Posts in on second thoughts

Somebody is thinking —somebody is translating. Embodiment in cognitive translation research

diciembre 26th, 2011 Posted by affordances, blog, cognitive translatology, embodiment, on second thoughts No Comment yet

I couldn't agree more with Sandra Halverson when, in her post of 19 October 2011, she claims that translation would benefit from adopting integrated, non-dualistic approaches. Perhaps such approaches could help us solve difficulties like those humorously depicted by Jorge Amigo (see his posts on 21 November and 6 December 2011) when confronted with the task to teach "the culture of a country" to translation students. Understanding and describing translation processes without preconceived divisions and dichotomies might be helpful in the design of syllabi for translation or interpreting training programs. This is an area where current wisdom needs to be thoroughly revised. The way translation learning is still now conceived of, as reflected in current program syllabi, might be arbitrarily dividing the acquisition and development of translator's complex skills.

In the last decades, cognitive science has challenged traditional dualistic oppositions such as those of input and output, mind and body, and subject and object. Embodied, embedded and distributed cognition is the (compound) umbrella label for various approaches that try to offer an integrated picture of human cognition. In this post, I would like to take a closer view at embodiment from different perspectives, and also to outline some possible implications of these views for translation process research. What does it mean for translation to be embodied? Let's dig it out a little bit.
One of the more frequently cited conceptual precursors to embodied approaches is James Gibson's notion of affordance. It is a relational notion, in that if focuses on the possibilities for action that environments offfer to agents. This early case of a challenge to dualisms focused on the pair stimulus-response (Costall 2007). For Gibson (1979), the visual system has legs, i.e. we are not static machines receiving information from the exjavascript:void(0)ternal world, but creatures that move around and interact with the envorinment. This interaction is what provides us with information about the world. Instead of depicting action in terms of stimuli and responses, Gibson describes information and affordances as resources for embodied action in the world. One and the same thing can offer very different affordances to different agents or to the same agent at different times. However, affordances are not completely subjective. They represent possibilities of action and are partly constrained by things themselves. For instance, the canonical use of a plastic bag is carrying things inside it, but it may be used to protect oneself from the rain. We live in a sophisticated world plenty of affordances.
The concept of affordance has implications for research into cognitive processes of translators, because we cannot isolate those processes from the affordances offered to translators by the environment without paying too high a price for it. It is not just a matter of ecological validity. Interactions with online search machines, databases, translation management programs and cooperative networks are part of translation processes, which thereby become distributed between different agents. Hence, when translating, cognitive processes cannot be separated from embodied interaction with a sophisticated world full of affordances.
The embodied mind
For second-generation cognitive science, such as embodied, embedded and distributed cognition, minds are extended into bodily action in the world. The most radical approaches, like Varela, Thomson & Rosch (1991), base cognition only on embodiment and interaction, with no need for mental representations. According to this approach, we are intelligent agents within the world, and we don't need to internally represent what is out there. The environment affords scaffoldings like language, mnemotechnic rules, pen and paper, computers. From this viewpoint, embodied minds don't need representations at all. At least, they don't need abstract representations with arbitrary referents in the world. Representations, in any case, could be depicted as flexible patterns of interactions with the environment, but not as inner entities with external referents (Johnson & Lakoff 2002).
But don't we represent things internally with our imagination—and not only in dreams? Is language not inherently representational? (Ikegami & Zlatev 2007). A trivial understanding of embodiment such as relating cognition with a physical support system would blurr the difference between traditional and second-generation cognitive science. An embodied approach should have some consequences not only for sensorymotor processes involving the body, but also for "higher-level" cognition. Based on experimental evidence, Barsalou, Solomon & Wu (1999) and Svensen, Lindblom & Ziemke (2007) argue that the core of embodiment is that sensorimotor processes and "higher-level" cognition share neural mechanisms. This shared neural circuitries are the key to understand embodiment, because through their activation we can mentally simulate al kinds of sensorimotor processes, and these simulations are the matter that concepts and thoughts are made of. From this perspective, our mental representations are embodied in a non-trivial sense: they do not just have to deal with a physical support; they are constructed through the same kind of processes as bodily perception and motion.
Now, what does all this mean for translation?
If translation process research were to adopt a trivial notion of embodiment (that of considering the body just a supporting physical system), it would hardly have any implications for current research. But the notion of embodiment may have important consequences if its focus is widened to study the mind as extended through bodily action into the world. Under this scope, translators' mental processes cannot be studied as isolated sequences of thoughts taking place in individual brains. Translators also translate with their hands, their eyes and their computers. Research into cognitive translation processes is no limited any longer to what happens in the translators' brains, but needs to enlarge its scope to cover what happens in their bodily activities and interactions. The evolution of translation process research method for data-collection parallels this change of perspective: the focus has shifted from translators' thoughts (elicited through introspection) to their behaviour, which comprises interactive activities from different perspectives, such as typing on the keyboard, moving one's eyes on a screen, online searching for information, and interacting with other people. Translators' activities are viewed not just only as keys into, but also as part of, mental processes during translation, inasmuch as they demand control and monitoring and/or the activation of learned routines and entail a cognitive load for the system (Muñoz in press).
Adopting such an approach also raises several questions regarding the fact that the same mental circuitries are shared by perception and motor action, on one side, and "higher" mental activities, on the other. What does this mean for translation processes? Do translators work with mental images or mental simulations? How can this be researched? What can the consequences be, both for our understanding of translation and for our search for improvements in learning and practice?
Translation process research and technological developments
Under this light, translation process research may also be extended to studying the quality of current or developing translators' tools. Translation support systems make things easier for translators, but sometimes they may have counterproductive effects too. For instance, prompting dictionaries combined with eye trackers can provide a translation for a word when the reader keeps looking at it beyond a threshold in time (Stamenov 2009). Such a device should somehow be very rich and adaptive in order not to limit potential translation solutions, thereby curbing translators' creativity. Repeated interactin with such a devise might also train translators to move their focus of attention on isolated words.
The same concern of reducing potential options applies to the current trend of Internet personalization. Usually, Internet tools make things easier for us, but sometimes they may go beyond an optimal limit and have negative effects. For instance, search machines and web networks can become so personalized, that they filter out all information that supposedly represents a viewpoint different from that of the user, isolating her in an information bubble. This actually leads to a reduction of the potential options to choose from where the subject has nothing to say. This is bad news for translation (see fully developed argument here)
Barsalou, Lawrence W., Karen Olseth Solomon & Ling-Ling Wu. 1999. Perceptual simulation in conceptual tasks. @ Masako K. Hiraga, Christoper Sinha & Sherman Wilcox (eds). Cultural, Typological and Psychological Perspectives in Cognitive Linguistics. Amsterdam: John Benjamins. 209-228.
Costall, Alan. 2007. Bringing the body back to life: James Gibson's ecology of embodied agency. @ Ziemke,  Zlatev & Frank, eds., pp. 55–84.
Gibson, James. 1979. The Ecological Approach to Visual Perception. Boston: Houghton Mifflin.
Ikegami, Takashi & Jordan Zlatev. 2007. From pre-representational cognition to language. @ Ziemke,  Zlatev & Frank, eds., pp. 197–239.
Johnson, Mark & George Lakoff. 2002. Why cognitive lingustics requires embodied realism. @ Cognitive Linguistics 13/3: 245–263.

Muñoz Martín, Ricardo. (In press). Just a matter of scope. Mental load in translation process research. @ Translation Spaces 1.


Stamenov, Maxim I. 2009. Cognates in language, in the mind and in a prompting dictionary for translation. @ S. Göpferich, A. L. Jakobsen & I. M. Mees, eds. Behind the mind. Methods, models and results in translation process research. Copenhagen: Samfundslitteratur.
Svenson, Henrik, Jessica Lindblom & Tom Ziemke. 2007. Making sense of embodied cognition: simulations theories of shared neural mechanisms for sensorimotor and cognitive processes @ Ziemke,  Zlatev & Frank, eds., pp. 242–269.
Varela, Francisco, Evan Thompson & Eleanor Rosch. 1991. The embodied Mind. Cognitive Science and Human Experience. Cambridge, Mass.: MIT Press.
Ziemke, Tom, Jordan Zlatev & Roslyn M. Frank (eds) Body, Language and Mind. Volume 1: Embodiment. Berlin: Mouton de Gruyter.
by C. Martín

Mirror mirror in the brain

julio 20th, 2011 Posted by blog, cognitive translatology, mirror neurons, on second thoughts No Comment yet

The correspondences between cognitive and neural operations are difficult to characterize. In the field of neuroscience of language, some researchers favor a modular division between syntactic and semantic neuroanatomical areas (for a review, see Carota

& Sirigu 2008) while others like Hagoort (2005, apud Toni 2008) emphasize functional gradients. Empirical evidence is not a mirror of reality. Both empirical evidence and reality need to be interpreted; and empirical evidence has also to be obtained. And yet, there seems to be empirical evidence of some kind of internal mirror in our brains. Not a mirror of reality, but a mirror of others’ actions and feelings.
Mirror neurons were discovered in the brain of a macaque monkey by Rizzolatti and his colleagues and interpreted as an understanding mechanism (Di Pellegrino, Fadiga, Fogassi, Gallese & Rizzolatti 1992; Rizzolatti, Fogassi & Gallese 2001, apud Zwaan 2008). In this view, the macaque understands the researcher’s action of grasping some food by internally simulating this action. Mirror neurons fire when we do something and also when we see someone else doing the same thing. We run the same motor program in both cases, and this is how we understand others’ actions.
In humans, the macaque’s brain area where mirror neurons were found corresponds to Broca’s area, so simulation theory was connected to language from the start (Rizzolatti & Arbib 1998, apud Zwaan 2008:16). From then on, a large number of empirical studies have investigated the relationships between language comprehension and the activation of sensorimotor areas from different perspectives. The underlying common thread is that language-mediated comprehension consists of internal simulations of actions and perceptions. Internal simulations of actions are motor activations similar to those involved in the real execution of the actions. Internal simulations of perceptions are reactivations of prior visual, tactile, auditory, kinesthetic and olfactory experiences. This last line of research has its origins in the works of Damasio (1989) and Barsalou (1999).
The fairest of all
Translatology is an applied discipline. The main goal of translatology is to improve translation quality, practice and learning. So what has all this to do with translation? What is the use of simulation theories in translation research? Would it make any difference at all if some group of mirror neurons happened to fire during translation activities? Would it have any consequences for translatology if some motor areas were activated during text comprehension? Do these correlations take place at a level too low to have any consequences for translation at all?
The relationships between neuroanatomical findings and translatological research are not straightforward. However, language comprehension is central to translation, and gaining a deeper insight on how we construct meaning during translation processes is no doubt of great importance for translatology. Pinning down the role of imagination in translation could be an important leap forward in the design of didactic scaffoldings.
Translation through the mirror
From the perspective of the cognitive research on language comprehension, the peculiarities and complexities of translation practice make an interesting research field: when translating, the requirements of the assignment or translation brief can have and influence from the very start. Translators read and interpret texts for other people, and they must somehow put themselves in their shoes (the only comment of Lakoff about translation in WF&DT is precisely that), although they usually do not know who their addressees are. Mirror neurons and resonance mechanisms most probably play a role in the processes of anticipating readers’ comprehension possibilities and difficulties.
Researching mental simulations in the field of translation cannot be easy. Perhaps one of the main problems to face is that mental simulations are mainly unconscious. And even if they were made conscious or could somehow be brought to the surface through introspective methods, results would pose a lot of methodological and epistemological problems. To avoid these problems, experimental conditions might need to differ from translators’ usual practices. In doing so, they would lose ecological validity. However, it is worth trying, and cognitive psychology is definitely a source of inspiration. For example, experiments may be designed to study translation (not “memory”) priming effects in the translation process, with conditions such as translating a text with and without previous exposure to related images or sounds, or executing certain movements or actions before translating. The insights obtained with this kind of experiments could bring us to design more realistic situations to investigate how imagination works during translation.
Barsalou, Lawrence C. 1999. Perceptual Symbol Systems. @ Behavioral and Brain Sciences 22: 502–518.
Carota, Francesca & Angela Sirigu. 2008. Neural bases of sequence processing in action and language @ P. Indefrey & M. Gullberg, eds., pp. 179–199.
Damasio, Antonio R. 1989. The brain binds entities and events by multiregional activation from convergence zones @ Neural Computation 1: 123–132.
Di Pellegrino, G., L. Fadiga, L. Fogassi, V. Gallese & G. Rizzolatti. 1992. Understanding motor events: A neurophysiological study @ Experimental Brain Research, 91/1: 176–180.
Hagoort, P. 2005. On Broca, brain, and binding: A new framework @ Trends in Cognitive Sciences 9: 416–423.
Indefrey, Peter & Marianne Gullberg, eds. 2008. Time to Speak: Cognitive and Neural Prerequisites for Time in Language. Malden: Wiley-Blackwell.
Rizzolatti, G. & M. A. Arbib. 1998. Language within our grasp @ Trends in Neurosciences 21: 188–194.
Rizzolatti, G., Fogassi, L. & V. Gallese. 2001. Neurophysiological mechanisms underlying the understanding and imitiation of action @ Nature Reviews: Neuroscience 2: 661–670.
Toni, Ivan. 2008. Sequential event processing: Domain specificity or task specificity? Commentary on Carota & Sirigu. @ P. Indefrey & M. Gullberg, eds., pp. 201–205.
Zwaan, Rolf A. 2008. Time in language, situation models, and mental simulations. @ P. Indefrey & M. Gullberg, eds., pp. 13–26.
by C. Martín

La lexicalización del pensamiento en paralelo y otras metáforas

mayo 25th, 2011 Posted by arallel thinking, blog, metaphors, on second thoughts No Comment yet
Aunque la teoría del pensamiento en paralelo ha surgido hace relativamente poco tiempo, resulta muy curioso descubrir cómo en una gran cantidad de expresiones y frases hechas podemos encontrar indicios y veladas referencias a ella. Expresiones como «una parte de mí piensa que…», «la voz de la conciencia le decía que…» o la habitual división entre «lo que opina el corazón y lo que opina la cabeza» nos hacen sospechar de que en nuestro interior no estamos tan solos como pensamos, sino que hay otros que se comunican misteriosamente con nosotros, ayudándonos en algunas ocasiones, saboteándonos en otras y, a veces, sencillamente acompañándonos.

Bien pensado, el pensamiento en paralelo ofrece claras ventajas evolutivas que, en cierta manera y de manera ya poco sorprendente, pueden asemejarse a los modelos de programación informáticos. Resulta mucho más fácil crear una rutina que realice una operación muy sencilla, por ejemplo detectar tipos muy específicos de errores dentro de un texto, aunque esas rutinas no serán capaces de obtener una visión global de un problema, a no ser que una rutina central trabaje con los resultados de cada una de ellas. Por expresarlo con una metáfora más sencilla, el pensamiento en paralelo le permitiría a un depredador evaluar por separado con gran rapidez los diferentes aspectos de una situación de caza, no solo la información sensorial sobre la presa y el entorno, sino también las distintas posibilidades de acción, pero sin que ascendiera a la superficie más que la información más relevante para la situación. Como ejemplo de esto, algunos ajedrecistas dicen percibir en ocasiones una «sensación de peligro», después de determinadas jugadas por parte del contrincante, que les alerta de que pueden estar a punto de perder la partida, aunque son incapaces de expresar exactamente de dónde proviene dicha sensación.

Los ordenadores actuales, como si fueran animales que también evolucionan, parecen estar siguiendo dicha tendencia y desde hace pocos años los procesadores con varios núcleos son la moda. La consola PlayStation 3 se caracteriza por disponer de un procesador central relativamente poco potente, pero con varios procesadores adicionales a los que se puede delegar determinadas actividades, un modelo que se ha recibido muchos elogios por parte de los técnicos y críticas por parte de los programadores, que encuentran difícil trabajar de esta manera, por asombrosos que sean los resultados. Efectivamente, nos resulta muy difícil pensar de manera no secuencial, probablemente porque la centralización de la toma de decisiones ha demostrado ser tan fundamental para nuestra supervivencia como la existencia de multitud de opiniones en nuestro interior.

La teoría del pensamiento en paralelo plantea un buen número de interrogantes fascinantes, ¿cuántos «procesadores adicionales» tenemos en nuestro interior?, ¿dos, diez, cien, millones?, ¿son estables dichos procesadores (lo que demostraría, por ejemplo, que efectivamente hay una voz de la conciencia en nuestro interior) o se crean y se destruyen según la situación?, ¿nace el ser humano con todos los procesadores o se van formando a medida que se desarrolla?, ¿cuentan también otros animales con estos procesadores? Preguntas fascinantes cuyas respuestas podrían redefinir nuestra concepción del ser humano y permitirnos tratar determinados trastornos que podrían estar causados por problemas de acoplamiento entre estos procesadores.

Aún cuando esta teoría no resulte tan prometedora como parece, puede ser muy inspiradora a la hora de diseñar todo tipo de sistemas. Siguiendo los preceptos de la biomimética, si la naturaleza ha elegido esta vía para procesar la información, bien merece la pena tenerlo en cuenta. Un sistema muy complicado probablemente podrá ser más eficaz si se descompone en unidades básicas que intercambian inteligentemente información que, de por sí, no es tan inteligente. ¿Podría ser?

by J. Perea

Transfer? What transfer?

mayo 18th, 2011 Posted by blog, cognitive translatology, on second thoughts, transfer No Comment yet
Like memes (or as such), concepts sometimes seem to jump from generation to generation, from paradigm to paradigm, from approach to approach. This appears to be the case with the concept of “transfer”, when it is meant to point to a phase of the translation process. This transfer phase showed up in deductive models in the 60’s, such as Nida’s (also Seleskovitch’s, but we will not pursue that lead any further here). Nida was, in his turn, inspired by generative linguistics, and hence tended to explain translating as a sequential process, which could be modelled by serial computing. This model depicts human translation as a three-phase process, with a transfer phase between those of comprehension and production. There is no evidence of the alleged transfer activity or phase. However, the notion seems to have been inherited by some approaches in empirical translatology, and I think cognitive translatology should challenge it.
From a cognitive perspective, it does not seem likely that a boundary can be drawn between processes involved in source text comprehending and those used in target text production. Observing translators’ behaviors is not enough to manage to distinguish between what goes on in your mind when you are (only) reading and when are you (only) writing. Present-day tools allow us to peek once or twice into mental processing by inferencing moves from observable behavior, but only time and gaze can be properly measured thus far. But keylogging and eyetracking cannot (yet) be used to determine whether there was more or less attention devoted to the main focus and perhaps (simultaneously) some marginal work on that of a different phase, i.e., because translators may shift their attention totally or partially, to different degrees and perhaps component operations, between the two actions. Today, these changes will still go unnoticed to our data gathering tool.
Now, eyetracking shows that reading a text with the (conscious and explicit) purpose of translating it turns out to be very different from reading it just in order to understand the gist of it (Castro 2007). When translating, comprehension and production processes seem tightly intertwinned. We know that, during target text production, translators go back once and again to the source text, so the activities as a whole can properly be described as a a set of recursive processes. Where is the transfer phase to be located in relation with these processes?
Most Western languages are metaphorically endowed to deal with translating as a movement from a source to a target. We are, in fact, so used to it that we hardly question the psychological reality of what has traditionally been labelled “transfer”. However, it is very difficult to hypothesize and capture a behavior associated with it. What is the translator supposed to do in this phase? Does she look up in a bilingual dictionary? Does she mentally search for equivalent expressions? Is the transfer ended as soon as we begin to type? How can we discern the transfer from the production phase?
Interestingly, when translating, sometimes target text production takes place almost simultaneously with source text reading, or even before it, because profesional translators, like interpreters, will now and then anticipate parts of the text they are translating (Dragsted 2010). The phenomenon seems to be a part of what have been variously named automat(iz)ed translating, routine implementation, non-problematic or uneventful rendering, interiorized routines, I think we will all agree on having similar impressions. Is transfer in such cases previous to text comprehension? One way out of this reductio ad absurdum is to equate and identify “transfer” with “translation” as a whole. Transfer only makes sense when there are already two texts/text segments in different languages to compare and you consider that they correspond to a certain extent in certain aspects.
But then using “transfer” would only add confusion. Thus, in this case it would be better to give up the notion of ”transfer”, to cut it off from our map of translators’ minds with Occam’s razor. Sometimes fuzzy concepts will help to amass knowledge on the process of translating by conviniently carving out blurred areas so that advances may be proposed in other parts of it, but these concepts may later become a burden to the development of this knowledge. So let’s drop “transfer”, and face instead the complexities of translation processes.
Castro Arce, María. 2007. Procesos de lectura y comprensión al traducir @ M M Fernández & R Muñoz, eds. Aproximaciones cognitivas al estudio de la traducción y la interpretación. Granada: Comares, pp 31–54.
Dragsted, Barbara. 2010. Coordination of reading and writing processes @ G Shreve & E Angelone, eds. Translation and Cognition. Amsterdam: John Benjamins, pp 41–61.
by C. Martín

Situatedness without mind?

abril 4th, 2011 Posted by blog, concepts, dynamic systems, mind, on second thoughts, situatedness No Comment yet

The most radically dynamic approaches to cognition depict it as a series of embodied, situated processes closely intertwined with action and environment. Some advocates of these approaches claim, for example, that robots designed without representationalist programming techniques —which do not manipulate internal representations like symbols— can behave intelligently (Brooks 1991). Others explain early human cognitive development in terms of dynamic systems

theory (Thelen & Smith 1996). These scopes do not seem very interested in things such as minds, thoughts or mental representations. That is why they have been criticized as a return to behaviorism. After all, we have internal representations, don’t we? I can imagine the sea, and I can see the waves dancing before my mental eye. So, what is the point of denying it?

Perhaps it is our own way of speaking what confounds us. Most of our words about cognitive processes are count nouns. We understand minds, meanings and thoughts as things or, at least, as states, but not as processes or actions. There are reasons for it. An analysis of the metaphors we use to speak about some kinds of things, such as minds and their contents, hints that we intuitively find it easier to think about cognition artifacts as everyday objects in a tridimensional space rather than in terms of complex, dynamical and intertwined processes. At least, for everyday purposes. But what about our scientific concepts?

This state of affairs does not seem very different in academic speech: Perhaps it is easy for us to see that minds are not containers with thoughts and memories inside them, or that words are not packages of meaning. But seeing beyond metaphors becomes more difficult when we talk about frames, mental images, idealized cognitive models or even just concepts. We tend to reify these notions, to think about them as ready-made schemes that can be retrieved from memory and applied to each situation. Once again, thinking of states or objects is simply much easier than attempting to envision dynamic processes or actions.

In translatology, functionalist theories have made us grow used to think in terms of actions. In cognitive translatology, in contrast with traditional linguistics, we conceive of language from the very beginning as part of a situated communicative activity. However, we are not immune to the tendency to reify concepts. Perhaps the most notorious example of this tendency is the very concept of meaning. It is very difficult not to think about meaning as something somehow there, ambushed in the text, waiting to be deciphered. And yet, we know that meaning is not to be found within texts. It is not archived in our minds either. Meaning is always created anew in dynamic processes of interaction between text, embodied mind and situation.

However, this does not mean that we cannot agree to some extent about the meaning of a text, nor that translators do not create mental representations at all. As a matter of fact, when we study what goes on in the translator's hands, in her computer and her interactions (Risku 2010), we are also trying to find out what is going on in her mind, even though we do not think about that mind as an isolated (black) box, but as a series of processes that are, so to speak, open to the world.

A possible way to escape the dichotomy of behaviorism vs mentalism might be to try and think about mental processes—such as remembering, imagining, thinking or dreaming—not only as part of situated action, but as actions themselves. Instead of asking What is going on in the translator’s mind? we could ask What is the translator mentally doing? The problem with this strategy is that not all mental processes can be depicted as intentional, goal-directed actions, that most of them are unconscious and/or uncontrolled. After all, we are not inclined to speak about digestive processes as actions either. And yet, mental processes can only be explained in relation with action and social interaction. Our most elaborated constructions, like language or the self, can only emerge within a highly sophisticated sociocultural world which affords us the necessary stability to act in it.

Thinking in terms of dynamic processes does not mean that we have to give up the very notion of mental representation; it just means that we must rethink it. We can’t step into the same river twice, but we conceive of it as the same river we forded yesterday. In fact, we cannot construct the same concept twice (Barsalou 1993). Nevertheless, we construct ourselves as being the same as yesterday, and this construction allows us to interact in a meaningful way with our environment.

Barsalou, Lawrence D. 1993. Challenging assumptions about concepts. Cognitive Development 8/2: 169-180.
Brooks, Rodney A. 1991. Intelligence Without Representation. Artificial Intelligence 47: 139-159.
Risku, Hanna. 2010. A cognitive scientific view on technical communication and translate. Do embodiment and situatedness really make a difference? Target 22/1, 94-111.
Thelen, Esther & Linda B. Smith. 1994. A dynamic systems approach to the development of cognition and action. Cambridge: MIT Press.

Recent Comments