Route planning with regionalized spatial memory

According to the hierarchical theories of spatial representations, places are grouped together into regions which form higher level nodes of a graph-like representation of space. Support for these theories comes from a wide variety of experiments using different methods (distance- and directional judgments, priming, recall). However, the ultimate purpose of an internal representation of space is to allow navigation, route planning, and directed movements through the environment. To our knowledge it is still an open question whether or not hierarchically structured spatial representations influence navigational abilities such as route-planning behaviour of humans. By employing a navigational task in a virtual environment we provide additional evidence for the hierarchical theories of spatial representations, and their role in route planning. Subjects navigated through a virtual environment that was presented on a large half cylindrical projection screen. They learned the positions of twelve landmarks within the environment. According to the object category of the landmarks, the environment could be grouped into three different semantic regions. Subsequent to an exploration- and training-phase subjects were asked to navigate the shortest route connecting three of the objects. All test routes featured two alternative solutions of equal length that only differed in the number of region boundaries they passed by. Subjects preferred routes that minimized the number of region boundaries that were passed by. These results provide additional support for the hierarchical theories of spatial representations. Additionally the findings suggest that route planning is based on region-connectivity, not place-connectivity alone. We propose a planning mechanism that uses coarse space information (region-connectivity) for the goal(s) and fine space information (place-connectivity) for the current location. Such a planning mechanism would reduce computational effort by using a graph theoretic distance measure, i.e. the number of region boundaries crossed along a route