Dynamic Problem Solving in Space.

The paper presents and illustrates the strong spatial cognition paradigm in the example of catching a flying ball by means of an angular control strategy. The strategy makes constructive use of structural geometric properties of 2- and 3-dimensional space in a way that is similar to strategies used by athletes and animals to catch flying objects. We contrast this strategy to predictive approaches employed in robotics: In these approaches, geometric properties are formally reconstructed from analytical domain knowledge which athletes and dogs do not require for catching balls or frisbees. A familiar spatial constellation of a shadow projection helps to understand the geometric principles which enable a constant angular control strategy. We illustrate the structural geometry of the approach by an interactive implementation in a game engine environment that permits observing and tracing the spatial dynamics. We show how the environment can “compute" for us and illustrate how this structural geometry can be used to solve spatial cognitive tasks.