BACKWARD UPDATING OF MENTAL MODELS DURING CONTINUOUS READING OF NARRATIVES

Participants' ability to track the protagonist's position and surroundings, during continuous reading, was investigated. In Experiment 1, participants read passages involving either inside-outside or top-down topological relations. A typical story described the protagonist interacting with 1 object, which was either consistent or inconsistent with his or her location. The results show that it took longer to read the sentence in the inconsistent condition. Experiments 2-5 used recognition probe words to test the accessibility of both consistent and inconsistent objects. The results show that participants did not update the situation model when the last sentences did not mention again any target object (Experiments 2 and 3). However, the mention of an object by means of an ambiguous pronoun triggers the updating of the situation to resolve the antecedent. The updating starts immediately after reading the pronoun, and the target still remains activated at the end of the sentence (Experiments 4 and 5). The overall results establish boundary conditions for mental model updating.