Building spatial mental models: Encoding and retrieving descriptions of space

Seven experiments were conducted to examine the development and application of abstracted and perspective-flexible mental representations, or spatial mental models. Participants read spatial descriptions written in route and survey perspectives and undertook several memory tasks demanding the application of both declarative memories and spatial mental models. Experiment 1 manipulated reader experience by varying description exposure and found that while spatial mental models were available after only one read of a survey description, several reads were necessary to develop functionally similar spatial mental models from route descriptions; this is in contrast to declarative memories which varied only as a function of experience and not description perspective. Experiment 2 examined self-paced description reading times and found that readers tended to be slower overall while reading spatial sentences during route descriptions over the course of three read cycles, supporting Experiment 1 results; this is in contrast to non-spatial sentences which showed no differences as a function of description perspective. Experiments 3 and 4 used a dual-task methodology to examine the working memory mechanisms involved during spatial description reading, and demonstrated interactive involvement of visuospatial, articulatory, and central executive resources towards declarative memory and spatial mental model development. Experiments 5 and 6 used the same dual task methodology during retrieval of spatial mental models, and found that these models are not necessarily a collection of inferences developed during reading but rather a foundation from which to extract information towards inference generation; further, they demonstrate that both relatively coarse surface representations of the text and spatial mental models are available for retrieval. Finally, Experiment 7 examined the influence of goals and perspective preferences on spatial mental model development and found that overall those with a survey preference were better-able to develop perspective-flexible spatial mental models, while those with a route or landmark preference were better-able to learn landmark identities; goal instantiation did not show any effects. Taken together the results demonstrate that the development and application of spatial mental models are contingent upon multiple working memory systems that interact importantly with representational formats, extent of experience, and reader preferences.