1960) The origin of speech. Scientific ment and Intergration of Behavior, ed. P. Bateson, 41- American, 23: 89-96. 70. Cambridge: Cambridge University Press. Itkonen, E. (1978) Grammatical Theory and Metascience. Matsuzawa, T., Biro, D., Humle, T. Inoue-Nakamura, Amsterdam: Benjamins. N., Tanook, R. and Yamakoshi G. (2001) Emer- Ikegami, T. and Zlatev, J. (in press) From pre- gence of culture in wild chimpanzees: Education representational cognition to language. In by master-apprenticeship. In: Primate Origins of Human Cognition and Behavior, ed. T. Matzuzawa. Ziemke, Zlatev and Frank (in press). Springer-Verlag. James, W. (1890) The Principles of Psychology. New York: McNeill, D. (1992) Hand and Mind: What Gestures Henry Holt. Reveal about Thought. Chicago: University of Chi- Johansson, S. (2005) Origins of Language. Constraints on cago Press. Hypotheses. Amsterdam: Benjamins. Meltzoff, A. and Gopnik, A. (1993) The role of imita- Kendon, A. (1981) Geography of gesture. Semiotica, tion in understanding other persons and devel- 37: 129-163. oping a theory of mind. In: Understanding other Kita, S. and Özyürek, A. (2003) What does cross- Minds: Perspectives from Autism, ed. S. Baron- linguistic variation in semantic coordination of Cohen, H. Tager-Flusberg, and D. J. Cohen. Ox- speech and gesture reveal? Evidence for an inter- ford: Oxford University Press. face representation of spatial thinking and speak- Meltzoff, A. and Moore, M. K. (1977) Imitation of ing. Journal of Memory and Language, 48: 16-32. facial and manual gestures by human neonates. Knight, C. (2002) Language and revolutionary con- Science 198: 75-78. sciousness. In: The Transition to Language, ed. A. - (1983) Newborn infants imitate adult facial ges- Wray, 138-160. Oxford: Oxford University tures. Child Development, 54: 702-709. Press. - (1994) Imitation, memory, and the representa- Krause, M and Fouts, R. (1997) Chimpanzee (Pan tion of persons. Infant Behavior and Development troglodytes) pointing: Hand shapes, accuracy and 17: 83-99. the role of gaze. Journal of Comparative Psychology, Menzel, E. W. (1973) Chimpanzee spatial memory 111 (4): 330-336. organization, Science 182: 943-945. Laakso, A. (1993) Language equals mimesis plus - (1974) A group of chimpanzees in a 1-acre field: speech. Commentary to Donald: Origins of Mind. leadership and communication, In: Behavior of Behavioral and Brain Sciences, 14/4: 765-766. Nonhuman Primates, ed. A. M. Schrier and F. Landau, M. (1991) Narratives of Human Evolution. New Stollnitz. Academic. Haven: Yale University Press. Merleau-Ponty, M. (1962 [1945]) Phenomenology of Lewis, D. K. (1969) Convention: A Philosophical Study. Perception. Routledge and Kegan Paul. Cambridge MA: Harvard University Press. - (2003) La Nature: Notes, Cours du College de France. Leavens, D. A., Hopkins, W. D. and Bard, K. (1996) Trans. Robert Vallier, Northwestern University Indexical and referential pointing in chimpanzees Press. (Pan troglodytes). Journal of Comparative Psychology, Miles, H. L. (1990) The cognitive foundations for 110: 346-353. reference in a signing orangutang. In: “Language” Leavens, D. A. and Hopkins, W. D. (1999) The and Intelligence in Monkeys and Apes, eds. S T. whole-hand point: The structure and function of 38 Bodily mimesis as “the missing link” in human cognitive evolution

Parker and K. R. Gibson. Cambridge University Psychology and Psychiatry and Allied Disciplines, 36: Press. 459-474. - (1999) Symbolic communication with and by Piaget, J. (1945) La Formation du Symbole Chez l'enfant. great apes. In: The Mentalities of Gorillas and Delachaux et Niestlé. Orangutans, ed. S. Taylor Parker, R.W. Mitchell - (1953) The Origin of Intelligence in the Child. Rout- and H. L. Miles. Cambridge University Press. ledge. Miles, H. L., Mitchell, R. W. and Harper, S. E. (1996) - (1954) The Construction of Reality in the Child. Basic Simon says: The development of imitation in an Books. enculturated orangutan. In: Reaching into Thought: Pierce, C. S. (1931-35) The Collected Papers of Charles The Minds of the Great Apes, ed. A. E. Russon, K. Sanders Peirce. Vols. 1-4. Cambridge, Mass.: Har- A. Bard. and S. T. Parker. Cambridge University vard University Press. Press. Pika, S., Liebal, K. and Tomasello, M. (2003) Gestural Mitchell, P. (1997) Introduction to Theory of Mind: Chil- communication in young gorillas (Gorilla gorilla): dren, Autism and Apes. Arnold. Gestural repertoir, learning, and use. American Mitchell, R. and Miles, H. L. (1993) Apes have mi- Journal of Primatology, 60: 95-111. metic culture. Commentary to Donald: Origins of - (2005) Gestural communication in subadult Mind. Behavioral and Brain Sciences 14/4: 768. bonobos (Pan paniscus): Repetoire and use. Ameri- Myowa-Yamakoshi, M. (2001) Evolutionary founda- can Journal of Primatology, 65: 39-61. tion and development of imitation. In: Primate Pinker, S. (1994) The Language Instinct. William Mor- Origins of Human Cognition and Behavior, ed. T. row. Matsuzawa. Springer-Verlag. Povinelli, D. J. (2000) Folk Physics for Apes: The Chim- Myowa-Yamakoshi, M. and Matsuzawa, T. (1999) panzee’s Theory of How the World Works. Oxford Factors influencing imitation of manipulatory University Press. actions in chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 113/2: 128-36. Povinelli, D. J. and Davis, D. R. (1994) Differences Myowa-Yamakoshi, M., Tomonaga, M., Tanaka, M. between chimanzees (Pan troglodytes) and humans and Matsuzawa, T. (2004) Imitation in neonatal (Homo sapiens) in the resting state of the index chimpanzees (Pan troglodytes). Developmental Sci- finger: Implications for pointing. Journal of Com- ence, 7/4: 437-42. parative Psychology, 108: 134-139. Nagell, K., Olguin, R. S. and Tomasello, M. (1993) Premack, D. (1988) “Does the chimpanzee have a Processes of social learning in the tool use of theory of mind?” revisted. In Machiavellian Intelli- chimpanzees (Pan troglodytes) and human chil- gence. Social Expertise and the Evolution of Intellect in dren (Homo sapiens). Journal of Comparative Psychol- Monkeys, Apes and Humans, ed. R.W. Byrne and A. ogy, 107/2: 174-86. Whiten, 160-179. Oxford: Clarendon Press. Nelson, K. (1996) Language in Cognitive Development. The Preston, S. and de Waal, F. (2002) Empathy: Its ulti- Emergence of the Mediated Mind. Cambridge Uni- mate and proximal bases, Behavioral and Brain Sci- versity Press. ences, 25: 1-72. Nelson, K. and Shaw, L. K. (2002) Developing a Reddy, V. (2001) Teasing, joking and mucking about socially shared symbolic system. In: Language, in the first year. In: Natural Theories of Mind, ed. Literacy and Cognitive Development, ed. J. Byrnes A. Whiten. Blackwell. and E. Amseli, Lawrence Erlbaum. - (2003) On being an object of attention: Implica- Osvath, M. and Gärdenfors, P. (2005) Oldowan tions of self-other-consciousness. Trends in Cogni- culture and the evolution of anticipatory cogni- tive Science, 7/9: 397-402. tion. Lund University Cognitive Studies 122, Lund. - (2005) Before the “third element”: Understanding Patterson, F. (1978) The gestures of a gorilla: Lan- attention to self. In Naomi Klein (ed.) Joint Atten- guage acquisition in another pongid. Brain and tion, 85-109. Oxford: Oxford University Press. Language, 5: 72-97. Rizzolatti, G., Fadiga, L., Gallese, V., and Fogassi, L. - (1980) Innovative use of language in a gorilla: A (1996) Premotor cortex and the recognition of case study. In: Children’s Language, Vol 2, ed., K. motor actions. Cognitive Brain Research, 3, 131-141. Nelson 497-561. New York: Garnder Press. Rizzolatti, G. and Arbib, M. (1998) Language within Perner, J. (1991) Understanding the Representational Mind. our grasp. Trends in Neurosciences, 21: 188-194. Cambridge, MA: MIT Press. Robinson, P. (1996) Consciousness, Rules and Instructed Perner, J., Leekam, S. and Wimmer, H. (1987) Three- Second Language Acquisition. Peter Lang. year-old’s difficulty with false belief: the case for Russon, A. E, and Galdikas, B. M. F. (1993) Imitation a conceptual deficit, British Journal of Developmen- in free-ranging rehabilitant orangutans (Pongo pyg- tal Psychology, 5: 125-137. maeus). Journal of Comparative Psychology, 107/2 : Peterson, C. C. and Siegal, M. (1995) Deafness, con- 147-61. versation and the theory of mind. Journal of Child Saussure, F. de (1916) Cours de Linguistique Générale [Course in General Linguistics]. Payot. Zlatev, Persson and Gärdenfors, Lund University Cognitive Science 121, (2005). 39

Savage-Rumbaugh, S. E., Wilkerson, B. J. and Bake- - (1999) The development of spontaneous gestural man, R. (1977) Spontaneous gestural communi- communication in a group of zoo-living lowland cation among conspecifics in the pygmy chim- gorillas. In: The Mentalities of Gorillas and Orangu- panzee (Pan paniscus). In: Progress in Ape Research, tans – Comparative perspectives, ed. S. T. Parker, R. ed. G. H. Bourne. Oxford: Academic. W. Mitchell and H. Lyn Miles. Cambridge Uni- Savage-Rumbough, S. and Lewin. R. (1994) Kanzi: The versity Press. Ape at the Brink of the Human Mind. New York: Tanner, J. E. (2004) Gestural phrases and gestural John Wiley. exchanges by a pair of zoo-living lowland goril- Savage-Rumbaugh, S., Shanker, S., and Taylor, T. las. Gesture, 4/1: 1-24. (1998). Apes, Language and the Human Mind. Ox- Terrace, H. S., Petitto, L. A., Sanders, R. J. and Bever, ford: Oxford University Press. T. G. (1981) Science 211/4477: 87-88. Searle, J. (1992) The Rediscovery of the Mind. MIT Press. Thompson, R. K. R., Boysen, S. T. and Oden, D. L. Senghas, A., Kita, S. and Özyürek, A. (2004) Children (1997) Language-naive chimpanzees (Pan troglo- creating core properties of language: Evidence dytes) judge relations between relations in a con- from an emerging sign language in Nicaragua. ceptual matching-to-sample task. Journal of Ex- Science 305: 1779-1782. perimental Psychology, 23/1: 31-43. Sinha, C. (1988) Language and Representation. A Socio- Tomasello, M. (1992) First verbs: A Case Study of Early naturalistic Approach to Human Development. Harver- Grammatical Development. Cambridge University ster Press. Press. - (2003) The evolution of language: From signals - (1993) It’s imitation, not mimesis. Commentary to symbols to system. In: Evolution of Communica- to Donald: Origins of Mind. Behavioral and Brain tion systems: A Comparative Approach, ed. K. Oller Sciences 14/4: 771-772. and U. Griebel. MIT Press. - (1996) Do apes ape? In: Social Learning in Ani- Sonesson, G. (1989) Pictorial Concepts. Lund University mals: The Roots of Culture, ed. C. Heyes and B. Press. Galef. Oxford: Academic. - (2001) From semiosis to ecology. On the theory - (1999) The Cultural Origins of Human Cognition. of iconicity and its consequences for the ontol- Cambridge, MA: Harvard University Press. ogy of the Lifeworld. VISIO 6/2-3: 85-110. - (2003) Constructing a Language: A Asage-based The- - (in press) The “Symbolic species” revisited. Con- ory of Language Acquisition. Cambridge, MA: Har- siderations on the semiotic turn in cognitive sci- vard University Press. ence and biology. Tomasello, M. and Call, J. (1997): . Sperber, D. and Wilson, D. (1995) Relevance: Communi- Oxford University Press. cation and Cognition. Cambridge University Press. Tomasello, M., Call, J. and Gluckman, A. (1997) Stern, D. (2000) The Interpersonal World of the Infant: A Comprehension of novel communicative signs View from Psychoanalysis and Developmental Psychol- by apes and human children. Child Development, ogy. Basic Books. 68: 1067-1080. Stoinski, T. S. and Whiten, A. (2003) Social learning Tomasello, M., Call, J. and Hare, B. (2003) Chimpan- by orangutans (Pongo abelii and Pongo pygmaeus). zees understand psychological states – the ques- Journal of Comparative Psychology, 117/3: 272-82. tion is which ones and to what extent. Trends in Stoinski, T. S., Wrate, J. L., Ure, N. and Whiten, Cognitive Sciences, 7/4: 153-156. A. (2001) Imitative learning by captive western Tomasello, M., Call, J., Nagell, K., Olguin, R. lowland gorillas (Gorilla gorilla gorilla) in a simu- and Carpenter, M. (1994) The learning and use lated food-processing task. Journal of Comparative of gestural signals by young chimpanzees: A Psychology, 115/3: 272-81. trans-generational study. Primates, 35/2: 137-154. Stokes, E. J. and Byrne, R. W. (2001) Cognitive ca- Tomasello, M., Carpenter, M., Call, J., Behne, T., and pacities for behavioural flexibility in wild chim- Moll, H. (in press) Understanding and sharing in- panzees (Pan troglodytes): the effect of snare injury tentions: The origins of cultural cognition. Behav- on complex manual food processing. Animal ioral and Brain Sciences. Cognition, 4: 11-28. Tomasello, M., Kruger, A. and Ratner, H. (1993) Stokoe, W. (1960) Sign language structure: An outline Cultural learning. Behavioral and Brain Sciences, 16, of the visual communication system of the 495-552. American deaf. Studies in Linguistics, Occasional Pa- Tomasello M., Savage-Rumbaugh S. and Kruger, per 8, University of Buffalo, New York. A. (1993) Imitative learning of actions on objects - (2001). Language in Hand. Why Sign Came before by children, chimpanzees, and enculturated Speech. Wahsington, D.C.: Gallaudet University chimpanzees. Child Development, 64: 1688-1795. Press. Tomasello, M. and Zuberbühler, K. (2002) Primate Tanner, J. E. and Byrne, R. W. (1996) Representation vocal and gestural communication. In: The Cogni- of action through iconic gesture in a captive low- tive Animal: Empirical and Theoretical Perspectives on land gorilla. Current Anthropology, 37/1: 162-73. Animal Cognition, ed. M. Bekoff, C. Allen and M. Burhgardt. MIT Press. 40 Bodily mimesis as “the missing link” in human cognitive evolution

Trevarthen, C. (1992) An infant’s motives for speak- Zlatev, J., Persson, T. and Gärdenfors, P. (in press) ing and thinking in the culture, In: The Dialogical Triadic bodily mimesis is the difference! Com- Alternative, ed. H. Wold. Oslo: Scandinavian Uni- mentary to Tomasello et al. (in press). versity Press. Zuberbühler, K. (2000) Interspecies semantic com- Varela, F., Thompson, E. and Rosch, E. (1991) The munication in two forest primates. Proceedings of Embodied Mind. Cognitive Science and Human Experi- the Royal Society of London. Series B Biological Sciences ence. Cambridge, MA: MIT Press. 267: 713-718. Veà, J. J. and Sabater-Pi, J. (1988) Spontaneous point- ing behaviour in the wild pygny chimpanzee (Pan paniscus). Folia Primatologica, 69: 289-290. von Hofsten, C. (1983). Catching skills in infancy. Journal of Experimental Psychology: Human Perception and Preference, 9: 75-85. Whiten, A. (1998) Imitation of the sequential struc- ture of actions by chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 112: 270-81. Whiten, A. and Byrne, R. W. (1988) Tactical decep- tion in 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 of artificial fruit processing in children (Homo sapiens) and chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 110/1: 3-14. Whiten, A., Horner, V., Litchfield, C. A. and Mar- shall-Pescini, S. (2004) How do apes ape? Learn- ing and Behaviour, 32/1: 36-52. Wilcox, S. (2004) Gesture and language: Cross- linguistic and historical data from signed lan- guages. Gesture, 4/1: 43-73. Wimmer, H., Hogrefe, G.-J. and Perner, J. (1988) Children’s understanding of informational access as a source of knowledge. Child Development, 59: 386-396. Wittgenstein, L. (1953) Philosophical Investigations. Basil Blackwell. Woll, B. and Kyle, J. (2004) Sign language. Encylopedia of Language and Linguistics. Oxford: Elsevier. Woodruff, G. and Premack, D. (1979) Intentional communication in the chimpanzee: the devel- opment of deception. Cognition, 7: 333-362. Vygotsky, L. S. (1962) Thought and Language. Cam- bridge MA: MIT Press. - (1978) Mind in Society. The Development of Higher Psychological Processes. Harvard University Press. Zlatev, J. (2002) Mimesis: The “missing link” between signals 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 of a unified biocultural theory of meaning. Evolution of Communication, 4/2: 253-296. - (2005a) What’s in a schema? Bodily mimesis and the grounding of language. In: From Perception to Meaning: Image Schemas in Cognitive Linguistics, ed. B. Hampe and J. Grady. Mouton. - (2005b) The dependence of language on con- sciousness. Paper presented at the conference Towards a Science of Consciousness 2005, August 17- 20, Copenhagen.