The time course of propositional-logic inferences in text comprehension
Four experiments investigated a claim in a model of propositional-logic reasoning proposed by Braine, Reiser & Rumain (1984) that inferences such as p or q; not p/therefore q are made spontaneously by readers at the moment both premises are available. This claim is inconsistent with some evidence in the text processing literature which suggests that only those inferences necessary for textual cohesion are made spontaneously. Subjects read stories in which a logical inference was not necessary to maintain textual cohesion and their inference making was assessed using a priming technique in which subjects performed lexical decisions on targets that were semantic associates of the proposition that could be inferred. Latencies of these lexical decisions were compared with those of control stories in which the inference could not be made. Two experiments tested two logical forms central to the Braine et al. model. Results showed that latencies for targets in the inference condition were significantly faster than controls, thus indicating that subjects were making the logical inferences. The stories in these experiments had thematic titles; two further experiments investigated whether these results could be replicated if the titles were omitted. Results from the no-title experiments indicated that subjects were making inferences even when the theme of the story was not given in advance, though perhaps less frequently than when the title was present. These findings support the prediction in the Braine et al. model regarding the time course of model inferences, and suggest that some propositional-logic inferences may be made routinely in texts even when they are not required for cohesion.