Accelerated long-term forgetting in transient epileptic amnesia: an acquisition or consolidation deficit?

Accelerated Long-term Forgetting (ALF) is the rapid loss of newly acquired memories over days to weeks despite normal retention at standard (~30minute) intervals. It has recently been described in association with epilepsy, particularly the syndrome of Transient Epileptic Amnesia (TEA). The cognitive mechanisms underlying ALF remain uncertain, but disruption either of memory acquisition or consolidation processes has been postulated. To arbitrate between these accounts, we reanalysed data from an existing word-list recall data set to investigate whether ALF can be observed for words learned under precisely matched conditions in TEA patients and controls. We reanalysed the data of 24 patients with TEA and 24 matched healthy controls who learnt a 15-item word list to a learning criterion of 90% with a minimum of 5 learning trials. Free recall of the words was probed at delays of 30minutes and one week and three weeks after learning. In addition, a 'yes-no' recognition test was conducted after the three-week free recall. Forgetting rates across the first 30minute delay and the subsequent one week and three week delay were compared between patients and controls. To ensure that learning conditions were closely matched between patients and control participants, we excluded exceptionally fast (NTEA=1, Ncontrols=4) and slow (NTEA=6Ncontrols=2) learners. Furthermore, we analysed only words that were presented 5 or 6 times during learning and retrieved successfully on 4 or 5 occasions during learning. Recall performance on the last learning trial and 30minutes after acquisition were indistinguishable between TEA patients and controls. Over the delay interval of 30minutes to one week, however, accelerated forgetting of this newly learned verbal material was observed in TEA patients. This severe forgetting is also reflected in the three-week recognition test, where TEA patients performed significantly worse than controls. Moreover, whereas recall on the last learning trial correlated significantly with the 30minute delayed recall in both groups, recall on the last learning trial correlated significantly with one week and three week delayed recall only in the controls. In both groups, the three-week free recall performance correlated with the three-week recognition test. Patients with TEA demonstrate ALF even for verbal material that is learned under precisely matched conditions. These results are consistent with the hypothesis that ALF represents a disruption of memory consolidation rather than an acquisition deficit.