Dynamics, Control, and Cognition

Once upon real-time A dynamic object is an object whose properties change over time. A static object is an object whose properties do not change over time. Given such an idealization, the notion of ‘static’ lies at an extreme end of the spectrum of temporal relations between objects and properties. Indeed, modern physics tells us that no objects are truly static. Nevertheless, many of our physical, computational, and metaphysical theories turn a blind eye to the role of time, often for practical reasons. So, perhaps it is not surprising that in the philosophy of mind – where physical, computational, and metaphysical theories meet – there has been a consistent tendancy to articulate theories that consider function and time independently. As a result, contemporary theories in cognitive science consider time unsystematically (see the next section for specific examples). In this chapter, I suggest that the problem with this ‘ad hocery’ is that the systems we are trying to characterize are real-time systems, whose real-time performance demands principled explanation (a point on which many of these same contemporary theorists agree). After a discussion of the importance and roots of dynamics in cognitive theorizing, I describe the role of time in each of the three main approaches to cognitive science: symbolicism, connectionism and dynamicism. Subsequently, I outline a recently proposed method, the Neural Engineering Framework (NEF), that, unlike past approaches, permits a principled integration of dynamics into biologically realistic models of high-level cognition. After briefly presenting a model, BioSLIE, that demonstrates this integration using the NEF, I argue that this approach alone is in a position to properly integrate dynamics, biological realism, and high-level cognition.