Fig. 2: Rapid expression of detour theta sequences is compatible with network pre-configuration.

a Example showing rapid expression of detour sequence. (Top and middle) The animal’s position during the first 3 laps with zoomed-in windows showing theta-scale decoding results. Vertical lines: theta peaks. (Bottom) Decoding results averaged over detour theta cycles from 2 run directions on the first 3 laps. b Distribution of detour segment theta cycle decoding quadrant ratio across first 3 laps (Wilcoxon signed rank test). c Behavioral and theta time-scale cross-correlograms (CCG) from two example cell-pairs during the lap 1 detour run. d Correlation of CCG bias between theta time-scale and behavioral time-scale (top) or bias of place map (i.e., distance between place fields) on detour (bottom) across first 3 laps. Dots: cell pairs. e Four hypothesized mechanisms contributing to the rapid development of theta sequence: Spike theta-phase locking (top left); Spike theta-phase precession (top right); Waking-rest replay compression (bottom left); Pre-configuration (bottom right). The first 3 mechanisms were either not occurring or could not fully explain the early-laps theta sequence (Supplementary Fig. 2); thus, we focused on pre-configuration. f Examples of detour preplay during pre-detour sleep. g Proportions of significant detour preplay measured by weighted correlation (left; Wilcoxon signed rank test) or a combined criteria of absolute weighted correlation and normalized maximum jump (right; Z-tests for two proportions) compared against time bin shuffles. On combined criteria, the black box indicates the region of significant preplay. h Proportion of significant detour preplay in sleep and run sessions before Run2 measured in 10 min time windows. Orange and blue lines: sleep (orange) and waking rest (blue) preplay proportions for each animal and direction. Black line and shaded area: mean and s.e.m. i Example frames detected as forward detour preplay of a detour segment with highlighted activities from two example cells. j Transition probability matrix estimated from significant forward detour preplay. Pixels indicate the conditional probability of one pyramidal cell firing (x-axis) right after another one (y-axis); one example pixel indicated by the small square and arrowheads, color-coded as corresponding cells in (i). k Example theta cycle depicting a detour theta sequence. l Probability of spikes in theta cycle from (k) computed based on detour preplay transition probability matrix from (j) compared against 500,000 shuffles with the same length to calculate percentile value. m Distribution of detour theta cycle (> 3 active cells) probability percentile values against shuffle. Significant (> 95% of shuffles) proportion of theta cycles was above chance level (Binomial test). n Forward preplay had steady prediction power in predicting theta cycles from the first, middle, and the last 2 laps, while prediction power from forward replay accumulated over experience. o Recruitment of tuplets with length of 2 or 3 cells from pre-detour sleep into early theta cycles was significantly higher than from shuffled sleep (Wilcoxon signed rank test). Bar plots in (g, o): mean ± s.e.m with each dot representing one animal, direction, and detour session (n = 16). ***P < 0.01, **P < 0.01, *P < 0.05, n.s. = not significant. Source data are provided as a Source Data file.