Effects of task on the activation of predictive inferences

ABSTRACT Research on the activation of predictive inferences has provided inconsistent results that may be explained within a contextual view of reading. The present study tested whether the type of test for explicit memory would affect the activation of knowledge-based predictive inferences. The information necessary for the activation of a predictive inference was provided to readers in four different conditions (no inference, local processing, global processing, coherence). Manipulation was accomplished by varying the type of question asked after reading the passage (verbatim, factual, or inference). Analysis suggests predictive inferences are automatically activated and nor affected by contextual factors such as the question. Consequently, the current data do nor provide clear support for a contextual view of comprehension. These conclusions are supported by a two-stage view of elaborative processing