Figure 5.

Relevant past events are preferentially reactivated at event boundaries. We measured the level to which reactivation is modulated by semantic context (as defined by a Bag Of Words model, Figure 1d) across the entire brain. Rows show results from the Sherlock dataset (n=17 participants) (a), the 21^st year dataset (n=25 participants) (b), and a mega-analysis pooling together the two datasets (n=42) (c). Columns depict results obtained using different analysis methods (within/between-participant analysis). All maps were created using one-sided t-tests, and were cluster-corrected for multiple comparisons across the entire brain (p<0.005), except for the within-participant analysis of the 21^st year dataset (b, 1^st column), and the within-participant mega-analysis (c, 1^st column). Maps are projected onto a flat cortical surface. Across datasets and analysis methods, the precuneus/retrosplenial cortex showed a consistent positive modulation of reactivation by semantic context (scenes with semantic context similar to that of each event boundary were reactivated more than scenes with different semantic contexts). To confirm that the same regions that showed a reactivation effect also showed a significant modulation of reactivation by semantic context, we tested independent precuneus/retrosplenial cortex ROIs (depicted in black contours), separately defined for each dataset (Sherlock/21^st year) and analysis method (within/between participant) in the reactivation analysis (Figure 2a, Figure 3a). We calculated the participant-specific reactivation effect sizes in these ROIs as the Time (pats-future) main effect / Time X Congruency (same/different story) interaction in the Sherlock/21^st year datasets, respectively (see Online Methods). Single-participant effect sizes are represented as circles and group means are represented as bars (Sherlock within/between participant analysis, p<0.001; 21^st year within/between participant analysis, p=0.02/0.039, respectively; mega analyses, p<0.001, one-sided permutation tests). LH, left hemisphere; RH, right hemisphere; * p<0.05; *** p<0.001.