Predicting memory from the network structure of naturalistic events

Abstract

Human life consists of a multitude of diverse and interconnected events. However, extant research has focused on how humans segment and remember discrete events from continuous input, with far less attention given to how the structure of connections between events impacts memory. We conducted an fMRI study in which subjects watched and recalled a series of realistic audiovisual narratives. By transforming narratives into networks of events, we found that more central events—those with stronger semantic or causal connections to other events—were better remembered. During encoding, central events evoked larger hippocampal event boundary responses associated with memory consolidation. During recall, high centrality predicted stronger activation in cortical areas involved in episodic recollection, and more similar neural representations across individuals. Together, these results suggest that when humans encode and retrieve complex real-world experiences, the reliability and accessibility of memory representations is shaped by their location within a network of events.

A generalized cortical activity pattern at internally-generated mental context boundaries during unguided narrative recall

“…Here, we provide a selective overview of procedures and analysis methods. More detailed descriptions of participants, stimuli, experimental procedures, fMRI data acquisition and preprocessing can be found in Lee & Chen (2021).…”

“…To test whether the recall activation patterns evoked by between-movie boundaries were similar to encoding activation patterns evoked by event boundaries within a movie (Figure 4, 'Event offset' condition), we identified the strongest event boundaries within each movie. We utilized the fine-grained event boundaries defined in our previous study (Lee & Chen, 2021) which divided the ten movie stimuli into 202 events excluding title scenes (mean duration = 13.5 seconds, ranged 2 -42 seconds). We had four independent coders watch the movie stimuli and then choose which of the fine-grained event boundaries were the most important.…”

A generalized cortical activity pattern at internally generated mental context boundaries during unguided narrative recall

“…One important factor might be the presence or absence of inter-event connections. Even the most salient within-movie boundaries still demand some integration of information across events as the events are semantically or causally related, and ultimately constitute a single coherent narrative ( Lee and Chen, 2021 ; Song et al, 2021b ). In contrast, an entire cluster of related events, or the narrative as a whole, might be completely ‘flushed’ at between-movie boundaries; this difference could induce distinct cognitive states at the two levels of boundaries, giving rise to different PMC patterns.…”

“…Here, we provide a selective overview of procedures and analysis methods. More detailed descriptions of participants, stimuli, experimental procedures, fMRI data acquisition, and preprocessing can be found in Lee and Chen, 2021 .…”

“…For whole-brain pattern similarity analysis, we used an atlas ( Schaefer et al, 2018 ) that divided the cortical surface into 400 parcels (200 parcels per hemisphere) based on functional connectivity patterns (17 networks version). For ROI analyses, we defined the bilateral PMC by combining the parcels corresponding to the precuneus and posterior cingulate cortex within Default Network A as in our prior study ( Lee and Chen, 2021 ). The precuneus and posterior cingulate cortex together spanned the area that showed the strongest content-and task-general boundary patterns in the whole-brain analysis ( Figure 3C ).…”