Extended Data Fig. 5: Simple models do not explain the relationship between replay content and behaviour.

a, Most common flight hypothesis. Left: schematic illustrating the hypothesis that replays are more likely for the most frequently taken flight paths (e.g., blue path). Middle: fraction of flights belonging to the 1st, 2nd and 3rd most common trajectories, ranked by frequency within each session. Right: corresponding fraction of replays for each trajectory rank. Box plots show median (centre line), 25th–75th percentiles (box) and minimum and maximum (whiskers) across sessions (n = 9 sessions from 3 bats in which at least 3 different trajectory types were each replayed at least 15 times; Wilcoxon signed-rank test between 1st and 3rd ranks). Note that replay fraction does not scale with flight frequency. b, Top: schematic illustrating the hypothesis that replays reflect the immediately preceding (orange) or following (purple) flight. Bottom: fraction of replays classified as immediate previous (left), immediate next (centre), or other (right), compared with chance level (n = 17 sessions from 5 bats; Wilcoxon signed-rank test, Methods). Chance level for each session is calculated as one over the number of different flight trajectories in that session that could be replayed. Thin lines depict single sessions; thick lines show the averages. c, Rewarded flight hypothesis. Left: schematic illustrating the hypothesis that replays prioritize flights to rewarded locations (feeders). Right: fraction of replays corresponding to flights ending at feeders (black) or perches (blue), for all, forward, and reverse replays. Replay fractions are normalized by the number of feeder- or perch-bound flight types present in each session, so values do not sum to 1. Box plots as in a (n = 23 sessions from 6 bats; Wilcoxon signed-rank test). No significant difference was observed between replays for reward- and perch-directed flights. P values are two-sided. ***P < 0.001; **P < 0.01; *P < 0.05; ns: P > 0.05. Illustration in c adapted from ref. ¹⁴, Springer Nature Limited, under a Creative Commons licence CC BY 4.0.