The representation of space in metal models derived from text

Mental models of text are representations of what the text is about (i.e., situations), rather than representations of the text itself. Many mental model theories propose that mental models, like real situations, are played out in a medium analogous to a Euclidean space so that distance has functional consequences. For example, when one mentally manipulates one element of the representation, one will notice other elements that are spatially close to it, and its will enhance their short-term accessibility. In a test of this noticing hypothesis, participants read texts that described the object-by-object construction of a spatial layout. According to the text, a critical object ended up close to a target object (in the spatial layout) or far from the target object. In neither case, however, was the relation between the critical object and the target object explicitly described in the text. The noticing hypothesis predicts that the accessibility of the target object will be enhanced when the critical object is close to it. We tested this prediction in seven experiments in which we also manipulated the number of objects described, whether the description was accompanied by a diagram, the presentation modality of the description, the number of dimensions in the spatial layout, and the measurement of accessibility. We failed to find consistent support for the noticing hypothesis. The data compel the conclusions that (1) spatial representations can be formed when one is reading, (2) these representations do not support automatic noticing of implicit spatial relations, (3) it is likely that the spatial representation is more topological or functional than Euclidean.