The interaction of working memory performance and episodic memory formation in patients with Korsakoff's amnesia

Both neuroimaging work and studies investigating amnesic patients have shown involvement of the medial temporal lobe during working memory tasks, especially when multiple items or features have to be associated. However, so far no study has examined the relationship between working memory and subsequent episodic memory in patients using similar tasks. In this study, we compared patients with amnesia due to Korsakoff's syndrome (n = 19) with healthy controls (n = 18) on an associative working memory task followed by an unexpected subsequent episodic memory task. The computerized working memory task required participants to maintain two pairs of faces and houses for either short (3 s) or long (6 s) delays. Approximately 5 minutes after completion of the working memory task, an unexpected subsequent recognition task with a two-alternative forced choice paradigm was administered. By directly comparing working memory and subsequent episodic memory, we were able to examine long-term encoding processes that may take place after longer delays. As expected, patients performed at chance level on the episodic memory task. Interestingly, patients also showed significantly impaired working memory performance (p < .01), even at short delays. Longer delays did not result in better subsequent memory, indicating that they do not facilitate long-term encoding processes. Our results are discussed in relation to Baddeley's working memory model as the episodic buffer is assumed to be a short-term store for maintaining bound representations. In light of these results, the long-standing view that working memory and long-term memory are strictly dissociated may need to be revisited.

► Associative working memory and subsequent episodic memory formation were studied
► We examined 19 patients with Korsakoff's syndrome and 18 matched healthy controls
► Patients scored significantly lower on the associative working memory task
► Longer maintenance periods did not facilitate long-term encoding processes
► Working memory and long-term memory may not be as strictly dissociated as assumed