Thursday, June 26, 2008

Dieter Wunderlich on the Evolution of Language (Part 2): Mirror Neurons, Gestures, and Imitation

As I already said, according to Wunderlich’s review of the archaeological and primatological evidence, it is likely that homo erectus already possessed a substantial number of the design features that make human language possible. In his presentation, Wunderlich then goes on to speculate about the relationship between language evolution and mirror neurons, neurons that were first observed firing both when a monkey either saw someone perform a grasping movement or performed the grasping movement itself. For Wunderlich the mirror neurons system gives a possible explanation for two crucial features of language:

- Intentionality ( Speaker intends to produce a certain effect in the Hearer)

- Parity (Speaker and Hearer are able to seamlessly switch roles in conversation)

Mirror Neurons also make plausible a gestural origin of language, because (Arbib 2005, Corballis 2003):

- Chimps are much better at controlling their hands than controlling their vocal apparatus, indeed, chimps are able to learn some kind of sign language and other forms of symbolic communication (i.e. via symbol keyboards), but have problems with learning a vocal-based arbitrary communication system

- We still use gestures when talking, and, as Susan Goldin-Meadow has shown, these speech-accompanying gestures are part of an language-independent knowledge system, so that, for example, in some experimental settings children sometimes indicate the right answer with their gestures but still give the wrong answer verbally.

- There are a lot of sign languages spoken either by the deaf or in some religious communities, and these are just as morphosyntactically fully-fledged as any spoken language

- deaf children born to deaf parents learn sign language just as quickly as hearing children learn spoken language

- in deaf children born to hearing parents, who are thus unable to expose them sign language, there is even a phenomenon of so-called home sign, a kind of spontaneously produced “signed pidgin” ,with the intention to communicate and express needs and desires, which exhibits a very rudimentary grammatical and morphological structure, something which speaks for some kind of innate communicative structuring.

- Also, manual gestures often still bear a closer link to their original motivation, and although arbitrary, they may have the tendency to be more transparent as to their iconic motivation, which could make them easier to learn

- If you consider the cognitive implications of toolmaking described in the last post, gestural language would also be a more natural explication of these, especially because they can combine Topic and Comment in a very visual and illustrative manner, namely manual.

- It is also only a small step from ritualized gestures to symbols. In chimpanzees, for example, there is the phenomenon of ontogenetic ritualization, in which the demand to have access to the breast of her mother is ritualized by child and mother: At first the child pushes her mother’s arm away in order to get to the breast, but after a while she just has to touch her mother’s arm so that she lifts it for her child to be able to be breastfed. But this specialized gesture vanishes by the time the child doesn’t need to be breastfed anymore, and every mother-child generation has to develop this ritual anew. Humans somehow transcend this ritualistic confinement and turn behavior like that into long-lasting symbolic behaviors (see e.g. Tomasello 1999: 31f.)

Another advantage of the mirror neuron hypothesis is that it helps understand the instinctive imitation capacities of children, which can not only be seen as a motor of language acquisition and thus further language evolution, but also as a crucial factor in cultural evolution via iterative learning by members of successive generations, each time with slightly differing input. Indeed as recent experiments show, imitation may be one of the key cognitive specializations that sparked the success of the human evolutionary story. Particularly intriguing are experiments done by Victoria Horner and Andrew Whiten from the Unviersity of St. Andrews in Scotland, and Derek Lyons and his colleagues from Yale University: Children aged 3-4 and young wild-born chimpanzees were shown how to get a little toy turtle/ a reward either 1) out of a transparent puzzle box or 2) out of an opaque puzzle box. Here’s the catch: the right solution is much simpler as the one shown by the demonstrator, who includes a lot of unnecessary steps.

(You only have pull out the red thing at the bottom, to get at the reward, but the demonstrator also removes the tube on the tope and pushed a stick intot he box)

The stunning result: In this task, the chimps prove smarter than human children.

If the box is opaque, that is, if chimps and children are unable to look at the internal workings of the puzzle box and are thus unable to assess which steps are necessary and which are unnecessary, both copy all the actions done by the demonstrator.

If however, the puzzle box is transparent, and children and chimps are in principle able to make out which steps are unnecessary and which aren’t, chimps only copy the relevant actions – and human children still copy every action, including all the obviously unnecessary ones.

taken from Horner & Whiten 2005

This phenomenon is called overimitation. For most tasks in the real world whose causal or internal structure we don’t understand, it pays to closely match the behavior of a more knowledgeable person. However, although in general it is a very powerful learning strategy, it often carries a cost and, as it seems, in children this process is sometimes so automatic that it overrides any relevant causal reasoning about the task itself:

Children who observe an adult manipulating an unfamiliar object show a strong tendency to encode all of the adult’s purposeful actions as causally meaningful, revising their causal beliefs about the object accordingly.” (Lyons et al. 2007: 19756)

Whiten and Horner (2005: 179) concur:

“differences in the behaviour of chimpanzees and children can best be explained by a combination of a differential focus of attention on actions, results and goals, with the latter possibly influencing the interpretation of the actions of the demonstrator as purposeful.”

However, they also see this as part of a more general tendency of adhering to cultural conventions and norms (Whiten and Horner 2005: 164), something which, albeit in a rudimentary form, is also present in chimpanzees (Whiten et al. 2005: 164)

Thus it is certainly right to say that:

“Much of the success of our species rests on our ability to learn from others’ actions. From the simplest preverbal communication to the most complex adult expertise, a remarkable proportion of our abilities are learned by imitating those around us Imitation is a critical part of what makes us cognitively human and generally constitutes a significant advantage over our primate relatives.” (Lyons et al. 2007: 19751)

But is it right then to say that chimpanzees “Don't Give a Hoot About Unnecessary Actions”?

Well the problem is that the experiments of Whiten and Horner where done with wild chimpanzees.

But what about enculturated chimpanzees raised by humans?

Buttelman et al. (2007) tested just that. They used the “rational imitation” paradigm, which features two conditions a) the subjects are shown an action in which the specific manner of the action is not purposive but result from the demonstrator being occupied with something else, e.g. carrying something so that has to use his foot to turn on light. b) the subjects are shown an action in which the demonstrator chooses a specific manner of performance on purpose, e.g., he has his hands free but still chooses to turn on the light with his foot.

taken from Call & Tomasello 2008

Another instance of the paradigm is an experimental setup in which the subject observes someone sitting on a chair with a light-box on the table in front of her. The person has a blanket wrapped around her. Now in one condition a) her hands are occupied (she is shivering and holding the blanket) and she leans forward and illuminates the light with her head, in the other condition b) her hands are free and she leans forward and illuminates the light with her head.

taken from Gergeley et al. 2002

In the experiment done by Gergeley et al. (2002), most 14-Months old human children re-enacted the demonstrator's way to illuminate the light in condition b) but not in condition a). The explanation for this is that in a) they probably see the demonstrator as being forced to use her head because her hands are occupied, whereas in b) they see the demonstrator as rationally choosing to illuminate the light with her head, because had she wanted to use her hands, she would’ve done. They thus encode her performance as causally meaningful and purposive, and re-enact it.

Interestingly, in Buttelman et al’s (2007) study, in both experimental setups presented here, the enculturated chimpanzees also copied

the demonstrator’s action more often when he freely chose his action than when he was forced to use it by some constraint” (Buttelman et al. 2007: F37).
On a rich interpretation of these results, one could opt for the

“possibility that all great apes possess the ability to understand others’ intentions as rational choices of action plans to achieve goals, but that only some of them are able to show this understanding in these types of experiments due to special attentional or motivational qualities.” (Buttelman et al. 2007: F38).

Thus chimpanzees may even have a rudimentary Theory of Mind, which enables them to

"understand others in terms of a perception–goal psychology, as opposed to a full-fledged, human-like belief–desire psychology (Call & Tomasello 2008: 187)

As to the whole mirror neurons business, and Wunderlich’s remarks, I’m not sure how the socio-cognitive capacity of purpose-encoding rational imitation can bee seen as being motivated by a mirror neuron system, especially when it comes to language evolution. The language evolution link was motivated by the hypothesis that the premotor cortex of the macaque brain, which is the place where most mirror neuron activity takes place, was an evolutionary cousin, a homologue, to Broca’s Area in humans. But as recent research has shown, this is not the case, and Broca’s Area doesn’t show any mirror neuron activity. (Morrin & Grezes 2008).

I will return to Wunderlich’s presentation in my next post (If I proceed in this manner I’m going to spend the whole year writing about his presentation, given that today’s elaborated on a single slide and I’ve only reached page 10 of 29…)


Arbib, Michael A. (2005) From monkey-like action recognition to human language: An evolutionary framework for neurolinguistics. Behavioral and Brain Sciences, 28(2):105--124.

Buttelmann, David, Malinda Carpenter, Josep Call, and Michael Tomasello (2007). Enculturated chimpanzees imitate rationally. Developmental Science 10.4: F31-8. doi:DESC630.

Call, Josep, and Michael Tomasello. (2008): Does the chimpanzee have a theory of mind? 30 years later. Trends in Cognitive Sciences 12.5: 187-192. doi:10.1016/j.tics.2008.02.010.

Corballis, Michael C. (2003) From hand to mouth: The gestural origins of language. In M.H. Christiansen and S. Kirby, editors, Language Evolution: The States of the Art. Oxford University Press.

Gergely, György , Harold Bekkering, and Ildikó Király. 2002. Developmental psychology: Rational imitation in preverbal infants. Nature 415 (February 1): 755.

Horner, Victoria, and Andrew Whiten. (2005): Causal knowledge and imitation/emulation switching in chimpanzees (Pan troglodytes) and children (Homo sapiens). Animal Cognition 8.3: 164-81. doi:10.1007/s10071-004-0239-6

Lyons, Derek E., Andrew G. Young, and Frank C. Keil. (2007): The hidden structure of overimitation. Proceedings of the National Academy of Sciences 104, no. 50 (December 11): 19751-19756. doi:10.1073/pnas.0704452104

Morin, O., and J. Grezes. (2008) What is “mirror” in the premotor cortex? A review. Neurophysiologie Clinique/Clinical Neurophysiology 38, no. 3 (June): 189-195. doi:10.1016/j.neucli.2008.02.005.

Tomasello, Michael (1999): The Cultural Origins of Human Cognition. Cambridge, Massachusetts; London, England: Harvard University Press

Whiten, Andrew, Victoria Horner & Frans B. M. de Waal. (2005): “Conformity to Cultural Norms of Tool Use in Chimpanzees.” Nature 437: 737-740.

1 comment:

Marc Verhaegen said...

FYI, language origins & the littoral theory of human evolution:

When we analyse human speech into its elements, we get
1) early hominoid territorial song (cf gibbon duets) in Miocene swamp forests: music, dialog, tone, rhythm, sounds,
2) seafood consumption during the Pleistocene diaspora of archaic Homo along coasts & rivers:
- voluntary breathing for diving: from emotional to voluntary sound production (arbitrary meanings of sounds & later words),
- slippery seafood, suction feeding = ability to close the mouth at different places (labial, dental, palatal, velar etc.) = consonants,
- brain enlargement, facilitated by the abundance of brain-specific nutrients in seafood (eg, DHA).

Was Man More Aquatic in the Past?
Bentham Sci Publ ebook
Mario Vaneechoutte, Algis Kuliukas & Marc Verhaegen eds 2011

CHAPTER 12 - Seafood, Diving, Song and Speech
Mario Vaneechoutte, Stephen Munro & Marc Verhaegen

In this paper we present comparative data, suggesting that the various elements of human speech evolved at different times, and originally had different functions.
Recent work by Nishimura [1-6] shows that what is commonly known as the laryngeal descent actually evolved in a mosaic way in minimally 2 steps:
(a) a descent of the thyroid cartilage (Adam’s apple) relative to the hyoid (tongue bone), a descent which is also seen in non-human hominoids, and
(b) a descent of the hyoid bone relative to the palate, which is less obvious in non-human hominoids, and which is accentuated by the absence of prognathism in the short and flat human face.

Comparisons with other animals suggest that
(a) the 1st descent might be associated with loud and/or varied sound production, and that
(b) the 2d descent might be part of an adaptation to eating seafoods, such as shell fish, which can be sucked into the mouth and swallowed without chewing, even under water.

We argue that the origin of human speech is based on different pre-adaptations that were present in human ancestors, such as
(a) sound production adaptations related to the descent of the thyroid cartilage associated with the territorial calls of apes,
(b) transformation of the oral and dentitional anatomy including the descent of the hyoid, associated with reduced biting and chewing, and
(c) diving adaptations, leading to voluntary control of the airway entrances and voluntary breath control.

Whereas chimpanzee ancestors became frugivores in tropical forests after they split from human ancestors about 5 Ma (million years ago), human ancestors became littoral omnivores. This might help explain why chimpanzees did not evolve language skills, why human language is a relatively recent phenomenon, and why it is so strongly dependent upon the availability of voluntary breath control, not seen in other hominoids, but clearly present in diving mammals.