Unconscious influences on amnesics' word stem completion

Amnesic patients and control subjects were asked to complete three-letter word-stems under one of three retrieval conditions. In a direct condition, they were told to use the stems as retrieval cues for words that had just been presented in a study list. In an indirect condition, they were told to use the first word that came to mind with no reference made to the study list. Finally, in an oppositional condition, they were told to use the first word that came to mind unless it had appeared on the study list. During the study list presentation, the patients and controls had analyzed each word according to either semantic (associating to each word) or graphemic (counting letters with enclosed spaces) instructions. The results revealed that the control subjects produced a different number of study words during retrieval as a function of retrieval instructions and encoding condition. The amnesics, however, did not vary their performance as a function of retrieval instructions. Under all conditions, they completed the word-stems far more frequently with words from the study list than would be expected by chance and they consistently produced more semantic than graphemic responses. We concluded that semantic analysis might affect the fluency with which an item occurs for the amnesic, but that the item itself remains independent of the source of that fluency for these patients. Thus, the level of analysis performed on a word during study can affect the unconscious performance of amnesic patients but is unavailable for use during conscious retrieval. (C) 1997 Elsevier Science Ltd.