It is a well-documented fact that normal aging produces a substantial decline in explicit memory ability (e.g., Salthouse, 1982), but studies of implicit memory present a markedly different picture. In a recent review, Graf (1990) concluded that age differences on implicit memory tasks were very small, averaging around 4%, and that interpretation of even these small differences was confounded because young and old differ in the extent to which they use explicit memory to facilitate implicit performance. This point reflects the difficulty in constructing implicit memory tasks that are purely implicit in nature, that is, tasks that, even if explicit memory of the learning episode is available, receive no benefit from that memory in task performance. Because of these confounding explicit influences,

the accurate investigation of age-related changes in implicit memory is more difficult than would initially seem the case.