Fig. 4: Event-specific intersubject pattern correlation.

a Intersubject pattern correlation (pISC) was computed for each movie event by correlating the event-specific activation pattern (averaged across times within the event) of a participant and that of each of the other participants. b Posterior medial cortex (PMC; orange) and early visual cortex (EVC; green) regions-of-interest visualized on the inflated surface of a template brain (medial view). c pISC in PMC and EVC during movie watching (left) and recall (right). Black diamonds show the mean pISC averaged across all participants and movie events. Orange and green histograms show the null distributions of the mean pISC in PMC and EVC, respectively. Statistical significance reflects difference from zero based on one-tailed randomization tests (all ps = .000999). d Whole-brain surface map of mean pISC during recall. pISC was computed for each of 400 parcels in a cortical atlas³³. The pISC map was arbitrarily thresholded at r = .015 for visualization purposes. pISC values in all visualized parcels were significantly greater than zero based on randomization tests (FDR-corrected q < .05 across parcels). e, f pISC for High vs. Low semantic centrality events during recall and the difference (Diff) between the two conditions in PMC (e) and EVC (f). For High and Low semantic centrality conditions, white circles represent individual participants (N = 15). Black diamonds represent the mean across participants within each condition. Error bars show SEM across participants. For the difference between High and Low conditions (Diff), black diamonds show the true participant average, and histograms show the null distribution of the mean difference. Statistical significance reflects difference from zero based on two-tailed randomization tests (p = .037 and .012 in PMC and EVC, respectively). *p < .05, ***p < .001. Source data for c, e, and f are provided as a Source Data file.