Content-specificity of the neural correlates of recollection

It is widely assumed that episodic retrieval (recollection) involves reinstatement of cortical activity engaged during the processing of an episode when it was initially experienced. It follows from this assumption that the cortical correlates of recollection should differ with the content of what is recollected, and that retrieval of different content should be associated with activity in functionally distinct cortical regions. The present experiment investigated these predictions. Subjects (N = 17) studied a mixed list of words and pictures and were then presented with a test list comprised of words only. Test items were studied words, the names of studied pictures, and unstudied (new) words. Functional magnetic resonance images were acquired while the subjects made Remember/Know/New judgments to these test words. Independent of study material, studied items endorsed as Remembered elicited greater activity than correctly classified unstudied items in several regions, including left frontal, left lateral parietal, and posterior cingulate cortex. In addition, Remembered items elicited greater activity in the right hippocampus and parahippocampal gyrus than items accorded Know judgments, replicating previous findings. Analysis of content-specific effects demonstrated a regional double-dissociation within left fusiform cortex; recollected words elicited greater activity than recollected pictures in lateral fusiform, whereas the reverse effect was evident in an anterior fusiform region. The lateral and anterior fusiform areas correspond closely to areas held to be functionally specialized for the processing of visual words and pictures, respectively. Thus, the current findings support the cortical reinstatement hypothesis of episodic retrieval.