Assessing the Ability of Multidimensional Scaling and Pathfinder Networks to Measure Spatial Knowledge

Abstract

Cognitive maps, or mental representations of external environments, aid spatial navigation. Typically, researchers study cognitive maps by having participants provide a sketched map. However, multidimensional scaling (MDS) and Pathfinder, statistical techniques which represent a set of input proximities as a n-dimensional space or a network, respectively, can both be used as measures of cognitive maps. Previous research with semantic knowledge suggests that Pathfinder is better than MDS for mental modelling. In the present study, participants drew maps of a familiar environment from memory and provided pairwise distance ratings for landmarks present in those locations. Using those distance ratings as inputs for MDS solutions and Pathfinder networks, the extent to which MDS and Pathfinder related to the participant sketch maps was assessed. Results indicated that MDS solutions were more highly correlated with sketch maps than were Pathfinder networks.