Better preservation of memory span relative to supraspan immediate recall in Alzheimer's disease

It is suggested that Alzheimer's disease (AD) patients are able to recall more items on the digit span task than on immediate free recall from a supraspan word list. Two experiments were undertaken to verify this assertion and to understand the basis of the putative span/supraspan discrepancy. The first experiment, involving 35 mildly or moderately demented AD patients, confirmed that digit span significantly exceeded immediate recall from a 10-item supraspan word list. Although digit span also exceeded supraspan recall in 38 elderly non-demented control subjects, the discrepancy was significantly greater within the AD group. In a second experiment, 19 AD cases and 20 controls were assessed with a word span task that used nouns matched by frequency and word length to nouns on the supraspan task. The magnitude of the span/supraspan discrepancy was reduced, indicating that part of the initial discrepancy was due to differences in stimulus items (digits versus common nouns). As before, AD subjects recalled more words on the span task than the supraspan task. However, in striking contrast, NC subjects recalled more words on the supraspan task, further indicating that AD patients are particularly impaired on supraspan recall. Using combined data from 106 subjects in both experiments, digit span performance correlated significantly with supraspan recall for NC but not AD subjects. Moreover, within the AD group the magnitude of the discrepancy was inversely related to a working memory measure derived from the backward digit span. The magnitude of the span/supraspan discrepancy correctly classified 88 c of patients with mild dementia and 740 of controls. Results indicate that AD patients are specifically vulnerable to information overload inherent in the supraspan task, a view consistent with the perspective that AD is characterized by prominent disturbances in working memory. (C) 2002 Elsevier Science Ltd. All rights reserved.