Extended Data Fig. 3: Goal PCs encoded reliable distance but not time information after the running start and didn’t require reward per se.

a, Four example Goal PCs showing relatively consistent spatial but not temporal coding at varying running speeds. The heatmaps show Δf/f values for space and time across laps, sorted by the running speed from the running onset to the peak PF location. Group-averaged Δf/f traces in red, green, and blue traces represent fast, medium, and slow speeds, respectively. Red dashed line: Time after running onset to the PF peak. The across-lap PF peak timing against running velocity is also plotted. The black line: Two-sided linear fit. b, PF peak position in space during slow (26.1 ± 2.6 cm) versus fast (23.0 ± 2.3 cm) runs. p = 0.066. c, PF peak position in time during slow (1.3 ± 0.10 s) versus fast runs (0.91 ± 0.075 s). p = 1.2e–15; one-sample two-sided Kolmogorov–Smirnov test for panels b–c. For panels a–c, only Goal PCs with PFs within 90 cm after the running onset were included (n = 90 Goal PCs from 6 mice, Pre, familiar). d, Top: Mean running velocity around the start of rewarded runs. Middle: Peak-normalized mean activity (Δf/f) in time around the start of rewarded runs for Goal PCs (Familiar, Day 1, Pre) sorted by the peak timing. Goal PCs with PFs located more than 90 cm after the reward were excluded. n = 86 Goal PCs. Bottom: Population mean Δf/f in time. e, Same as d, but for non-rewarded runs. PCs are sorted by peak timing during rewarded runs. f, Comparison of the activity peak timing between rewarded and non-rewarded runs. Each dot indicates one neuron. Orange line: Two-sided linear fit. g, PC field similarity between rewarded and non-rewarded runs (0.56 ± 0.043; p = 8.9e–22, paired two-sided t-test). Each black dot represents one neuron. Horizontal dashed line marks 0. For panels d–g: n = 86 neurons from 5 mice. One mouse did not exhibit non-rewarded runs, resulting in 43 non-rewarded running episodes from 5 mice. Data: Mean ± SEM.

Source data