Modeling Human Knowledge of Routes: Partial Knowledge and Individual Variation

Commonsense knowledge of large-scale space (the cog-nitive map) includes several diierent types of knowledge: of sensorimotor, topological, and metrical spatial relationships. Sensorimotor knowledge is deened as that knowledge which is necessary to reconstruct a route from memory after travel along that route in a large-scale environment. A representation for route knowledge is proposed with suuciently robust performance properties to be useful as commonsense knowledge. Its states of partial knowledge are shown to correspond to those observed in humans. We also deene and explore the space of all possible variants of this representation, to derive empirical predictions about the nature of individual variation.