Item

Abstract

Two main classes of mental representations describe navigable environments: single coordinate systems and graphs. In coordinate systems each location can be assigned a specific coordinate (maybe on a hierarchal sub-level). Graphs represent local environments as nodes (e.g. a street or room) and the relations between close-by nodes (e.g., two meters straight, one to the left). One difference between graphs and coordinate systems is that coordinate systems have to be consistent; each coordinate refers only to one location and vice versa. This is important, for example, when walking round a block encountering the start location again which was thought to lie further down the street. The start and the current location are identical and represented at the same coordinate. The erroneous estimation of the streets in the loop must be adjusted to fit into the coordinate system. In graphs such consistency is not required; local street length and turns do not have to be consistent in a global metric. To test these predictions participants walked multiple times through a virtual corridor environment roughly 80% round the loop when they visually encountered the start again. Afterwards, they judged the spatial relation between adjacent locations along the loop. The majority of participants adjusted the spatial relations as if they were in a smaller loop than the one walked. However, they adjusted them not far enough to fit into a coordinate system, which would have required an even smaller loop. These results are inconsistent both with classical graph and coordinate representations.