Geoffrey F. Miller, Jennifer J. Freyd
Recently, some theorists (e.g. Port & Van Gelder, in press)have suggested that cognition is more a matter of `dynamics' than of `computation',such that concepts borrowed from dynamical systems theory will prove more useful in psychology than concepts such as `internal representation' and `information-processing' borrowed from traditional epistemology and computer science. Although we sympathize with the growing excitement surrounding this `dynamical cognition view' (Skarda & Freeman, 1987; Smolensky, 1988; Van Gelder, 1992) we believe that such a view complements and reinforces,rather than contradicts, a broadly defined representationalist view (e.g. Shepard, 1984).Although Freyd (1987, 1992, 1993) has already advanced a theory of `dynamic mental representations' and substantial evidence for their existence, the relationship between dynamic representations and the new dynamical cognition view has not yet been investigated. This paper argues that a theory of dynamical representations can combine the methodological and theoretical strengths of the dynamical and the representationalist approaches to understanding cognition. Section 1 presents a historical introduction to the issues, and section 2 shows how the dynamical systems view and the representationalist view might be reconciled at the theoretical level. Section 3 shows how they have already been reconciled at the experimental level by Freyd's research on dynamic representations and representational momentum, and section 4 reviews recent extensions of this experimental research showing that some dynamic representations are specially tuned to entrain and predict the behavioral dynamics of animate motion. Section 5 outlines how representational momentum experiments could be used as a general experimental technique for investigating cognitive dynamics. Finally, because these issues become clearer when put in their proper evolutionary-functional context, section 6 explores the interplay among evolutionary, cognitive, and behavioral dynamics by examining (1) the evolutionary origins of dynamical representations, arguing that they are generally the most efficient and adaptive way for cognitive systems to entrain and extrapolate real-world dynamics that are relevant to important problems of survival and reproduction, particularly the behavioral dynamics of animate motion, and (2) the evolutionary effects of dynamic representations, arguing that the widespread phenomenon of `protean' (adaptively unpredictable) behavior (Driver & Humphries, 1988) results from the unique selective pressures imposed by dynamical representations capable of entraining, extrapolating, and predicting the behavior of other animals.
This paper is not available online