This paper is a survey of the main issues surrounding the role of embodiment in cognition and development. For most of the history of cognitive science, a "disembodied" model of cognition-according to which cognitive processes are independent of sensorimotor processes-dominated the field. More recently, the field has taken a turn toward embodiment, the view that sensorimotor processes routinely influence (and perhaps even constitute) cognitive processes. This view has diverse historical origins in American pragmatism, phenomenology and ecological psychology. However, in recent years, researchers have adduced many arguments and a great deal of empirical evidence in favor of the embodied cognition hypothesis. Developmental psychologists have played an important role in this transformation of cognitive science, adopting and adapting Piaget's views as a means of explicating the role that embodiment plays in development, as well as collecting developmental data that support the embodied cognition hypothesis. Despite the enthusiasm for embodiment among cognitive scientists generally and developmentalists in particular, however, the embodied cognition hypothesis still faces formidable challenges.
KEYWORDS: cognition, development, embodiment, sensorimotor systems
Over the last 20 years or so, an enormous amount has been said in the cognitive science literature about the topic of embodiment, some of it specifically about the role that embodiment plays in development. A comprehensive review of the evidence and arguments is not practical in this forum. Instead, the goal of this paper is to give the reader a sense of the main issues regarding the role of embodiment in cognition and development. The first section traces the history of the "disembodied" perspective that dominated cognitive science from the origins of the discipline through the 1980s. The second section traces the history of the "embodied" perspective. In addition, it surveys some arguments for the embodied cognition hypothesis and describes some of the empirical evidence that weighs in its favor. The third section examines some of the evidence and theoretical contributions that have come out of the developmental literature in particular. The fourth section points out important challenges for supporters of the embodied cognition hypothesis.
THE DISEMBODIED PERSPECTIVE IN COGNITIVE SCIENCE
For many years, the orthodox view in cognitive science was that the study of cognition, including the study of conceptual processes and mechanisms, could proceed independently from the study of sensorimotor processes and mechanisms. Researchers assumed that it was sufficient to use discrete, amodal symbols to represent not only the contents of thoughts but also the sensory activities that provide information about the environment and the motor activities that constitute behavior. They considered these sensorimotor processes irrelevant to the core project of cognitive science-explaining cognitive processes, which were thought to take place in an abstract information-processing device. Sensorimotor processes, on this view, were relegated to peripheral input and output devices (M. Wilson, 2002).
The view that cognition can (and should) be studied independently of perception and action had its roots in the computational metaphor of mind, beginning with Turing's concept of a universal computing machine (Turing, 1936). Turing essentially reduced the problem of creating a thinking machine to the problem of calculating mathematical functions (Rohrer, 2007). From its origins in Turing's work, the universal computing machine became the preferred metaphor for the mind through the 1980s (Johnson & Rohrer, 2007).
There was a natural affinity between the computational metaphor of mind and the philosophical position known as functionalism (Putnam, 1960), the view that mental states inherit their contents from their functional roles, that is, their roles in a computational system. …