This interdisciplinary project intends to empirically identify factors that influence human spatial reasoning and planning processes, formally specify these with the help of formal and computational techniques, and develop a cognitive complexity measure for human spatial reasoning.
This involves methods and techniques from psychology, computer science, artificial intelligence and cognitive science.

We establish factors responsible for reasoning and planning errors and latencies that allow identifying fallacies during mental reasoning processes. On the empirical side, behavioral, eye-tracking, and fMRI experiments are conducted to identify the underlying cognitive processes and their neural correlates. We will investigate the nature of the mental representations involved in the spatial reasoning process. The results will help understanding the nature of the underlying functions in order to implement these findings about the human spatial reasoning and planning processes into the general complexity theory.

Our cognitive complexity theory is formalized using AI methods, like parameterized complexity, in order to classify the variety of spatial problems according to their cognitive complexity. One major interest is the comparison of the cognitive complexity measures to formal complexity measures used in algorithm analysis. The goal is to develop a theory of cognitive complexity on multiple levels from biological implementation to a computational theory. The behavioral studies, eye tracking and fMRI are necessary in order to take into account the nature of the underlying representation and functions. This will lead to an extended and adapted computational theory about general complexity in humans while performing spatial reasoning and planning tasks.

We are interested in the operational complexity induced by the task representation, and its interaction with inter-individual differences, like visual intelligence, as well as reasoning and planning ability. For a detailed analysis of the planning processes, we will investigate problems of a high transformation complexity, like Tower of London and Rush Hour. The results will be incorporated in a computational model, which learns to predict operations and planning moves.