Hippocampus-Seeded Functional Connectivity during Mnemonic Discrimination

Pattern separation is a neurocomputational mechanism thought to orthogonalize similar inputs into distinct, non-overlapping representations (Yassa & Stark, 2011). This process is also known as mnemonic discrimination (MD) on a behavioral level, and it is commonly assessed by the Mnemonic Similarity Task (MST). Previous studies have shown that activation of the hippocampus during the MST contributes to pattern separation. However, few studies have examined the role of hippocampus-seeded cortical connectivity in mnemonic discrimination, which may provide new insights at a brain network level. Inspired by the cortico-hippocampal pattern separation (CHiPS) framework proposed by Amer and Davachi (2023), I hypothesized that functional connectivity between the hippocampal head and other brain regions, such as the medial temporal lobe and the visual cortex, would be significantly different between lure correct rejection (LCR) and lure false alarm (LFA) conditions during successful mnemonic discrimination. To test this hypothesis, a publicly available fMRI dataset (DOI: 10.1811/openneuro.ds003711.v1.0.0) from OpenNeuro was used, and a seed-to-voxel analysis was conducted using the generalized psychophysiological interaction (gPPI) in CONN toolbox of MATLAB. Significant ipsilateral hippocampal head - amygdala functional connectivity was found in both LCR and LFA conditions. However, contrary to my hypothesis, while the left hippocampal head showed higher connectivity with left insular cortex and left posterior superior cortex for the LCR-LFA contrast at cluster level uncorrected p < .05, this difference was not found at an FDR-corrected threshold. An exploratory ROI-to-ROI analysis with 22 ROIs selected from the medial temporal lobe (MTL) and the visual cortex was conducted. When setting the cluster threshold to the ROI-level uncorrected p < .05, several connections involving the right anterior middle temporal gyrus (aMTG) were found, but these connections did not survive at a rigorous cluster FDR-corrected p-value.