ABSTRACT
According to the minimalist hypothesis (McKoon & Ratcliff, 1992), the only inferences routinely generated during reading (without special goals or strategies), are those based on quickly and easily available information and those required for local coherence of the text being read. The minimalist hypothesis summarizes the past 20 years of text processing research. During that time, little uncontestable evidence has been found to support the notion that readers engage in the constructive processes that would be required to generate a large number and variety of inferences. Most of the data that does show constructive inferences has been collected in situations where readers could plausibly be thought to use special efforts, goals, or strategies directed toward the kind of inference under investigation. For the great majority of experiments, in which there is no reason to think such efforts took place, the minimalist hypothesis is a description of the few kinds of inferences that might be drawn in the absence of special goals or strategies.
