From low to high level approach to cognitive control

Abstract

In this Chapter the application of dynamical systems to model reactive and precognitive behaviours is discussed. We present an approach to navigation based on the control of a chaotic system that is enslaved, on the basis of sensory stimuli, into low order dynamics that are used as percepts of the environmental situations. Another aspect taken into consideration, is the introduction of correlation mechanisms, important for the emergence of anticipation. In this case a spiking network is used to control a simulated robot learning to anticipate sensory events. Finally the proposed approach has been applied to solve a landmark navigation problem. © Springer-Verlag Berlin Heidelberg 2009.