The Mechanics of Embodiment

The Mechanics of Embodiment Ken McRae (kenm@uwo.ca) Department of Psychology, University of Western Ontario London, Ontario, Canada N6A 5C2 Martin H. Fischer (m.h.fischer@dundee.ac.uk) School of Psychology, University of Dundee DD1 4HN Scotland UK Keywords: embodied cognition; computation; concepts. Overview and Motivation Embodied cognition is a theoretical stance which postulates that sensory and motor experiences are key parts of the representation of our knowledge. This view has challenged the longstanding assumption that knowledge is represented abstractly in an amodal conceptual network. There now exist a large number of interesting and intriguing demonstrations of embodied cognition. Examples include changes in perceptual experience or motor behaviour as a result of semantic processing. These demonstrations have received a great deal of attention in the literature, and have spurred many researchers to take an embodied approach in their own work. There are also a number of theoretical accounts of how embodied cognition might work. One influential proposal is “perceptual symbols system” theory, according to which the retrieval of conceptual meaning involves a partial re- enactment of experiences during concept acquisition. However, to a large extent, embodied theories are still developing, particularly in terms of computational implementations, as well as specification with regard to moment-by-moment on-line processing. Given the established empirical foundation, and the relatively underspecified theories to date, many researchers are extremely interested in embodied cognition but are clamouring for more mechanistic implementations. This symposium aims to address this specific need for more detailed explorations of the specific processing mechanisms involved in embodied cognition. Four speakers from varying backgrounds and approaches will describe how they think the human mind is embodied, and what they view as the critical current and next steps toward mechanistic theories of embodiment. Toward Implementing Embodiment Lawrence Barsalou (Emory University, Atlanta, Georgia, USA) and Ken McRae (University of Western Ontario, London, Canada) will address issues concerning the construction of embodied computational models. One general set of issues concerns the computational architecture, aside from whether it takes the form of neural networks, Bayesian approaches, production systems, classic AI architectures, or another form. To implement a truly embodied system, multiple modalities are essential. In particular, intelligent action coupled with perception epitomizes embodied approaches, beyond basic response production. Other modalities are also essential from the embodied perspective, including affect and motivation, as well as abstract thought. Another architectural issue concerns the hierarchical structure of feature areas, the hierarchical structure of association areas, and the connectivity patterns among them (Simmons & Barsalou, 2003). Also important are the unique areas associated with bottom-up activation versus top-down simulation, along with shared areas. Finally, issues associated with the architecture’s development and plasticity are important, including genetic and experiential contributions, and how epigenesis is realized (Elman et al., 1996). A second set of critical issues surrounds specific forms of functionality to implement in the architecture. Barsalou (2003) argues that selective attention and categorical memory integration are essential for creating a symbolic system. Once these functions are present, symbolic capabilities can be built upon them, including type-token propositions, predication, categorical inference, conceptual relations, argument binding, productivity, and conceptual combination. Another key aspect is the implementation of space and time. Perception, cognition, and action must be coupled in space and time, and simulations of non-present situations must be implemented in space and time, perhaps using overlapping systems. Because situated action in the environment is fundamental for all organisms, implementing embodied cognition that supports intelligent activity in a few critical situations may be a good place to start (Robbins & Aydede, 2008). By focusing on a complete embodied approach to achieving goals in specific situations, modelers must not only implement specific capabilities, such as goal setting, planning, perception, action, cognition, affect, reward, and learning, but implement interfaces that allow all these processes to interact effectively. These are lofty goals indeed, and the remaining talks will describe current projects that are working toward them. Computational Explorations of Perceptual Symbol Systems Theory The second speaker is Giovanni Pezzulo (National Research Council, Rome, Italy) who has worked extensively

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.