Pattern separation involves regions beyond the hippocampus in non-demented elderly individuals: A 7T object lure task fMRI study.

Investigating the mechanism of differentiating similar representations, known as pattern separation, has primarily focused on the hippocampus. The roles of cortical regions and their interaction with the hippocampus remain largely unclear. In this study, we address this topic by analyzing whole-brain, high-resolution mnemonic similarity task (MST) fMRI data collected with a 7-Tesla MR scanner. Structural and functional MRI data were acquired from 55 non-demented elderly subjects. During the encoding phase of the MST task, participants responded with "indoor" or "outdoor" to 66 everyday objects. In the recognition phase, participants were asked to make "same" / "similar" / "new" judgments about objects that were either the same as previously seen objects (targets), similar but different from previously seen objects (lures), or new objects (foils). A general linear model was conducted on hippocampal regions of interest (ROIs) and at the whole-brain level with five conditions, including "new" response to foils (foil), "same" response to targets (hit), correct "similar" response to lure (lure correct rejection, LureCR), false "same" response to lure (lure false alarm, LureFA), and one condition for all others. The activity difference when lures were identified as "similar" compared to "same" (LureCR vs. LureFA) was used to assess if a region is involved in pattern separation. An association analysis was conducted to test if lure discrimination performance was correlated with activity difference of hippocampal ROIs between LureCR and LureFA, as well as age. Task-based functional connectivity between hippocampal ROIs and other regions involved in pattern separation was examined. In the hippocampal ROI analysis, left anterior CA3/DG showed greater activity in LureCR condition compared to LureFA. All other hippocampal ROIs did not show differential activity. Better lure discrimination performance was associated with larger activity difference between LureCR and LureFA at left anterior CA3/DG and right anterior subiculum. In the whole-brain analyses, regions in the frontoparietal network (FPN) consistently showed increased activity in the Hit, LureCR, and LureFA conditions, and the activity was right-lateralized for Hit and LureFA conditions but bilateral for LureCR condition. Eleven clusters, mainly located in the left hemisphere, were identified to show significant activity difference between LureCR and LureFA condition, including left FPN, middle temporal lobe, and subcortical regions. In summary, with the whole-brain high-resolution MST fMRI data, regions exhibiting the pattern separation signature were found to be lateralized to the left hemisphere in elderly participants. The left and right FPN are suggested to have distinct functional roles in the MST. The right FPN contributes to retrieving previously viewed same or similar objects, while the left FPN is preferentially involved in pattern separation. Furthermore, the pattern separation process might require the coordinated effort of FPN and hippocampus, with their interaction potentially mediated by subcortical regions.