Umys³ - Mózg - Sztuczna Inteligencja
In English - Origins of Language: Between Nature and Culture
Slawomir Wacewicz - 2006-11-07, 19:21
Temat postu: Origins of Language: Between Nature and Culture
*** TOMASZ KOMENDZINSKI
we invite you to take part in a discussion on the evolutionary and cultural origins of language. This thread is, in a way, an extension of my proseminar on this topic held at the Institute of Philosophy (UNC, Toruñ), and is based on texts by a number of specialists in this field.
Our Guest of Honour in this thread is Professor Jordan Zlatev of Lund University, Sweden, who has kindly agreed to answer questions by the users of our board, time permitting (please note bear in mind his tight schedule).
Also participating in the discussion will be a co-author of the first text, Tomas Persson, a doctoral student at the Department of Philosphy in Lund.
We also count on the participation of other Guests of Honour, namely Professor Merlin Donald of The Department of Psychology, Queen's University, Canada and Professor Daniel D. Hutto of Philosophy at the University of Hertfordshire, UK.
The first text for discussion is "Bodily mimesis as "the missing link" in human cognitive evolution" (2005. Lund: Lund University Cognitive Studies 121), by Jordan Zlatev, Tomas Persson, and Peter Gärdenfors from SEDSU Research Group at Lund University, Sweden.
Our in-class discussion of the text will extend over several weeks (we have only managed to cover a few pages today), but we will be posting questions here as we go.
In the meantime, anyone is welcome to join the exchange of ideas with their own questions or comments.
*** SLAWOMIR WACEWICZ
I take the liberty to start the discussion with a couple of questions. But first I'd like to express my appreciation of the text: it's very informative and also quite difficult by being so informationally dense. I subscribe to many of the ideas it contains, such as the focus on the underlying cognitive abilities rather than on the communicative code (as was the case with Hockett's famous design features); the idea that understanding others (thoery of mind) does not have to conform to a propositional format (as suggested by folk psychology), but might be based on more primary empathetic mechanisms; the appreciation of gestures rather than concentrating on oral communication; seeing sign use itself as a selcetion pressure; and others.
The first question is a warmup one. You cite 'neonatal face mirroring' as an example of proto-mimesis. In a popular (but good) book Mind Hacks it was stated that the direction of the imitation might not be that obvious after all: a study they quote (O'Toole, R., & Dubin, R. (1968)) found out that during feeding, 80% of the time it was actually the mother who copied the facial gestures of the infant. How would you comment on that?
The second question is about the imitation-emulation distinction. Apes do not seem to be capable of imitation proper (copying both the goal and the action itself). Is it reasonable to speculate that what they do is simply use the final state of the demonstrated process as a cue? I.e. in a manipulation task, the chief difficulty for them is to form the mental image of the goal (final state of this task = the puzzle as solved), and they only need to 'pay attention' to this final state, disregarding the demonstrated actions that led to it. (A prediction would be that an ape that is shown just the final state of a manipulation task would perform on it just as well as an ape that was shown the final state plus the action leading to it).
The third question is a more general and speculative one that I'd like to offer just in case there is a temporary 'slowdown' in the thread so all participants have some more time.
Do you think that our intuitions regarding what language is are valid in the study of the evolution of language? Or might it be, perhaps, that they are fundamentally 'marred' (in a way) by our extant cultural environment: e.g. we tend to think of language as a means of transferring abstract scientific knowledge, as a foundation of our technological society, etc.
But this might be totally unlike how things were in early hunter-gatherer societies back 100-200.000ya, where language might have been something very intimately tied to the surrounding reality. Also, we might fall prey to another bias that I'd call "Take any language - take English". E.g. I am told there are languages that lack recursion (Pinker and Jackendoff 2005). Do you think we should bear the above in mind, or are such reservations really marginal?
*** JORDAN ZLATEV
Dear Slawomir (if it is you who asked the questions... I am new to this)
Let me reply in the reverse order. Dan Everett (2005) has recently argued that Piranha (a language spoken in the Amazonas) not only lack recursion, but number concepts, myths, etc etc. First, note the difference between compositionality and recursion (Janckefoff makes it too) - you can have the first but not the second. Then, even if Everett may be overstating his claims, he nevered dared state that Piraha lacks CONVENTIONAL semantic knowledge. All the Piraha speakers (presumably) know what "boat" means in Piraha and know that the other Piraha speakers know this, (and that they know this). It is this third-order knowledge that the paper claims is crucial for be able to say that you "know Piraha". But we may have overstated ourselves in the paper... Knowing that X meaning "this" in a NORMATIVE way (implying that other speakers should know this as well) seems to be suffcient for accounting for COMMON semantic knowledge.
Concerning ape imitation, Tomas Persson is the "expert" in our paper. But my answer is that it is now yet clear what apes can and cannot do in this respect. Tomasello (1999) was much too schematic in this respect... We are currently carrying out experiment with Josep Call to test the limits of ape imitation.
Concerning neonatal imitation, it has been replicated by so many researchers, in so many different contexts, that I have little doubt that it is a human universal. But notice that by classing it as "proto-mimetic" we attribute to it less intentionality and other higher-order cognitive capacities than Meltsoff and Moore. On the other hand, ape neonatal imitation needs stronger support, since the publications in its favour are coming (only) from Myowa-Yamakoshi...
Finally - the kind of "mutual imitation" you mention is importnat, but I think that it occurs at a later stage, and in less careful procedures than the neonatal imitation work.
P.S. Sorry for all the typos in my reply... I am very busy this week, but will have time for better spelling, and for more reflective replies next weekend! :-)
*** SLAWOMIR WACEWICZ
Dear Prof Zlatev,
thank you for finding time to answer. Yes, the questions were mine (indeed I'm a bit hard to recognise on the photo [but so is Tomas Persson on his webpage :)]), but I hope others will join in soon. In the seminar, we're not yet even halfway through the text (not least because it's so informationally dense), so most ideas are not yet discussed and digested.
I hasten to say that my third question was a general one and only very loosely related to your text. I take it that Piraha is a valid human language (which, as you observed, necessarily involves compositionality and conventional semantic knowledge), and more importantly, that any neonate of this Amazon people is capable of mastering any human language, recursion posing no particular difficulty.
I only wanted to remark that generally in studying the evolution of language we might be (mis)guided by our very specific understanding of language.
Also the first question (the warmup one) was given more as a 'curiosity' than anything directly relevant to the text. By the way, I think you are right in recognising the importance of proto-mimesis. On the face of it, it looks insigificant and obvious ("ok so you just do what you see, big deal!"), but when one stops to think about it, even such a mechanical translation across modalities (from vision to motor commands) is anything but obvious.
Thank you once again for your answers; more detailed questions are bound to appear this week.
*** MERLIN DONALD
I am very pleased to join this discussion from New Haven, Conn. I have a good internet connection here, so I will take advantage of it!
I think the main reason why mimesis is so crucial to language is not only that it provides a "bottom-up" foundation for language invention and acquisition, but also represents the first solution to the main problem facing the central nervous system: its solipsism.
The only thing a brain can actually do, in terms of acting on the world, and escaping its solipsistic isolation, is to send out motor commands and move muscles. So the evolving brain could not communicate anything of its specific knowledge to anoher brain until it could find a way to re-enact or re-create its event-percepts in actions of the muscles whose meaning could be made clear to another brain.
That is what mimesis entails, in neurobiological terms: a map of event-perceptions projected onto the voluntary musculature, producing a pattern of action whose meaning can become transparent to another.
I don't think mimesis evolved in a simple step or saltation. It is a multiple cluster of skills, and it was probably the case that mimetic mastery of the whole body came first, to support the systematic rehearsa of skills. Species Homo is, above all, a skilled creature who acquires and refines dozens of skills during a lifetime.
Voco-mimesis, a very specialized cluster related directy to the evolution of speech, came later, as testified by the fact that language itself is amodal, while the normal infant is clearly biased toward speech. Vocomimesis was the precursor to prosodic vocalization, and a precondition for phonological and morphophonetic invention.
Apes and monkeys may not be good imitators in Tomsello's terms, but I think imitation is not digital, that is, either on or off. Rather, there is a mimetic gradient on which humans fall at one end, primates in the middle, and most other mammals at the other end. Songbirds are mimetic, but only in one modality; this is a clear case of a parallel modular evolution of a specialized kind of mimetic capacity.
In any case, it is hard to imagine an evolutionary scenario whereby the order could have been different.
However, that did not prevent a number of critics (usually from a linguistic background) in 1991 telling me that language must have preceded mimesis, rather than following its emergence! I think that line of argument has died down, but the top-down bias was so strong in that discipline then that people actually expected the nervous system to function like a computer, forgetting that, on the contrary, the fuzzy nonsymbolic neural-net systems of the brain managed somehow to evolve symbolic skills despite, not because of, their basic design.
*** SLAWOMIR WACEWICZ
Dear Prof. Donald,
welcome to the board; we are very pleased to have you here as our Guest.
If I may chip in one comment: I think that a lot of disagreement in the field (i.e. evolution of language) is caused by the parties talking at cross purposes. There seems to be a set of beliefs tied to a particular theoretical perspective that pretty much comes as a package. Generative linguists tend to be nativistic and modularistic, seeing language as relatively autonomous and isolated from the rest of cognition. I think this is because when they talk about 'language', what they have in mind is the generative machinery, especially recursion (which is also easier to formalise than semantics). Both their theoretical apparatus and their explanatory goal are mostly related to syntax, which is why they fail to recognise the crucial importance of more basic steps on the 'road to language'.
On the other hand, they are probably right in saying that the emergence of generative syntax cannot be smoothed over.
Also, they see language as their theoretical remit... so we have to excuse their tendency to admonish 'nonsepcialists' (as they perceive nonlinguists)...
*** TOMAS PERSSON
Hello all, and especially Slawomir whose question (31 march) on imitation and emulation I will try to answer/expand upon.
The definition of imitation which demands copying both goals and actions to bring about those goals is Tomasello's definition where goals are being the goals in the mind of the model, not the end states per se. In that case apes can fall short. But more often than not the goal and the end state is one and the same thing, and apes can clearly copy end states (which is the/a definition of emulation, although "copy" might be a questionable term here.)
In early experiments apes exclusively used their own methods to bring about end states (a further part in the/a definition of emulation). However, later work by Andrew Whiten and others have shown that actions leading up to end states also can be copied. The problem in that analysis, with regards to mimesis, is to separate copies of the behaviors of the objects involved from copies of the model's bodily actions on those objects. The former are different cases of emulation (there are many types of emulation, of which object movement re-enactment is one, and sequence copying another) and the latter is, I would say, bodily imitation (regardless of goals). Apes are not blind to bodies and they can imitate bodily movements in for example Simon Says games. Which leads me to the second part of Slawomir's comment.
I don't think apes ONLY pay attention to end states. How would they know that the end state is possible to bring about, without devine intervention, if all they can represent is an open box infront of them? I think their attention works full time and they probably learn a lot about a task by observing. It is just that in experimental set-ups there are so many variables for the subject to attend to. The problem in imitation seem to be to combine all the above competences which all works fine by themselves into one solution for a problem: Working towards an end state, structuring observed actions and copy bodily movements. If this is a cognitive or motivational problem I don't know.
There is an enlightening recent study by Victoria Horner and Whiten, "Causal knowledge and imitation/emulation switching in chimpanzees and children" (Animal Cognition 2005:8), which put the finger on the main problem I think, that emulation and imitation works side by side, and to prove imitation in apes one has to focus the task design on the motivating factors for imitation, i.e. making emulation costly. Human children, on the other hand, seem to find their motivation outside of the task. (Human adults, by the way, seem to use emulation more than do children.)
From the Horner & Whiten abstract:
"[...] These results suggest that emulation is the favoured strategy of chimpanzees when sufficient causal information is available. However, if such information is not available, chimpanzees are prone to employ a more comprehensive copy of an observed action. In contrast to the chimpanzees, children employed imitation to solve the task in both conditions, at the expense of efficiency. [...]"
*** SLAWOMIR WACEWICZ
I am very pleased you joined us, and thank you very much for a comprehensive reply.
At first glance it seems action copying sequence-by-sequence is cognitively less demanding that (other types of) emulation, where you need, as it were, to come up with an 'original' solution, to invent motor sequences 'of your own'. So, the finding that apes are much better at emulation is counterintuitive. My idea was that to apes, when they are shown the solution to a task, the end-state is the most salient part, and all their cognitive resources get allocated to it.
But if I read you correctly, this might be simply a matter of innate biases: "without an innate bias, why would you take pains to copy someone's movements in addition to the goal, if you can just focus on the goal?"
But on our part (seminar students) this is all armchair speculation; that's why we're so eager to hear from real experts.
*** JORDAN ZLATEV
Dear Merlin and All,
I am very glad that Prof. Donald is aboard. I still think that his 1991 book is the most inspiring book in cognitive science that I have read, and if it was not for reading it after my thesis in 1997, I would be doing very different work from the one I am doing now. I also think that Merlin's theory, and in particular the concept of mimesis has been underappreciated! Have a look at the program for EVOLANG 2006 in Rome next week! (One good thing is that you can download all the accepted papers, and not only the abstracts, for all but the plenaries.)
Many papers deal with GESTURES, which is also one of Tomasello's favourite topics nowadays: ape gestures being "dyadic" and "imperative", unlike those of children in the second year of life, etc. But I have not seen any paper (but ours) which refers to Merlin's work. When Merlin's book was presented as a BBS "precis" in 1993, many people attacked it, unjustly. Tomasello's commenatry was called (I believe) "It's not mimesis, it's imitation"... When I, Tomas Persson and Peter Gärdenfors managed to get in a commentary to the recent Tomasello et al 2005 in BBS on "Shared intentionality", we thought of calling it "It's not imitation, it's mimesis" - but it would probably not have been accepted then.
The thing is that even Tomasello now accepts that apes imitate (public discussion after pleanry lecture in Vancover 2005), but sees their difficulties in motivational, social-cogntion terms. The key idea of the paper that you are reading and discussing, is that the difference between apes and human beings is above all in terms of "triadic mimesis", which is - at least I believe - the first form in evolution in which the "sign function" came about. But what do we mean by these terms? I can understand if you are having difficulties in the first 10 pages or so...
I will be presenting these ideas in Rome next week, and for the purpose have come up with some new definitions, in part different from the ones in the paper. Can I try them out on you - now for the first time in public?
X stands for Y for S, in a way that:
(a) X is directly perceived by S, but non-thematic, Y is non-directly perceived by S, but thematic
(b) Asymmetrical (X ? Y, but not Y ? X)
(c) Differentiated: X is qualitatively different from Y for S
Relatively stable periods in either evolution and development (without presuming any isomorphism, or �recapitulation�)
characterized by the *onset* of a particular type of semiotic capacity,
subsuming and integrating all previous capacities.
A particular bodily act of cognition or communication is an act of bodily mimesis if and only if:
(a) It involves a cross-modal mapping between exteroception (normally dominated by vision) and proprioception (normally dominated by kinesthetics)
(b) It is under conscious control and corresponds to � either iconically or indexically � to some action, object or event, but at the same time are differentiated from it by the subject.
(c) The subject intends for the act to stand for some action, object or event for an addressee (and for the addressee to recognize this intention).
But not if:
(d) The act is conventional (normative) and
(e) divides (semi)compositionally into meaningful sub-acts that systematically relate to other similar acts.
The mimesis hierarchy:
1. Proto-mimesis: only (a), e.g. neonatal mirroring, contagion, mutual gaze ( cf. primary intersubjectivity, Trevarthen)
2. Dyadic mimesis: only (a) and (b), e.g. imitation, mirror self-recognition, shared attention
3. Triadic mimesis: only (a), (b) and (c): e.g. declarative pointing, iconic gestures, full joint attention (cf. protosign, Arbib)
4. Post-mimesis1: (a), (b), (c) and (d): vocabulary (cf. protolanguage, Bickerton)
5. Post-mimesis2: (a), (b), (c), (d) and (e): e.g. grammar (cf. symbolic language, Deacon)
The major differences with the paper are that
- I have decided to remove them terms "representation", "communicative sign function", "symbolicity" in the definition of bodily mimesis - since they are so abstract, and have so many meanings for different people, that they confuse, more than help.
- I have separated two forms/stages of "post-mimesis" - since it is at least conceivable to have a rather simple form of "protolanguage" with very little systematicity (pace Deacon 1997, and my earlier definitions of "symbols")
- I have take away "imperative pointing" as an example of Dyadic mimesis - since it always confuses people how something the per definition involves three intities can be "dyadic" (the answer is that I wished to offer is that S points to X and looks to Y - but these form 2 overlapping dyads and not a real referential triangle: S is NOT pointing out X FOR Y, which would make it clearly triadic, but aslo declarative). But I may have to rethink this...
So to finish for now - I think that Prof. Donald's theory - also on other aspects than mimesis - has great potentials for evolutionary studies and for cognitive science. Our attempt is mostly to tease appart differnet forms/stages of mimesis (in semiotic terms), to link these more explicitly to the body (hence the term "bodily" mimesis), and to explore the transitions beteen the stages (in both evolution and ontogeny) in some more detail.
*** SLAWOMIR WACEWICZ
Dear Prof. Zlatev,
everyone seems to be taking their time, so I'll try to offer a short comment [I'm a bit pressed for time at this particular moment; nb. I'm going to Rome, too, and hope we meet there].
Although my vote goes to Terrence Deacon's Symbolic Species, I also think Prof. Donald's concept of mimesis to be groundbreaking - although I only read his papers and fragments of A Mind So Rare.
As to the definition of 'sign', my impression is: most of the work seems to be done by the opposition 'thematic-nonthematic', but the terms might be a bit unclear.
As to the cross-modal mapping, my gut feeling is less stress should be laid on proprioception (which I construe as somehow passive) and more on motor programs - but I'm no expert on how these two are differentiated and interact, so I'm only speaking of my intuitions.
To me, removing 'loaded' terms like 'representation' is indeed helpful for the reasons you mentioned. This extends to terms outside your list, with 'intentionality' being particularly confusing: it has a distinct, different meaning in philosophy that I feel is often conflated with the more 'everyday' use.
As to protolanguage, I think Bickerton's protolanguage, as I remember it, is in fact symbolic (it's language minus generative syntax). I also think Deacon is right when he says symbolicity of the signs arises partly (but indispensably) from the system of their mutual relations, so some simple combinatoriality is a must.
As to imperative pointing. I agree that the terms dyadic - triadic can be counterintuitive at times. (and might be worth relinquishing). But undoubtedly there is a qualitative difference between imperative and declarative pointing - this, by the way was first made clear to me when I read texts on (the lack of) ape cooperation by Peter Gärdenfors. To me, this difference can be accurately captured by a picture similar to that on page 17 of your text.
Dear Prof. Zlatev,
My question concentrates on the concept of imitation, or better - imitation skill. In the texts concerning language development (not only in a phylogenetic, but also in an ontogenetic sense) we often read about infants' imitation skills. One proof that is offered is that infants can imitate our mimics, such as smiling. I'm not so sure if this ability is a sufficient explanation for imitation skills. Perhaps it is an exageration? I think it
would be much safer if we looked for the origins of imitational skill in theory of mind development, at the age of 2 or 3. Therefore, we should treat an infant's smile (as a response to our smile) as a result of some genetic program, not as an imitaional skill (the child has no skill in repeating specific movements or sounds). Its purpose is to build a specific relation between mother and child, but nothing more. The child has no idea about what it is doing and why. As evidence, we could point
to the research by Eibl-Eibesfeldt, which showed that blind infants can look (with their blind eyes) at the mother's face, which is often crucial for "a mother's love". There is no doubt that it is a fully automatic and non-conscious act. Why do we have to use different terms to describe an infant's smile and mimics? Perhaps there's nothing more to it than a simple Stimulus-Reaction scheme, which is a complete oposition to imitation?
*** JORDAN ZLATEV
Dear Slawek and Tomek,
I am sorry for the long delay, but after coming back from EVOLANG had many local deadlines. I wonder what impression from Rome Slawek brought back, but since both he and Tomasz (in an email) asked me of my impressions, I have to say that I am rather disappointed. There were TOO MANY modelling contributions, while basic issues, such as "what is language", what is semantics, what are the capacities of apes and prelinguistic children etc etc are left unanswered. Since I hold that language is "a conventional (normative) symbolic system for communication and thought", if you have not said anything about the capacity to share conventions and social norms, you have not said anything (more interesting) about language, or its evolution. Most contributions, including those of Hurford on "protopropositions" would fall in this category, for me. Knight's (in)famous "mensturation-masking theory" does deal with norms, but in a rather baroque way - why so much focus on this particular "just so story", especially since body painting appeared so late.
The best presentation was that of Fitch - that we need to combine moe evidence from different sources - but also the most general one. Note that he stated, and I asked him the pause to confirm it, that "Faculty of Language Narrow" can turn out to be (a) not specific not humans or (b) not specific to language - and that even Chomsky accepts this! So we are back to square 1.
Given all this - I am quite happy with our "bodily mimesis" model... :-)
Here are some specific replies:
1. The definition of sign: this is based on semiotic terminology which I admit is not always too clear... But I work with Göran Sonesson, who I believe is one of the clearest semioticians there is. You can download his paper on "The embodiment of meaning" from the SEDSU homesite, where he clarifies the sign-concept that we use here in Lund. If you think this is unclear - just go to Pierce! BTW "?" should be an arrow pointing to the right...
2. Cross-modal mapping: Gallese was in the audience when I presented the model, and in the question session claimed that cross-modality is more general capacity of the brain, and need not imply mirror neurons. I don't know - the way they were originally presented, MNs did involve mappings between own actions and perceived actions, which is more specific. But now they have explanded to all over the brain, and involve pains, audio-visual mappings (those which fire when you either see or just hear a peanut being crushed etc). So I wonder if we are not talking about cross-modal mappings in the general case (say what allows infants to recognize visually the pacifier they have sucked on) and action-perception mapping being a special case of this.
3. Proprioception - notice that I do not bind myself to any particular model of proprioception. Kinaethethics is inportant, but so is haptic sense, and even visual sense of one's own body. Maybe Gibson is right that proprioception is just an aspect of embodied perception itself... The point to emphasize it is ... consciousness. I am interested in the qualitative experience of action, and then it cannot be a linking directly to "motor programs", as Slawek suggests. If that was all there is, mimesis would remain on the "proto-mimetic" level.
4. Intentionality is probably THE concept in cognitive science, and I agree with Slawek that it should be handled with care. But I don't agree that the "phylosophical sense" is completely seperate from the "psychological sense" or the everyday sense. It is a matter of polysemy rather than homonymy, as a linguist would say. (Chris Sinha has a recent paper relating the senses...) In all cases we have form of Bertrano's "directedness of the mental" - directedness to a goal, a referent, a communicative goal. I use it the last sense in the definition of triadic mimesis. Remember Grice 1957 and his definition of "Meaning", which Sperber and Wilson 1986 picked up in Relevance: A communicative intention is an intention to have the informative intention recognized, roughly. Ergo: a third-level intention (I intend that you recognize my intention). I think this is both a prerequiste from true language, and a futher developmed by it. (A chicken and egg story...)
5. "Protolanguage" - note that in my latest definition of the mimesis hierarchy I split up (d) and (e). Unlike Deacon, and Zlatev (2003) I think that it is possible to have symbols with VERY LITTLE system there. In fact that is what the apes are learning (or approximating). The story that Deacon gives in Chapter 2 on who the chimps Sherman and Austin "learned symbols" does not correspond to the story that Savage-Rumbaugh and Lewin 1994 give - in did not have to do with them grasping any "internal symbolic relations" but rather with getting the communicative point, as it were, and being able to form (proto-)conventions. At least that is my understanding now...
Now it may be an empirical fact that Proto-language is not stable - if the iconic/indexical ground is lost, then internal relations may be the only way to preserve it from disintigrating. But we wish to make this hypothesis testable, and not part of our definition.
6. Imitation. Tomek - I really don't see how you can suggest that imitation, even TRUE imitation, comes at the age of 2-3 years... I can understand that one can be suspitious of the first smiles and mouth protrusions, and I agree that Meltzoff and Moore tend to give the neonatal imitation a VERY rich interpretation. But even for what we called "proto-mimetic" imiation, roughly untill the age of 9 months, there are so many behaviours and modalities involved that it is impossible to explain them as "innate responses" or SR-pairings. I am not an expert on infant cognition, but reading for example Stern and Bruner, not to mention Trevarthen, has convinced me that the infant is an intersubjectively attuned agent, communicating first emotional states, and then referential information through the resources that are available - vocal and gestural. We know from Tomasello et al that 14 month old babies "inform" others of where things are, and where did they learn the pointing gestures from? While many people have suggested that the come from "ontogentic ritualization" or some other non-imitation mechanism, no one has succeeded. I think it is through triadic mimesis - imitating communicative signs. Note also that before they can talk, children can understand and perform iconic gestures (Acredolo et al).
So my general strategy is: rather than sharply dividing betwene imitation and non-imitation, or language and non-language, or "theory of mind" (a term I hate) and non theory of mind - to look for developmental and evolutionary progresssions. I think that the mimesis hiearchy model has helped us find several "missing links" in all these cases, and futhermore to bind them to a single social-cognitive capacity: bodily mimesis. Note the twist that we come to at the end of the paper - it is not "theory of mind" or "imitation" that is so diffuclt for apes as intentional gestural communication. So at least my evolutonary hypothesis is that it was (communicative) triadic mimesis that privided the niche for selecting for better imitation and intersubjectivity, rather than vice versa.
I have to rush home now - so please excuse the many typos I have certainly committed.
*** SLAWOMIR WACEWICZ
I'd like to ask just three questions regarding the texts, and later I'll try to add a brief comment on the conference.
1. In the text, you stick to the traditional notion 'displacement'. I'm curious, because in a few of his texts, Gärdenfors prefers to speak of 'detachment', which builds upon displacement but adds the possibility of referents not only being distant in space/time but also not existing at all (imaginary/hypothetical/negated, etc.).
2. It seems the bottom line of the text is there are two 'thresholds' on the evolutionary path from the ape ancestor to the articulate sapient humans of today: 1. diadic mimesis -> triadic mimesis (communicative intention), and 2. mimesis -> post-mimesis (conventionality). Which, would you say, is more 'significant' (in any sense)?
3. The last question is a tricky one. Since it appears that enculturated apes are in principle capable of 'mastering' post-mimesis; what is the qualitative difference between them and us - because that some such difference exists is unquestionable? Wouldn't it follow that this ultimate dividing line ('Rubicon that no brute dare cross' ) is in fact syntax? And if so - that syntactocentric nativists like Pinker, Chomsky and Bickerton are right after all?
Also I'd like to add two short remarks on the replies:
(2) cross-modality (unless clearly defined) is potentially ambiguous. E.g. synaesthesia can be seen as cross-modal, without involving the mirror system. Such intuitions aren't relevant to mimesis, but may provoke confusion.
(5) I still think that it's not possible to have just one or two symbols and that much of a symbol's meaning arises from its paradigmatic and (potential) syntagmatic relations - but of course I can't 'prove' it in any way.
As to the conference, my feelings are essentially the same. Far too much modelling (if this indeed is the way the field develops, why not rename the conference, and leave the general label [evolution of language] for general topics?). I particularly liked the presentations by Tomasello, Seyfarth and Fitch, while I particularly disliked the one by Knight.
The FLN/FLB distinction is an important one, only no-one knows what it is. The original Hauser/Chomsky/Fitch text was extremely ambiguous on it, and I have a very strong suspicion each of the authors has a different idea. One interesting presentation related to this was Anna Parker's.
PS: I almost forgot. I am really greatful for pointing out the difference between arbitrariness and conventionality. People sometimes say 'not all language is symbolic' because 'onomatopoeia etc. is iconic, and icons are no symbols'. I have always felt there is something profoundly wrong with such statements, but until now I have had no way of sorting it out.
*** JORDAN ZLATEV
As always you raise very pertinent points. And again, I am late in responding. But instead of doing one of my "quick replies" - with tons of typos which are there for all to read - let me take some more time, and do a proper reply.
But just a question: you refer to the "texts" - apart from Bodily mimesis text with Persson and Gärdenfors, do you also meen the "On intersubjectivity and mimetic schemas" text that I sent you by email. Is that made public to the group? Otherwise we should maybe discuss it by email... In any case, in it, and my entry on April 7th, I separated between Post-minesis 1 and Post-mimesis 2. Just so that people do not get confused, this is definitely not Bickerton's "protolanguage" + "grammar" distinction, but that between less-systematic (e.g. Kanzi) and very systematic (e.g. a la Deacon) symbol use. (This is in relation to your "tricky question").
Finally - on September 29 and 30, 2006 we will have a small conference here in Lund called "Comparison and Interaction of Semiotic Resourses in Evolution and Development", with Terry Deacon and Jean Mandler as plenary speakers. Participation will be free of charge, and there is relatively cheep accomodation in Lund (Hotel about 60 euro, Hostel about 20). So you, and anyone else in this forum are welcome to participate, however only as listeners. There will be more information on this on our home site SEDSU-Lund quite soon.
*** TOMASZ KOZLOWSKI
Dear Prof. Zlatev,
I have a question about a possibile connection between your theory and Noam Chomsky's Generative Grammar Theory. Do you think there is a reasonable chance to make such a combination? I wonder where you would place Chomsky's Generative Grammar Module: before or after post-mimesis period. In other words, is it possible to have post-mimetic abilities WITHOUT a generative grammar module?
Tomek Kozlowski, Nicolaus Copernicus University, Torun
*** JORDAN ZLATEV
Dear Slawek and Tomek – and all other readers of this forum. (But I wonder if there are others, why is it only Slawek and Tomek who pose questions… :-)
First, I am sorry for my long delay. May has been a long stretch of meetings, exams and deadlines, but luckily – it is ending. Here are my replies to the last 6 questions.
1. Displacement/detachment etc. It is true that in the text we refer to Hockett’s classical term “displacement”. But this does not take on any heavy theoretical burdens. The crucial term is rather differentiation (between expression and content within the sign) – which is the condition for dyadic mimesis. Think of the ability to recognize yourself (or anything else) in the mirror – you need to be able to appreciate both the similarities between mirror-image and your body-image AND their spatiotemporal distinction. Children “formally” pass Galup’s “mirror test” at the age of 21-24 months, but starting from 1 year they show signs of recognizing themselves – and that something has been added to their faces – even when they don’t try (or manage) to coordinate their hand movements to wipe it out. Here the “correspondence” between expression and content is on the basis of similarity, hence the ground (in the Pierceian sense) is one of iconicity. In the case of the transition from reaching to pointing, we also get differentiation, but the ground is one of indexicality. It is not that the child reaching does not sensomotorically distinguish between its body and the desired object, but rather that it is so “involved” in the activity to make any explicit distinction between activity and object nonexistent. In the case of pointing (both imperative and declarative) the child again appreciates both the proximity of its gesture and object, and their spatiotemporal separation, i.e. their differentiation. Note that in all these cases we have both X (expression) and Y (content) in the same situation, so strictly speaking there is no “displacement”. What the latter requires is for the content to be in the (long term, declarative) memory of the subject – but also in the case of communicative (triadic) mimesis for the subject to realize that it can be not only in one’s memory, but in the addressee’s as well. So triadic mimesis with “displacement” does require some form of meta-representational capacity, but as Dan Hutto argues, this does not need to be a matter of “theory of mind” or “propositional attitudes”. What he calls “intentional attitudes”, evoking shared experiences (in the past) would suffice. Finally: I am not very happy with Peter Gärdenfors’s term “detachment”, since it applies to many phenomena and not primarily to signs (cf. “The detachment of thought”). It is too individualist and not sufficiently Vygotskyan for my approach. Like Lev Semeonovich, I would claim that the social precedes the individual, and the overt precedes the covert in the child’s “cultural development”. So to Slawek’s question: I would predict an ontogentic progression: differentiaion > “displacement” > non-existing referents - “detachment” (The paper you read was not explicit on these points since Peter and I had to compromise. But we have come to the conclusion since then that our differences are big enough to require separate elaborations…)
2. The transitions are rather three, or even four – given the reformulation of the mimesis hierarchy in 5 stages, separating between post-mimesis1 and post-mimesis2. The reason we downplayed the importance of the first – between proto and dyadic mimesis – is that is seems that the differences between apes and humans in dyadic mimesis are more quantitative than qualitative. But still, I agree with Arbib (2005) that e.g. “complex imitation” seems to have been selected for in hominid evolution (the transition between his stages 3 and 4). To your question: I (and the paper) would emphasise the transition to triadic mimesis (and communicative intention), because here first the differences between “us” and “them” are qualitative. Think of the ease with which one year (plus) old children begin to intentionally gesture for and establish joint attention with others – and the difficulty in getting apes to do so, or to understand the intentional gestures of others. We are currently con ducting experiments with picture understanding with infants and apes – and the results are similar. In simple tasks children seem to understand the sign function (X stands for Y) very early, while apes just don’t get it. So I (and Göran Sonesson) would probably be willing to call Homo Sapiens “The Signing Species”… As for conventionality – given (a) an advanced imitation capacity, (b) advanced declarative memory and (c) sign function, this would probably require a relatively smaller transition. Notice that triadic mimesis furthermore another prerequisite for forming conventions: (d) third order mentality (though not concerning beliefs, which is post-mimetic). One problem, however, is if add (as I do now) normativity to the definition of post-mimesis 1. This means understanding that it not only is “common” for X to mean Y, but that it is correct to do so, rather than say Z (even if it does not make any pragmatic difference in a given situation). Children begin to “correct” oth ers from 2 years of age – if you call a ball a “stone” and they will protest – if they don’t think you are playing some sort of pretend game. I know of no evidence that Kanzi would do so…
3. Now to the “tricky question”. First – I don’t think that we ever said, or even implied that apes are capable of “in principle capable of mastering post-mimesis”. If anything we said the opposite – that they are “in principle” capable of achieving some degree of triadic mimesis (Koko, Kanzi and the other enculturants) but their failures in developing grammar, solving false-belief tasks and learning new (sign language) signs through imitation rather than molding – and more generally using the protolanguage that they have acquired for different communicative functions (informing, asking, promising etc) and producing narratives showed that (full) post-mimesis was beyond their “ZPD” - and analogously was beyond the “zone of proximal evolution” (Donald 2001) for the common ancestor. But I agree that the paper was not sufficiently clear on this point, and this is one reason that I now distinguish between post-mimesis1 (PM1) and postmimesis2 (PM2). In separating between PM1 and PM2, we can ask – along with Slawek – whether the “Rubicon” is conventionality-normativity or combinatorics-compositionality. As I wrote above, the difference between triadic mimesis and PM1 is above all normativity, but (as Dan notes) this by no means implies the evolution of a “normativity module” as suggested by Caruthers. In reading carefully Savage-Rumbaugh and Lewin (1994) is seems that after (very) much training Sherman and Austin were induced to understand – not the “internal relations of symbols” as Deacon claims – but something of the conventionality of reference (as we also suggest in the paper). So I am willing in part to accept Slawek’s challenge and say that IF we have to place a Rubicon somewhere, it would probably be between PM1 and PM2. But even if we accept this hypothesis it in no way validates the claims of the “innate syntax” people since grammar is complex systems of signs, and NOT a formal system for moving around meaningl ess “symbols” like TNS, AGR etc. But the question is: do we really have to look for a strict “Rubicon”? It is one thing to say that apes fail to acquire grammar and produce narratives – which almost everyone agrees with – but quite another to say that this is because they lack a “grammar (or whatever) module”. The latter is an error of “misplaced concreteness” and of setting up “discontinuities” in both evolution and development which are impossible to match to the empirical data of neuroscience, evolutionary theory, child development etc. So instead of placing a discrete boundary between “us” and “them” I prefer to visualize the differences in terms of “developmental profiles”. Normal human children progress through the stages of the mimesis hierarchy chronologically. In the case of apes, who have difficulties with the features (a)-(d) listed above, and hence triadic mimesis, massive exposure to human culture and language can (in certain circumstances) induce a rather primitive form of post-mimesis1 (some degree of conventionality, but hardly normativity). In a sense, triadic mimesis is “bypassed”. Now, is it surprising that under these circumstances they fail to develop lexica of thousands of words, grammar and narrative? Despite reservation, I think enculturated apes’ language learning can be compared with those of severe language impaired children with austim. After much intervention these come out of “mutism” and learn some one-to-one mappings between labels and objects, and even seem to get the sign funct ion: they start to ask for “names for things”. But they don’t develop grammar and are far from any forms of narratives.
4. “Cross-modality”. I completely agree that this is to general. That is one reason why I removed it as technical term for (a) in the definition of mimesis (and condition for proto-mimesis) in more recent papers. Notice that a mapping between proporioception and exteroception is not just “any kind of” cross-modality, but one which involves correlating the embodied self and the environment. There is neuroscientific work (e.g. by Ramachandran) which has specific hypotheses for this is carried out – involving not only centers for “proprioception” but the limbic system, and their mappings to frontal and pre-frontal areas. Interestingly, these – and related mirror neuron systems – seem to be the areas that are disfunctioning in (many cases of) autism. That is why during the past months I have been reading about autism, and will be conducting experiments on imitation, picture understanding and language with Pam Heaton in London. Our working hypothesis is that we can “explain” autism as an impairment in bodily mimesis, and in the process of doing so will spell out the kind of cross-modality that is involved in bodily mimesis.
5. “One or two symbols” - note that whether this is possible of not is not so much an empirical but a conceptual (definitional) issue. If symbols are defined as signs co-defined in terms of their internal relationships (as done by Deacon, and Saussure before him) then it becomes “impossible” by definition. But if symbols are conventional mappings between expressions and contents – whether these are arbitrary or not (as we agree) – then there is no reason why a limited inventory of symbols should be impossible. In fact – this is exactly what Sherman, Austin, Kanzi and some autistic children appear to achieve: They get what Deacon calls the “indexical” relationships, but not the system of relations. Since it is an empirical possibility – and seeming factuality – to have PM1 without PM2, which should not conflate the two in our definition of symbolicity, as I did myself before.
6. I view Chomsky’s many different models – there as been at least 6 or 7 of them from 1956 – as deeply misguided… None of Chomsky’s arguments for an innate grammar (LAD, UG, FLN etc) holds, and the more we know about the brain, evolution and development, the more implausible they become. The last bastion seems to be “recursivity”, but there are both semiotic (Deacon), cognitive (Arbib) and functional (Harder) ways to explain this in language. Furthermore, I would claim with Arbib, Tomasello and others that the complexity of human languages is mainly “post-biological”. Homo Sapiens of 200,000 years ago possibly (and probably) had a simple language with terms for objects, people, locations and actions – but all the wonderful things like tenses, agreement, modality, cases, prepositions etc. that linguists are fascinated by came about as a result of cultural evolution and grammaticalization – without any genetic mutations or recombinations. The big mistake of Chomsky and many other linguists is that when they describe grammatical patterns in and across languages, they think they are uncovering hidden unconscious structures and processes, while they are simply producing decriptions of the linguistic/cultural conventions. Explanation of any universals, or near-universals need to appeal to functional and cognitive factors, as done within functional and cognitive linguistics. There is much to be done in this respect, and I think that we will soon think of Chomsky’s large influence in linguistics in the second half of the 20th century as a historical curiosity.
*** SLAWOMIR WACEWICZ
Dear Prof. Zlatev,
Thank you for the reply. Yes, the 'mastering' was a mistake (and I've noticed that I generally overuse this word), what I meant was something like 'acquire the basics of'. I've just thought of yet another question, possible the last one:
|But still, I agree with Arbib (2005) that e.g. “complex imitation” seems to have been selected for in hominid evolution |
What do you think the evolutionary pressures might have been?
Derek Bickerton is adamant about his now-unpopular claim that "the cognitive prime mover" in hominid evolution was something related to the hominid ecology rather than broadly understood 'social realm'. Specifically, he points to hunting (and dismisses 'Machiavellian intelligence'). His positive evidence, I think, is weak (I listed a few criticisms in a review of Lingua ex Machina), but there is one bit of negative evidence that one can't just shrug off:
if what was important were the pressures from 'within the society' [I take it such is the motivation for mimesis], how come only the genus Homo has ultimately developed language, of all the primate species that have faced and still face intense selection pressures on social skills?
*** SLAWOMIR WACEWICZ
And it is also high time to start asking questions to our next Guest of Honor, prof. Daniel Hutto.
Dear Prof. Hutto,
let me begin with a simple (maybe even a bit naive) question for a start:
throughout your paper, you use the acronym TOMM alongside TOM. Does the other 'M' stand for 'module' (as it usually does in evolutionary psychology), and if so, does it suggest a modular understanding of the mind in any way?
Daniel Hutto - 2007-06-10, 06:33
Temat postu: Re: Origins of Language: Between Nature and Culture
Sorry for the painfully slow reply - my life has been busy of late.
The extra M in ToMM can be understood as 'module' - though notoriously, there is much debate about what features define modules (see for example Carruthers 2003, 2004. He attempts to go beyond Fodor's version which insist on information encaspulation). In my hands the acronym is used to stand for 'Theory of Mind Mechanism', where 'Theory of Mind' denotes a class of abilities and 'Mechanism' refers to any posited device that allegedly makes use of mental state representations when doing its work. As such I have in my sights any kind of Inherited Mindreading Device (or IMD). This would include, for example, the Early/Low-Level Minreading Systems posited by Stich and Nichols (2003) and Goldman (2006).
Officially, then, it would exclude 'smart behaviour' reading as opposed to mind reading devices - such as those posited by Povinelli and Vonk (2004) as part of their re-interpretation hypothesis. But since I have a master argument against representationalism tout court I doubt the existence of such mechanisms too.