Dissociations between word priming effects in normal subjects and patients with memory disorders: multiple memory systems or retrieval?

Repetition priming was measured in two different tasks within a single experiment--one in which subjects named briefly (tachistoscopically) presented words, and one requiring naming of visually fragmented/degraded words. Thirteen amnesic patients, 12 patients with Alzheimer's disease (AD), and 15 normal control subjects were tested under 4 experimental conditions involving the factorial combination of two variables: delay of test (10 minutes and 24 hours), and number of prior occurrences of the primed items (1 and 4). The two tasks produced very different patterns of priming effects, despite the fact that common study phases were employed. In one task the priming effect showed no decay and virtually no effect of the number of prior occurrences of the primed items, whereas both these variables affected priming in the other task. The AD patients evidenced impaired priming in both tasks. However, in the degraded-word-naming task the deficit was only apparent under some experimental conditions. The amnesics produced priming effects that in absolute terms were similar to those produced by control subjects. However, when group differences in overall performance level were taken into account in the tachistoscopic task, these patients also showed clear evidence of impaired priming. It is argued that the complex pattern of priming effects obtained is best explained by the characteristics of the retrieval cues provided in the tasks, and, generally, that such characteristics may determine whether or not experimental variables will affect measured priming.