Assessing the occurrence of elaborative inferences: Lexical decision versus naming

This research extends previous attempts (e.g., G. McKoon & R. Ratcliff, 1986, Journal of Experimental Psychology: Learning, Memory, and Cognition, 12, 82-91; M. Singer & F. Ferreira, 1983, Journal of Verbal Learning and Verbal Behavior, 22, 437-448) to determine whether subjects infer highly likely consequences of an event while reading. For example, if subjects read about someone falling off a 14th story roof, will they infer that the person died? The present experiments differ from previous research in that they use procedures-lexical decision and word naming-which we feel permit a clearer assessment of whether the inference occurred at encoding, they directly compare the effects of requiring versus not requiring the inference for coherence in the absence of the usual confounds with semantic relatedness, and they examine the effects that disconfirming evidence has on the activation level of the inferred concept. The lexical decision task appeared to indicate that subjects do, indeed, infer the likely consequences of events while reading. However, the naming task suggested that such inferences were not drawn. We conclude that such inferences are not drawn while reading and that the lexical decision results, as well as previous results using a recognition task, reflect context checking that occurs at the time of the test.