Reinstatement of causal information during reading.

An on-line word naming probe was used to test whether information presented earlier in a text, and then backgrounded by several sentences, would be reinstated when Ss were required to understand the cause of a currently processed action or event. In Experiment 1, Ss named a probe word that represented an earlier-mentioned cause more quickly when it followed a causal coherence break than when it followed a neutral sentence. Experiments 2 and 3 replicated this effect and examined 2 conditions that may affect the process of reinstating a cause: (a) Inclusion of part of the context in which the cause was originally presented was not necessary to obtain reinstatement of the cause and (b) reinstatement of the cause was not evidenced when it had been disconfirmed earlier in the text.