Natural Rationality | decision-making in the economy of nature

1/9/08

Embodied, Situated and Distributed Cognition

[I am currently starting to write an introduction for a collective book on that subject; I try here to summarize the big ideas of these approaches]

Blogging on Peer-Reviewed Research
From 1990 to 1999, the U.S. Library of Congress and National Institute of Mental Health sponsored an interagency initiative that designated the 90’s as the Decade of the Brain. Paradoxically, the cognitive science of this decade was marked by a major methodological and conceptual change that one can summarizes as “cognition beyond the brain”. Whereas the brain was traditionally conceived as being the only seat of intelligence, many trends of research emphasized the functional entrenchment of the brain in the body, environment and culture. In neuroscience (Churchland, Ramachandran, and Sejnowski 1994), psychology (Barsalou 1999; Thelen and Smith 1994) Artificial Intelligence (Ballard 1991), robotics (Pfeifer and Scheier 1999; Brooks 1999), Artificial Life (Langton 1995), linguistics (Lakoff and Johnson 1999) and philosophy (Clark 1997), researchers began to be dubious of the “standard picture”:

Perception is commonly cast as a process by which we receive information from the world. Cognition then comprises intelligent processes defined over some inner rendition of such information. Intentional action is glossed as the carrying out of commands that constitute the output of a cogitative, central system. (Clark, 1997, p. 51)

According to the new approaches, cognitive processes are not limited to the symbolic processing of internal information structures, but implemented in various sensorimotors processes (motricity, perception, emotions, coordination, imagery, emulation, simulation) and various substrates (members, bodies, artifacts, environmental regularities). The major theme of these new approaches is the embeddedness of cognition: in the body and in the world. If one can, conceptually, distinguish the brain from the body and from the environment, a flood of dense and continuous information binds the three together.

Whether it is problem-solving or perception, decision-making or memory, many cognitive processes “piggiback” on lower-level processes. The best example is probably Damasio’s discovery of the importance of emotions in cognition. As he and his colleagues discovered, subjects impaired in affective processing—due to damage to the orbitofrontal, an area involved in emotions and memory integration—are unable to cope with everyday tasks, such as planning meetings (Damasio 1994). These lose the ability to anticipate and sometimes event experience negative feelings. They thus engage in behaviors detrimental to their own well-being (loss of money, friends, family, social status, etc.), However, they are completely functional in reasoning or logical tasks. Moreover, they do not feel sad for their situation, even if they perfectly understand what “sad” means. Damasio thus suggested that many cognitive processes need affective ones, such as “somatic markers”. Somatic markers are bodily states, such as skin-conductance, cardiac rhythm, etc., that ‘mark’ options as advantageous/disadvantageous. Through learning, bodily states become linked to neural representations of the stimuli that brought about these states. These neural structures may later reactivate the bodily states or a simulation of these states and thereby indicate the valence and magnitude of stimuli. Our well-documented preference for sure lotteries over risky ones thus seems not to be a product of logical deduction, but of gut feelings. Emulation theory of motor control (Kawato 1999) and simulationist accounts of perception (Barsalou 1999) essentially make the same point: high-level processes such as voluntary action guidance or categorization can be seen as complex redeployments of sensorimotor processes (Poirier, Hardy-Vallée, and De Pasquale 2005). In every case, bodily structures are prerequisite for entertaining decoupled representations run offline. In the end, there may be no such thing as a purely cognitive process, if one adopts the radical position of Barsalou and colleagues: cognition is coordinated non-cognition (Barsalou, Breazeal, and Smith 2007).

The somatic markers illustrate in which sense cognition can be said to be beyond the brain: just like the hammer “fuses” with the hand of the blacksmith and become literally an extension of it, our brains fuse with others, non-neural processes (e.g. visceral, perceptual, muscular). More than a century ago, the founder of American pragmatism, Charles S. Peirce had the intuition that, in certain case, we may recruit external structures as “instrument of thought”:

The old chemist's maxim had been, "Lege, lege, lege, labora, ora, et relege." Lavoisier's method was not to read and pray, but to dream that some long and complicated chemical process would have a certain effect, to put it into practice with dull patience, after its inevitable failure, to dream that with some modification it would have another result, and to end by publishing the last dream as a fact: his way was to carry his mind into his laboratory, and literally to make of his alembics and cucurbits instruments of thought, giving a new conception of reasoning as something which was to be done with one's eyes open, in manipulating real things instead of words and fancies. (Peirce 1877).


Whereas embodied cognition refers to the embeddedness of the brain in the body, situated cognition refers to the embeddedness of the brain-body complex in an environment (see essays in Aydede and Robbins, forthcoming). Just like our brains can “consult” visceral processes, they can also use external structures or “epistemic structures” as Chandrasekharan & Stewart call them (2007): road signs, post-it notes, color codes are different means by which memory, categorization or inference can be simplified. In a manifesto for the “extended mind” thesis, Clark and Chalmers argued that retrieving information in a notebook or in long-term memory involve essentially the same functional process, although in the first case cognition extends in the environment (Clark and Chalmers 1998). Notebook are epistemic structures that make retrieval simpler: why having your whole shopping list in your head while you can conveniently carry it in your pocket?

These structures, however, are not just convenient appendage but inherently tied to our mental life. Andy Clark referred these environmental and social regularities that support and extend cognition as external scaffoldings (Clark 1997). We grow, learn and act everyday through these, language being the ultimate cognitive artifact. Human language, cultures and institutions allow our species to extend cognitive processes beyond the brain, the body and the immediate environment. Cognition is not only embodied and situated, but also distributed: in certain cases collectives (or networks of situated brains) process information conjointly (Hutchins 1995). Teams, families, groups and organizations sometimes implement cognitive activities where no one in particular can be said to implement them as a whole. For instance, when a jury deliberate, information flows from one brain to another for hours. One can say that as an entity, the jury remembers, evaluates and chooses a verdict. However, the substrate of remembering, evaluating and choosing is not an individual but a collective of brains whose interaction is made possible and constrained by two external scaffolding: language and procedural rules. In the era of Internet and groupware, distributed cognition and collective intelligence are key notions that need to be carefully defined and studied.

Cognition is therefore embodied, situated in an environment and distributed among agents, artifacts and external structures. The constant sensorimotor dynamics, the role of action in perception and learning, the agent/environnement coupling and collective intelligence are cognitive determinants which one must take into account to describe adequately natural cognitive systems and to build artificial cognitive systems. If this picture of the mind is correct, if we are “better at Frisbee than at logic” (Clark, p.60), then how should the scientific study of cognition be conducted? What can we learn frome it? And what are the limits of this so-called new paradigm? As Adams and Aizawa and made clear in their book The Bounds of Cognition (2008), “extracranalists” theoreticians of the mind face several problems. For instance, how are we to analyze a fuzzy category such as “systems made of brain coupled with tools”? Why should pencils or computers figure in a theory of cognition? [to be continued...]


References
  • Adams, Frederick, and Kenneth Aizawa. 2008. The bounds of cognition. Malden, MA: Blackwell Pub.
  • Aydede, Murat, and Philip Robbins, eds. forthcoming. Handbook of Situated Cognition. Cambridge: Cambridge University Press.
  • Ballard, Dana H. 1991. Animate vision. Artificial Intelligence 48 (1):57-86.
  • Barsalou, L.W. 1999. Perceptual symbol systems. Behavioral and Brain Sciences 22 (4):577-609; discussion 610-60.
  • Barsalou, Lawrence, Cynthia Breazeal, and Linda Smith. 2007. Cognition as coordinated non-cognition. Cognitive Processing 8 (2):79-91.
  • Brooks, R.A. 1991. Intelligence Without Representation. Artificial Intelligence Journal 47:139-159.
  • ———. 1999. Cambrian intelligence : the early history of the new AI. Cambridge, Mass.: MIT Press.
  • Chandrasekharan, Sanjay, and Terrence C. Stewart. 2007. The Origin of Epistemic Structures and Proto-Representations. Adaptive Behavior 15 (3):329-353.
  • Churchland, Patricia S., Vilayanur S. Ramachandran, and Terrence J. Sejnowski. 1994. A Critique of Pure Vision. In Large-Scale Neuronal Theories of the Brain, edited by C. Koch and J. L. Davis. Cambridge: Bradford Book, The MIT Press.
  • Clark, Andy. 1997. Being there : putting brain, body, and world together again. Cambridge, Mass.: MIT Press.
  • Clark, Andy, and David Chalmers. 1998. The Extended Mind. Analysis 58 (1):7-19.
  • Damasio, Antonio R. 1994. Descartes' error : emotion, reason, and the human brain. New York: Putnam.
  • Dewey, John. 1906a. Reality as Experience. The Journal of Philosophy, Psychology and Scientific Methods 3 (10):253-257.
  • ———. 1906b. The Experimental Theory of Knowledge. Mind 15 (59):293-307.
  • Hutchins, E. 1995. Cognition in the wild. Cambridge, MA: MIT Press.
  • Kawato, Mitsuo. 1999. Internal models for motor control and trajectory planning. Current Opinion in Neurobiology 9 (6):718-27.
  • Kirsh, David. 1991. Foundations of AI: The big issues. Artificial Intelligence 47 (1-3):3-30.
  • Lakoff, George, and Mark Johnson. 1999. Philosophy in the flesh : the embodied mind and its challenge to Western thought. New York: Basic Books.
  • Langton, Christopher G. 1995. Artificial life : an overview. Cambridge, Mass.: MIT Press.
  • Peirce, Charles S. 1877. The Fixation of Belief. Popular Science Monthly 12:1-15.
  • Pfeifer, Rolf, and Christian Scheier. 1999. Understanding intelligence. Cambridge, Mass.: MIT Press.
  • Poirier, P., B. Hardy-Vallée, and J.-F. De Pasquale. 2005. Embodied categorization. In Handbook of Categorization in Cognitive Science, edited by C. Lefebvre and H. Cohen: Elsevier.
  • Thelen, Esther, and Linda B. Smith. 1994. A Dynamic systems approach to the development of cognition and action, MIT Press/Bradford Books series in cognitive psychology. Cambridge, Mass.: MIT Press.
  • Varela, Francisco J., Evan Thompson, and Eleanor Rosch. 1991. The embodied mind : cognitive science and human experience. Cambridge, Mass.: MIT Press.



3 Comments:

Anonymous said...

"Whereas embodied cognition refers to the embeddedness of the brain in the body"

I don't think that's what embodied cognition means. It is how our cognitive processes are based on action. Even perception is based on action. See for example Alva Noe and others.

Anibal said...

For the sake of the argument suppose that i ´m neither an advocate of embodied cogntion nor an advocate of the more traditional approaches to cognition.
My simple question, and surprise, is, why, given the fact that the brain it is the most complex system in the universe (or so it is said)and that we don´t have a complete understanding of it, we have to left to study this wonderful organ and put our attention to processes beyond its reach, or even the skin and skull as the lemma of emodied cogntion states.
Nöe´s latest book "Action in perception" in order to defend the embodied approach presents a counterargument made by John Searle about how we can understand the workings of the gastrointestinal system by means of studying "food" as an analogy of why we have to study the workings of the brain by means of studying the "world".
It is in this sense, that perhaps we try to unload too much the distintictive parts and contents of the machinery of our brains to look elsewhere which seems terra incognita.
Despite of it, the sensoriomotor approach of Noe and Hurley and others, including Clark and even you, deserves much consideration because might answer many interrogations pose by traditional approaches.

D Silverman said...

Anonymous said...

"Whereas embodied cognition refers to the embeddedness of the brain in the body"

I don't think that's what embodied cognition means. It is how our cognitive processes are based on action. Even perception is based on action. See for example Alva Noe and others."

It comes to the same thing, however.

Noe argues that the only thing that can carry out the action-based perceptual processes humans can has to be a human, or at least has to have a body that resembles a human one in all crucial respects.

This would be a rejoinder to someone thinks the processes could be reproduced in something radically different like a metal box.