Fig. 6: Replay sequences exhibit greater diffusivity in the awake state compared to the subsequent sleep state.

a Replay diffusivity can be changed by tuning adaptation strength in the CAN model. b Identifying SWRS during the sleep state. Top, aggregated LFP from CA1, CA2, and CA3 tetrodes for low/high-amplitude LFP (SIA/LIA) period detection; middle, theta to delta ratio for REM period detection; bottom, running speed for immobile period detection. Red bars mark candidate sleep periods lasting at least 90 s with extended (>5 s) continuous LIA periods (“Methods”). Only ripple events within these periods were analysed further. c Examples of an awake replay (left) when an animal performed the W-track spatial alternation task and a subsequent sleep replay (right) when the animal was in the resting box. d Replay step size distribution (time bin  = 2 ms) for awake (orange) and subsequent sleep (blue) states from two successive sessions for one animal. The inset panel illustrates step size versus time bins on a log-log scale for both awake and sleep replays, showing greater diffusivity in awake replay. e Comparison of diffusivity of awake replay (orange) and subsequent sleep replay (blue) across all recording sessions and animals. Each dot represents a replay diffusion exponent calculated from a recording session, with grey lines connecting values from successive running and sleep sessions. Awake replays show significantly higher diffusivity than sleep replays (Wilcoxon signed-rank test, P = 4.6 × 10⁻⁴).