Westphal, A.J., Monti, M.M., Reggente, N., Yazdanshenas, O., & Rissman, J.
Andrew Westphal presented this work at an SfN Nanosymposium in Washington D.C (2014)
SfN 2014 Abstract:
The act of retrieving a memory for a specific episode of one’s past requires the coordination of brain networks involved in controlling access to mnemonic contents and representing and monitoring the stored information. This has been shown to invoke a brain connectivity profile that diverges somewhat from the brain’s intrinsic resting state organization (Fornito et al., 2012). However, it is not yet clear to what degree this “retrieval mode” brain state differs from that observed during other complex cognitive tasks. In order to examine this further, we performed a graph theoretical analysis on fMRI functional connectivity data patterns measured while participants (N = 20) alternated between the performance of episodic source memory retrieval, analogical reasoning, and visuospatial perception tasks. In order to avoid systematic confounds, we ensured that the tasks were matched for response demands, reaction times, and bottom-up visual processing. Following preprocessing, we extracted fMRI time-courses from each 40 sec task block and concatenated these across runs to generate task-specific time-courses. We next reduced our whole brain data set to 264 functional areas, identified by resting state parcellation and meta-analysis (Power et al., 2011) and defined as spherical regions of interest (5mm radius). Pairwise correlations were then computed between all pairs of nodes for each cognitive task, and the weakest connections were thresholded out at a range of sparsity values. To capture a key global property of brain network dynamics, we analyzed how much each task-set expressed a graph theoretic measure known as modularity (Newman, 2006), which assesses the amount of connectivity within identified networks versus between networks. Our data revealed that the memory retrieval task showed significantly reduced modularity in comparison to the reasoning and perception tasks, an effect that replicated across sparsity thresholds. This suggests that the memory task-set is characterized by more widespread connectivity across the brain. Strikingly, reduced modularity in individual subjects was diagnostic of fewer memory errors and improved source monitoring. Taken together, our results suggest that memory retrieval may benefit from lower modularity, presumably because otherwise competitive brain networks supporting externally-directed and internally-directed attention must work together to link environmental stimuli with an introspective mnemonic search process.