Abstract

Three experiments investigated level of processing effects on a variety of direct and indirect memory tasks, in the context of a processing theory of dissociations. Subjects studied words in five encoding conditions and received one of ten memory tests. In Experiment 1, four tests previously classified as conceptual showed a robust LOP effect, as did a direct perceptual test of graphemic cued recall. An indirect perceptual word fragment completion test was unaffected by LOP. Experiment 2 showed that a new indirect version of a graphemic cued test was not affected by LOP. In Experiment 3, guided by a generation/recognition model, we constructed three new direct tests in which subjects identified words that were graphemically, phonologically, or semantically similar to studied words. The three tests differed in their sensitivity to study conditions, but LOP had no effect in any case, despite the involvement of deliberate conscious recollection. Contemporary explanatory frameworks couched as dichotomies do not provide an adequate account of the results. It seems necessary instead to specify the types of information activated by each encoding condition, the types of information required by each test, and how encoding and retrieval processes are modified by task instructions