The processing of predictive and bridging inferences, with reference to age- and task-related differences

Five experiments explored inferencing in young and elderly subjects. Experiments 1 and 2 used the contradiction reading paradigm, a frequently used paradigm to study the formation of elaborative inferences. Subjects read short passages followed by a continuation sentence that either contradicted or was neutral to the elaboration. In Experiment 1, 1-2 sentences separated the predicting context and the contradiction sentence. In Experiment 2, 5-6 sentences separated the predicting context and the contradiction sentence. Overall, elderly subjects were slower than young subjects. Both young and elderly subjects read the contradiction sentence more slowly than the neutral sentence in Experiment 1. In Experiment 2 the effect was seen on the sentence following the contradiction sentence.;Experiments 3, 4 and 5 explored the possibility that the results in Experiment 1 and 2 arose not because subjects formed elaborative inferences on-line but only because bridging inferences were formed upon presentation of the contradiction sentence. Only young subjects were tested in these experiments. Experiments 3 and 4 were replications of experiments 1 and 2 with the use of a different paradigm---a probe-word naming paradigm. In this paradigm, a probe word that was either related to the elaborative inference or was neutral was presented either 1-2 sentences or 5-6 sentences after the predicting context. In this paradigm faster naming times for the related than the neutral word are expected if subjects form elaborative inferences (i.e., facilitation). Results in both experiments were unexpected and showed a marginally significant inhibition effect (i.e., longer naming times in the related compared to the neutral condition). These results raise questions about whether elaborative inferences are formed on-line. In Experiment 5, the stimuli from Experiment 3 were altered to include a sentence that would encourage the formation of bridging inferences. In this experiment, there was no significant difference between the related and neutral conditions. Overall, the results call into question the hypothesis that elaborative inferences are formed on-line.