WORKING-MEMORY AND INFERENCE REVISION IN BRAIN-DAMAGED AND NORMALLY AGING ADULTS

This study examined the association between estimated working memory ( WM) capacity and comprehension of passages that required revision of a n initial interpretation. Predictions stemmed from the recently elabor ated theory of capacity-constrained comprehension (Just & Carpenter, 1 992, Psychological Review, 99, 122-149), which includes as a major fea ture the principle that WM influences comprehension only as processing demands approach or exceed the limits of capacity. As anticipated fro m task analysis, correlations between unilaterally brain-damaged patie nts' estimated WM capacity and discourse comprehension performance wer e minimal for nondemanding measures, and increased in magnitude with t ask processing requirements. Most notably, a meaningful correlation (/ r/ greater than .50) emerged only for the task judged to involve the m ost demanding comprehension processes, for adults with right hemispher e brain damage. No meaningful associations between estimated WM capaci ty and task performance were observed for normally aging subjects, who were not expected to have difficulty with any of our comprehension me asures. The nature of WM deficits in brain-damaged adults (total capac ity, vs. resource allocation, vs. slow or otherwise faulty component p rocessing operations) is considered, and some existing work is interpr eted from a cognitive resource perspective. Theoretical implications a nd clinical applicability of the working memory/resource framework are also discussed.