Functional magnetic resonance imaging (fMRI) was used to compare directly e
pisodic encoding and retrieval. During encoding, subjects studied visually
presented words and reported via keypress whether each word represented a p
leasant or unpleasant concept (intentional, deep encoding). During the retr
ieval phase, subjects indicated (via keypress) whether visually presented w
ords had previously been studied. No reliable differences were found during
the recognition phase for words that had been previously studied and those
that had not been studied. Areas preferentially active during encoding (re
lative to retrieval) included left superior frontal cortex, medial frontal
cortex, left superior temporal cortex, posterior cingulate, left parahippoc
ampal gyrus, and left inferior frontal gyrus. Regions more active in retrie
val than encoding included bilateral inferior parietal cortex, bilateral pr
ecuneus, right frontal polar cortex, right dorsolateral prefrontal cortex,
and right inferior frontal/insular cortex.