Strength and duration of priming effects in normal subjects and amnesic patients.

In three separate experiments, we assessed the strength and duration of word completion effects in amnesic patients and two control groups. In Experiment 1 subjects studied words under a semantic orienting condition and were given tests of word completion and recognition memory after an immediate, 2-hr or 4-day delay. In the word completion test for Experiment 1, we presented three-letter word stems that could be completed to form several common words, one of which had been presented previously (e.g. MOT for MOTEL), and subjects completed each stem with the first word that came to mind. Priming effects were equivalent in amnesic patients and control subjects and they reached baseline levels within 2 hr. In Experiments 2 and 3, subjects studied words under either a semantic or a nonsemantic orienting condition, and word completion was tested at the same three delays using cues that uniquely specified the study words (e.g. JUI for JUICE; or A--A--In for ASSASSIN). In these experiments, amnesic patients exhibited both smaller and shorter lasting word completion effects than control subjects. Specifically, amnesic patients exhibited word completion effects that seldom lasted as long as 2 hr, whereas control subjects usually exhibited completion effects lasting 4 days. An important additional finding was that control subjects exhibited larger and longer-lasting word completion effects when tested under the semantic orienting condition than when tested under the nonsemantic orienting condition. Amnesic patients were not affected by this manipulation. Moreover, under the nonsemantic orienting condition, control subjects and amnesic patients performed similarly. The results show that word completion performance is not always fully intact in amnesic patients. Long-lasting word completion effects found in normal subjects may be mediated by declarative or elaborative retrieval processes, which are impaired in amnesic patients. If so, priming as measured by word completion methods is shorter lasting than recent studies of normal subjects would suggest.