Posts Tagged ‘Embodied Embedded Cognition’
The term Embodiment refers to the idea that the internal milieu of the body (such as hormone levels or other homeostatic functions) plays an important role in the processes usually attributed to more higher cognitive processes. This influence is probably achieved through manipulation of emotional states, as suggested by Damasio (1994).
“Embedded” describes the quality of reciprocal interaction between the body and the physical world, which in turn gives rise to cognitive processes.
For a long time the human mind was thought to be totally different from those of animals. It was assumed that sensory data is elaborated by the human mind and stored in a network of abstract representations that are of a semantic structure. The currently dominant paradigm sees the human mind as essentially being a computational-representational system. Within the current paradigm the ultimate explanation for behavior lies within the virtual cognitive functions (software) that are computed by the brain (hardware). Those virtual cognitive functions handle the sensory input, compute a solution and perform output (behavior).
In opposition to this paradigm, the theory of Embodied Embedded Cognition postulates that the difference between the hardware and the software is a semantic one. The metaphor Hardware describes the materialistic, biological aspects of the brain, whereas the metaphor software focuses on the functional aspects. This does not mean that these are two different “things”. Body, brain and world form a system. The intelligent behavior arises from the interaction of the different parts. No a-modal representational system is required to connect a meaning to a symbol, but only a modal system of representations (see also Rolls (1997)).
Specific neurons are activated when they perceive a stimulus, let’s say a car. Through repetition, neuronal activity gets connected to the “real thing” (the car). When enough items of one category have been perceived, we are then able to generate a prototype (Simulators). The idea of this prototype consists of the neural activity that most of the items in one category share. You can also start with the prototype and imagine how an unknown face would look like, by slightly changing the neuronal “fingerprint” of the prototype.
Is there scientific evidence for the Embodied Embedded Cognition Theory?
First of all the a-modal representation has, per definition, all capabilities of a Turing-machine. It is therefore able to explain everything and thus nothing, so the merit as scientific theory is questionable.
Secondly, Embodied Embedded Cognition Theory has postulated some specific hypotheses that have been tested experimentally. For example, specific predictions have been made concerning the spread of related words in an a-modal, a semantic network and a modal network. Those specific predictions have been shown to be true for the human representational system (see Wong & Yon (1991). This has greatly increased the scientific weight of the theory. In comparison, the a-modal theory has never generated this kind of specific hypotheses.
Hopefully greater resolution of brain scans will help to uncover more of the many secrets the human mind still has to offer.
Damasio, A.R. (1994). Descartes’ Error: emotion, reason, and the human brain. New York: Grosset/Putnam.
Rolls, E.T. (1997). “Consciousness in neural networks”. Neural Networks, 10, 1227-1240.
Wong, S.K M. & Yao, Y. Y. (1991). A probabilistic inference model for information retrieval. Transactions on Information Systems 16, 301-321.