The ensemble that plays together, stays together

It has long been known that the medial temporal lobe is crucial for the formation and retrieval of episodic memories. This region includes the hip- pocampal formation (the hippocampus proper and subiculum) and the un- derlying entorhinal, perirhinal, and parahippocampal cortices, all of which receive a unique complement of cortical inputs. How best to characterize the precise roles of these different brain regions and their interactions is of central concern in understanding the medial temporal lobe contributions to episodic memory. Using functional magnetic resonance imaging (fMRI) to assess brain activity while subjects performed an episodic retrieval task, Giovanello et al. (2004) have contributed two important pieces to this puzzle. First, their findings show that the hippocampal formation is differentially recruited when subjects are asked to make memory decisions about the conjunctive relationship (in this case, the co-occurrence) of two stimuli compared to decisions about each stimulus independently. Second, their data demon- strate that the extent to which the hippocampus is activated also depends critically on whether the two stimuli were actually previously paired. In this study, subjects encoded a series of word pairs by forming sentences that contained the two words. At retrieval, subjects were asked to engage in two different tasks: an "Item" task that asked subjects to report whether the two test words had been previously presented and an "Associative" task that asked subjects to report whether the two items had been previously pre- sented together. Giovanello and colleagues collected fMRI data during this retrieval phase, demonstrating that the hippocampus shows greater BOLD signal increase when subjects performed the Associative task compared with the Item task. Furthermore, greater BOLD increase within the Associative task was seen to "intact" word pairs (i.e., test words were paired during the study phase) compared to both "rearranged" (i.e., both test words were presented during the study phase but paired with a different word) and to novel word pairs. The implications of these findings are twofold. First, they add to a grow- ing literature demonstrating the involvement of the hippocampus in con- junctive processing (i.e., making decisions about conjunctions between stimuli) above and beyond item processing (i.e., making decisions about individual stimuli). More specifically, these findings underscore the hypoth- esis that the role of the hippocampus in episodic memory is in the creation of an ensemble of activity that is a filtered representation of all activity in the cortical and subcortical mantle during the moment to moment processing of events. In this way, each singular hippocampal represen- tation might, in essence, act as a mirror reflection of all other activity in the brain and it is through this process that multiple elements of an episode might get bound into a single, conjunctive representation (Norman and O'Reilly, 2003). Second, these results demonstrate not only that the hippocampus is preferentially engaged dur- ing conjunctive processing, but also that activity in the hippocampus signals when previously associated stimuli appear together again or are "intact." Taken together, these findings suggest that hippocampal processes can be