Abstract
Patients with Alzheimer's disease (AD) often exhibit an abnormally liberal response bias in recognition memory tests, responding "old" more frequently than "new." Investigations have shown patients can to shift to a more conservative response bias when given instructions. We examined if patients with mild AD could alter their response patterns when the ratio of old items is manipulated without explicit instruction. Healthy older adults and AD patients studied lists of words and then were tested in three old/new ratio conditions (30%, 50%, or 70% old items). A subset of participants provided estimates of how many old and new items they saw in the memory test. We demonstrated that both groups were able to change their response patterns without the aid of explicit instructions. Importantly, AD patients were more likely to estimate seeing greater numbers of old than new items, whereas the reverse was observed for older adults. Elevated estimates of old items in AD patients suggest their liberal response bias may be attributed to their reliance on familiarity. We conclude that the liberal response bias observed in AD patients is attributable to their believing that more of the test items are old and not due to impaired meta-memorial monitoring abilities.