IC‐P‐060: Conversion from mild cognitive impairment to Alzheimer's disease over 12 months: Predictive value of Aβ imaging with 18F‐Florbetaben

in mild cognitive impairment patients subsequently diagnosed with probable Alzheimer’s disease (Olichney et al., 2008), and a feedback-locked P300 which we believe is reflective of participants’ confidence in their category learning. Younger adults showed similar patterns for the late positive complex and P300 event-related potentials as in our original study (Morrison, Reber, and Paller, 2009). Because of the previously described differences in accuracy between the rule and no-rule subgroups, we examined event-related potentials separately in these two subgroups. The rule subgroup showed a larger late positive complex for correct than incorrect trials, similar in magnitude to that observed for younger adults. The no-rule subgroup showed no late positive complex differences between correct and incorrect trials. The rule subgroup also showed a feedback P300 difference for correct/incorrect trials, but it was smaller than that in younger adults, suggesting decreased rule-learning confidence (despite similar accuracy across groups). The no-rule subgroup showed event-related potentials characteristic of chance performance, with no differences between correct and incorrect trials in the feedback-locked P300, paralleling the lack of an effect on the stimulus-locked late positive complex.There were no significant differences in the control task between younger and older adults. Rule and norule older adults also performed similarly showing performance approximately equal to the asymptotic categorization performance for the rule-subgroup during the primary task. This suggests that no-rule subgroup categorization performance was likely reflecting deficits in long-term memory or executive functions. Conclusions: Preliminary results in this study showed that a simple rule-based category-learning task elicited differential performance and event-related potential findings as a function of age. We focused on findings from a group of older adults known to have long-term memory within the normal range on standard neuropsychological tests. Yet, some members of this group were successful in learning to categorize stimuli and some were not. We thus identified two distinct subgroups of healthy older individuals, one subgroup with category learning similar to college students and the other with dramatic deficits. Event-related potentialss recorded during categorization confirmed this stratification. Specifically a stimulus-locked late positive complex believed to index long-term memory was preserved in older adults who learned to categorize, but not in the others. Secondly a feedback-locked P300 believed to reflect the degree to which the participant is confident in their category learning showed a graded decrement. P300 differences for feedback on correct versus incorrect trials were greatest in amplitude in younger adults, intermediate in rulelearning older adults, and smallest in those older adults who failed to learn to categorize. The current study suggests that rule-based category learning, a task utilizing long-termmemory and executive functions, may be an effective means for identifying individuals at increased risk for mild cognitive impairment and probable Alzheimer’s disease. Event-related potentials recorded during this task may provide a more sensitive measure of changes in cognition compared to simple neuropsychological tests by measuring neural function even when participants perform well, thus providing an objective measure of confidence in learning. Future work will be directed at testing patients with mild cognitive impairment and Alzheimer’s disease using this paradigm.