Three experiments are reported comparing memory for sentence sets with perfectly correlated predicates (correlated sets) to memory for sets in which predicates are not perfectly correlated (standard sets). Subjects learned sentence sets then received timed recognition tests on old sentences intermixed with distractors formed by re-pairing subject-terms and predicates. In the first two experiments, standard sets gave rise to the fan effect, that is, longer recognition latencies as the number of facts learned about a concept increased, whereas correlated sets did not. The third experiment indicated that predicate correlation reduces the fan effect only if subjects give the same encoding to separate tokens of a predicate; with variable encoding of predicate tokens, correlated sets give rise to a fan effect of the same magnitude as that for standard sets. The results are consistent with the idea that predicate correlation produces integrated conceptual units parallel in certain respects to natural categories.
ASJC Scopus subject areas