R2-[BackSpace] - Overview

Effects of Background Knowledge on Human Spatial Reasoning


The starting point of this project is the hypothesis that background knowledge can significantly influence the performance of individuals and computational systems when these solve spatial reasoning problems. This influence can either be a facilitation or an impedance of the reasoning process, depending on whether the described arrangement is congruent or incongruent with prior knowledge. The aim of the proposed project thus will be to determine the precise circumstances under which background knowledge supports or hinders spatial reasoning. For this purpose, the project will combine experimental methods from cognitive psychology, functional brain imaging studies, and computational investigations. In the psychological experiments, the difficulty of the reasoning problems and the availability of background knowledge will be varied systematically. The experiments will also take into account that spatial reasoning problems usually occur either in a verbal or a pictorial format. The result will be an exact determination of the circumstances under which an effect of reasoning facilitation or impedance appears. The brain imaging studies will be performed as such methods provide a powerful tool to investigate cognitive processes on the cortical level. In these studies, we will determine under which circumstances a reasoning problem activates the areas of the brain that are typically related to reasoning processes in working memory, or areas related to retrieval of prior knowledge from long-term-memory. The psychological studies will be complemented by computational investigations to determine the relation between cognitive and computational complexity and to investigate how spatial reasoning difficulty can be reduced by background knowledge. The output of the project will be a detailed theory of human reasoning with spatial relations and background knowledge. This theory will cover all three levels of explanation: cognitive performance, algorithmic realization, and physical implementation. As these findings will be important for a wide range of technical issues, the project will provide psychological facts that can be used within the SFB/TR to develop cognitively plausible computational systems. The project will especially collaborate with the projects R1-[ImageSpace] and I2-[MapSpace] in which psychological data are needed for a fine-grained and cognitively plausible model of spatial inference processes.