Two experiments investigated the proposition that the amount of cognitive effort expended to encode information will be directly related to recall of that information. While previous research has shown that amount of processing may affect recall, these studies have generally drawn on the notion of an elaborated memory representation to explain their results. In this study, the amount of processing required to correctly interpret anaphoric relations was varied while the elaboration of the memory trace was held constant. These experiments employed a self-paced reading paradigm in which subjects read a series of short paragraphs and later were cued to recall the final sentence of each paragraph. It was found that recall was significantly improved when more processing was required to correctly interpret the anaphoric relationship expressed in the final sentence. These findings suggest that encoding processes can affect recall performance without elaboration of the memory representation.Although the original formulation of the levels-ofprocessing framework (Craik & Lockhart, 1972) has been the subject of considerable criticism (see Baddeley, 1978;Nelson, 1977), revised versions of the model have been proposed that attribute differences in recall to various encoding activities. These versions can generally be divided into two groups. The elaboration model, proposed by Anderson and his colleagues (Anderson & Reder, 1979), claims that deeper processing at encoding results in the creation of a set of multiple propositions that facilitate recall of the information. The distinctiveness theory, advanced by Jacoby Jacoby, Craik, & Begg, 1979), predicts that deeper processing at encoding results in a more distinct memory representation that has a greater probability of being recalled at time of retrieval. In the present article, both of these models will be outlined and the results of two experiments will be reported that test the adequacy of the models to account for differences in the retention of prose.The elaboration model, as proposed by Anderson and Reder (1979), suggests that different encoding processes result in more or less elaborated memory traces. Passages or sentences vary in the degree to which they encourage readers to draw inferences about consequences, causes, or outcomes. When the reader is able to draw an inference, the inferred information is stored along with the explicitly stated information. The storageWe are grateful to Jeff franks, Miriam Goldberg, and John Limber for their helpful comments on earlier drafts of this manuscript. The second author is now at the Professional Examination Service, 475 Riverside Drive, New York, New York 10115. of inferred and explicit information results in an elaborated memory trace. The degree to which a passage or sentence is elaborated upon depends in part on task demands, readers' prior knowledge, and other nontextual factors. Bradshaw and Anderson (1982) suggest that elaborated traces are more easily recalled for two reasons. First, the presence of an elaborated trac...