DESCRIPTION (adapted from investigator's abstract): Computation models of memory storage have recently gained increased popularity because they account for a variety of findings observed in patients with human amnesia. These models also lead to two predictions regarding the pattern of deficits seen in patients with semantic dementia. This application requires six months of support to test these predictions in a well characterized sample of semantic dementia patients followed at the Medical Research Council in Cambridge, England. The first two studies use both explicit and implicit probes of memory to test the prediction that semantic dementia patients should show better preservation of semantic knowledge acquired in recent time periods than in more remote time periods. The second two studies test the prediction that new learning is intact in these patients when the need for semantic elaboration at encoding is minimized. These studies examine the status of two qualitatively different processes than can mediate new learning, namely recollection and familiarity. Taken together with the existing findings in patients with amnesia, these studies of patients with semantic dementia will provide a powerful test of current models of memory storage in the human brain.