ABSTRACT
Inferences are a major source of coherence in reading. For more than 20 years considerable research has been devoted to understanding what sorts of inferences contribute to coherence during reading, and the fecund collection of results issued from this research has lead to several recent attempts to produce a large-scale account of text comprehension (e.g., Graesser & Bower, 1990; Graesser, Singer, & Trabasso, 1994; Kintsch, 1988; McKoon & Ratcliff, 1992). Missing from this work, however, is any account of how propositional-logic inferences are made during reading. In fact, even the most promiscuous (with regard to inference-making) of these theories takes a rather pessimistic view of the ease with which logical inferences can contribute to comprehension; in discussing the sorts of inferences that are made during reading (online), Graesser and his colleagues stated: “some classes of inferences [are] normally difficult to generate and are therefore off-line. First, there are logic-based inferences that are derived from systems of domain-independent formal reasoning, such as propositional calculus, predicate calculus, and theorem proving” (Graesser et al., 1994, p. 376). More minimal hypotheses about inferencemaking claim that very few inferences are made during reading-only those needed for one sentence to make sense in the context of the previous sentence, and those that can be derived from easily available information (McKoon & Ratcliff, 1992, 1995). Thus, current theories of text processing present a somewhat inhospitable frame from which
to consider the poten-tial contribution that logical inferences might make to text comprehension.
