Making Sense of Embodiment: Simulation Theories and the Sharing of Neural Circuitry Between Sensorimotor and Cognitive Processes

Abstract

Making Sense of Embodiment: Simulation Theories and the Sharing of Neural Circuitry Between Sensorimotor and Cognitive Processes Henrik Svensson ([email protected]) University of Skovde, School of Humanities and Informatics Box 408, 54128 Skovde, Sweden Tom Ziemke ([email protected]) University of Skovde, School of Humanities and Informatics Box 408, 54128 Skovde, Sweden Abstract Although an increasing number of cognitive scientists are convinced that cognition is embodied, there still is relatively little agreement on what exactly that means. Notions of what it actually means for a cognizer to be embodied range from simplistic ones such as ‘being physical’ or ‘interacting with an environment’ to more demanding ones that consider a particular morphology or a living body prerequisites for embodied cognition. Based on experimental evidence from a range of disciplines, we argue that one of the keys to understanding the embodiment of cognition is the sharing of neural mechanisms between sensorimotor processes and higher-level cognitive processes. The latter are argued to be embodied in the sense that they make use of (partial) simulations or emulations of sensorimotor processes through the re-activation of neural circuitry also active in bodily perception and action. Introduction Although an increasing number of cognitive scientists are convinced that cognition is embodied (e.g. Varela et al., 1991; Clancey, 1997; Clark, 1997; Lakoff & Johnson, 1999; Ziemke, 2002), there still is relatively little agreement on what exactly that means. Notions of what it actually means for a cognizer to be embodied range from simplistic ones such as ‘being physical’ or ‘interacting with an environment’ to more demanding ones that consider a particular morphology or a living body prerequisites for embodied cognition (cf., e.g., Chrisley & Ziemke, 2002; Wilson, 2002; Anderson, 2003; Ziemke, 2003). This lack of agreement or coherence, after more than a decade of research on embodied cognition, has unfortunate consequences. Firstly, critics commonly argue that the only thing that embodied cognitive theories have in common is in fact the rejection of traditional, computationalist and supposedly disembodied cognitive science. Secondly, there is a certain trivialization of embodiment, not least among many AI researchers who consider as embodied any physical system, or in fact any agent that interacts with some environment, such that the distinction between computationalist and embodied cognitive theories disappears since, in some sense, all systems are embodied, and thus cognitive science has always been about embodied cognition (Chrisley & Ziemke, 2002). Thirdly, there is the ‘misunderstanding’ that perhaps embodiment is only relevant to sensorimotor processes directly involving the body in perception and action, while higher-level cognition might very well be computational in the traditional sense and only dependent on the body in the sense that mental representations ultimately need to be grounded in sensorimotor interaction with the physical environment. This paper, on the other hand, argues that one of the keys to understanding the embodiment of cognition, in an important, non-trivial sense, is to understand the sharing of neural mechanisms between sensorimotor processes and higher-level cognitive processes. Based on experimental evidence from a range of disciplines, we argue that many, if not all, higher-level cognitive processes are body-based in the sense that they make use of (partial) simulations or emulations of sensorimotor processes through the re- activation of neural circuitry that is also active in bodily perception and action (cf. Clark & Grush, 1999; Grush, in press; Hesslow, 2002). As Barsalou et al. (2003) put it, the main point is that “simulations of bodily states in modality specific brain areas may often be the extent to which embodiment is realized”. The next section elaborates the key idea of this paper, i.e., cognition as body- and simulation-based in the above sense, in more detail. In the following sections then supporting empirical evidence from a range of disciplines is presented. The final section then presents a brief summary as well as some open questions and directions for future work. Cognition as body-based simulation The idea that even higher-level cognitive processes are in a strong sense grounded in bodily activity and experience is, of course, hardly new, but was developed already in the 1980s, most influentially by Maturana and Varela (1980, 1987) from a neurobiological perspective, and by Lakoff and Johnson (1980, 1999) from a linguistic perspective. Lakoff (1988) summarized the basic idea as follows: Meaningful conceptual structures arise from two sources: (1) from the structured nature of bodily and social experience and (2) from our innate capacity to imaginatively project from certain well-structured aspects of bodily and interactional experience to abstract conceptual structures.