HPC-DMN integration at neural event boundary affects across-boundary BOLD representations and memory recollection

Abstract

Humans naturally divide their continuous experiences into discrete events, with event boundaries playing a critical role in this segmentation process. These boundaries are marked by significant shifts in brain activity, particularly in the hippocampal complex (HPC) and default mode network (DMN), which are key to learning and memory formation. We hypothesized that neural event boundaries in specific HPC-DMN regions contribute to episode recollection through across-boundary BOLD pattern shifts and neural event network formation, which may be influenced by two factors: network integration and boundary alignment. To test this, we used the widely recognized Sherlock fMRI dataset, which involves 22 human participants (17 subjects were used in this study). One key finding was that across-boundary BOLD pattern dissimilarity in the inferior parietal lobule (IPL) was greater for successfully recalled events compared to unsuccessful ones, specifically at boundaries with a high participation coefficient (indicative of high integration) or those aligned with the anterior superior temporal gyrus (aSTG) or the ventrolateral prefrontal cortex (vlPFC). Moreover, the entorhinal cortex played a crucial role in linking events into a network to facilitate subsequent recollection, particularly at boundaries with low participation coefficients or those aligned with the parahippocampus (PHC). These findings highlight the vital role of neural event boundaries in aiding comprehension and memory in naturalistic contexts through interactions with other brain regions.