Mental models contribute to foregrounding during text comprehension

A primary property of mental models is that they represent what the text is about (the events, objects, and processes described in the text), rather than features of the text itself. We used this property to demonstrate that mental models are operative during text comprehension. Subjects read texts that were propositionally equivalent, but described events in which the main actor was either spatially associated with a target object or spatially dissociated from the object. Pronominal reference kept the actor foregrounded throughout the text, but the target object was never repeated. The question of interest was whether the target object remained foregrounded when the text described events in which the actor and the object were spatially associated. Data from experiments using item recognition and reading time measures provided an affirmative answer. Thus the mental model controlling foregrounding reflected the structure of the events described by the text, not just the structure of the text.