It is well known that human long-term memory is not a unitary system. Evidence from various approaches indicate that an explicit memory system that supports the ability to consciously remember a past experience is disrupted by damage to the medial temporal lobe. In contrast, repetition priming — the non-conscious facilitation of processing that results from implicit memory for previous stimulus presentation — is relatively intact following medial temporal lobe damage. Very little is known about the interactions between these systems. A recent paper in the Journal of Cognitive Neuroscience by Wagner, Maril and Schacter provides intriguing evidence of 'crosstalk' between these two systems.

The authors investigated the novel and somewhat counter-intuitive hypothesis that priming may hinder new episodic learning. This hypothesis is based on neuroimaging data demonstrating that greater activation of regions of inferior prefrontal cortex during episodic encoding is associated with superior long-term retrieval, and that priming results in reduced activation in the same brain regions during subsequent re-encoding of the (primed) stimulus.

On day one, subjects incidentally encoded a list of novel words by reading the words on a screen and making a judgement as to whether the word was concrete (chair) or abstract (love). The next day a shorter list of novel words was presented and incidentally encoded before the subjects were scanned using fMRI while they incidentally encoded a third set of words. The key variable during the scan was that some of the words were from the first list (re-encoding with long-lag between presentations), some were from the second list (re-encoding with short-lag between presentations) and some were novel. Varying the lag in this way facilitated a comparison between different levels of priming during re-encoding, and a subsequent explicit memory test. The latter was performed two days later in the form of an old/new explicit recognition memory test for all of the previously seen words.

Greater priming during re-encoding in the scanner was associated with lower levels of subsequent explicit memory. Moreover, those subjects who demonstrated the strongest priming effects during the re-encoding of the long-lag word list tended to demonstrate the least benefit during this re-encoding for subsequent explicit remembrance. These data indicate that priming for past experiences may disrupt new episodic encoding. The authors also provide an elegant possible explanation for these counter-intuitive findings that addresses the 'crosstalk' between memory systems. Further research that will shed new light on this neglected area is awaited with interest and will be facilitated by the decision to deposit these data in the newly formed National fMRI Data Center for public access (accession number 2-2000-11142).