Long-term spatial representations from pictorial and textual input

Abstract

Spatial information is typically represented in memory in a form that facilitates operations like spatial reasoning, mental navigation, search, and perspective shifts. This apparently happens whether learning is through primary experience or description, and it appears to happen whether the learned configuration is large- or small-scale. Typical methods used in the cognitive map and mental models literatures diverge, however, in several ways that might affect representation (e.g., medium of input at learning, size of the referent situation, and delay from learning to test). The cognitive map literature shows that spatial representation of large-scale environments persists, though often biased toward spatial simplification over time. This represents one optimal solution to the problem of retaining information in a form that facilitates likely cognitive operations while minimizing resources devoted to storing the information. The current studies investigate whether mental models learned from description are subject to the same tendencies or whether they revert with time to a less spatial format that better optimizes cognitive economy. Secondly, these studies investigate the extent to which visual input at learning affects how the information is represented.