Dissociations between implicit and explicit memory have been observed across a wide variety of tasks and conditions, as documented by recent review articles (cf. Richardson-Klavehn & Bjork, 1988; Roediger, 1990; Schacter, 1987) and by other chapters in this volume. Despite the apparent ubiquity of such dissocia­ tions, it is probably safe to say that the most striking separation between implicit and explicit memory is observed in the amnesic syndrome: Densely amnesic patients perform poorly on explicit tests of memory, but they perform remarkably well, and frequently normally, on numerous implicit tests (e.g., Cohen & Squire, 1980; Graf, Squire, & Mandler, 1984; Milner, Corkin, & Teuber, 1968; Moscovitch, 1982; Schacter, 1985; Shimamura & Squire, 1984; Warrington & Weiskrantz, 1974). In addition to providing some of the strongest empirical grounds for distinguishing between implicit and explicit memory, these dissocia­ tions can provide potentially important insights for both cognitive and neurobiological theories of mnemonic processes. On the one hand, observations of preserved implicit memory in amnesia provide important constraints for cog­ nitive theories: If a theory does not speak to or cannot accommodate the amnesia data, then it fails to explain a critical aspect of implicit memory. On the other hand, data concerning implicit memory in amnesic patients can aid neurobiological formulations by providing insights into the function of the hippo­ campus and related limbic structures that are typically damaged in amnesia (e.g., Milner et al., 1968; O’Keefe & Nadel, 1978; Squire, 1992), and can also be informative regarding the cortical structures that are typically preserved in am­ nesia (cf. Schacter, 1990, 1992a; Squire, 1992). Indeed, attempts to fully charac­ terize the computations that these structures perform should be informed by, and

must be consistent with, the known implicit memory abilities of amnesic pa­ tients.