Qualitative and Quantitative Reasoning and Instance-Based Learning in Spatial Orientation

This paper describes an experiment and a computational cognitive model involving a spatial orientation task. The experiment tests participants’ ability to identify their location on a map, given a view of the space. The model performs this task by applying an instantiation of a general strategy that has been shown to be effective in other spatial orientation tasks. It uses perceptual grouping to organize the space into recognizable elements (clusters), combined with qualitative (left versus right) and quantitative (egocentric bearing) information about those clusters to determine its response. The model is compared to the empirical data, showing good overall correspondence to human performance, including response times and errors.