Stochastic independence between two implicit memory tasks.

Two experiments tested the statistical relation between performances on two implicit, priming tests of memory: word-fragment completion and perceptual identification. If performance on implicit priming tests is mediated by a single memory system, then stochastic dependence between them should be found. Contrary to this prediction, stochastic independence was obtained between performances on the two tests in all but one condition. For that condition (Experiment 1) the results suggest that subjects may have knowingly used information derived from the first test to complete the second. When perceived similarity was eliminated by altering contextual cues (Experiment 2), stochastic independence was found in this condition as well. The results are not consistent with current multiple memory system interpretations that assign all implicit, priming tests to one system and explicit tests to another. An alternative interpretation is that the degree of dependence between performances on memory tests is determined by the similarity of the component processes that the tests engage and of the information they use. Evidence for a dissociation between performances on explicit and implicit tests of memory has been based on demonstrations of functional and statistical independence between the two types of tests. In this article we show that statistical independence can also occur between two implicit tests of memory. Explicit memory tests, such as recognition and recall, are