Hippocampal signals for strong memory when associative memory is available and when it is not.

The paired-associate task has been used with functional magnetic resonance imaging (fMRI) in studies that assessed the role of the medial temporal lobe (MTL) subserving recollection and familiarity.Some researchers have interpreted their results to mean that the hippocampus selectively subserves recollection and not familiarity[cf., Eichenbaum et al., (2007) Annu Rev Neurosci 30:123–152]. Yet many of these results confound recollection and familiarity with strong and weak memories, and it is not clear whether the conclusions represent differences between memory processes or memory strength. In the current study, participants were scanned with fMRI during retrieval in a paired-associate task, and a new approach separated the analysis of memory strength from the analysis of memory processes. The data were sorted by confidence level in an old/new task, and the high-confidence responses were compared in categories when associative memory was highly accurate and when it was not available. The results show that high-confidence memory produced increased activity in the hippocampus,relative to the level for forgotten pairs, both when associative memory was available and when it was not. Two interpretations are discussed for the behavioral results for when associative memory was not available: one account based on familiarity and the other account based on noncriterial recollection. The conclusion is that recognition of the word-pairs was based on familiarity when associative memory was not available. Together with the fMRI results that activity in two regions associated with cognitive control (left ventrolateral prefrontal cortex and left inferior parietal lobule) was greater when responses were based on associative memory than when based on familiarity, the findings suggest that the hippocampus supports strong memory and that cortical regions make an additional contribution to recollection.