Fig. 4: Relationships between cue and pull activities and session types.

a, b Early-pull rate (a), pull latency (left in b), success rate (middle in b), and lever trajectory correlation (right in b) in 14 H, 14 M, and 14 L sessions (from eight mice). The same color indicates sessions from the same mouse. Bars indicate mean ± SEM. ***P = 7.38 × 10–6 between H and M sessions, 7.38 × 10–6 between H and L sessions, and 7.47 × 10–6 between M and L sessions for early-pull rate, ***P = 1.50 × 10–4 between H and M sessions, 7.47 × 10–6 between H and L sessions, and 5.81 × 10–5 between M and L sessions for pull latency, ***P = 9.25 × 10–6 between H and L sessions and 5.80 × 10–5 between M and L sessions for early-pull rate, Wilcoxon rank sum test with Bonferroni correction (α = 0.0167), two-sided. c Proportions of cue-preferring and pull-preferring axons, and axons that were classified into both types. The mean ± SEM are also shown. In two-sided Wilcoxon rank sum tests with Bonferroni correction (α = 0.0167), no significant differences were detected for each type. n = 14 for H, M, and L sessions. d Cue, pull–0.3s, and pull activities. The mean ± SEM are also shown. Cue activity: n = 12, 12, and 10 for H, M, and L sessions, respectively. Pull–0.3s and pull activity: n = 13, 14, and 14 for H, M, and L sessions, respectively. In pull–0.3s activity, ++P = 0.00244 and +++P = 0.000244 in comparison with zero, Wilcoxon signed-rank tests with Bonferroni correction (α = 0.0167), two-sided. e Cue and pull–0.3s activities in trials grouped by pull latency (green, 700–1000 ms; blue, 1000–1300 ms; purple, 1300–1600 ms). The mean ± SEM are also shown. Wilcoxon signed-rank tests with Bonferroni correction (α = 0.0167, two-sided) and Wilcoxon rank sum tests with Bonferroni correction (α = 0.0167, two-sided) were used for comparisons between trials with different pull latencies and between different sessions, respectively. In cue activity, *P = 0.00507. When pull–0.3s activity was compared to zero (Wilcoxon signed-rank test with Bonferroni correction, α = 0.0167, two-sided), ++P = 0.00171 for 700–1000 ms and 0.00244 for 1000–1300 ms in H sessions, and ++P = 6.10 × 10–4 for 700–1000 ms, 3.66 × 10–4 for 1000–1300 ms, and 0.00305 for 1300–1600 ms in M sessions. Cue activity: n = 12, 12, and 10 for H, M, and L sessions, respectively. Pull–0.3s activity: n = 13, 14, and 14 for H, M, and L sessions, respectively. f Schematic of the time courses of combined activity of cue-preferring and pull-preferring axons in trials grouped by pull latency (green, 700–1000 ms; blue, 1000–1300 ms; purple, 1300–1600 ms). Line thickness represents trial frequency. Source data are provided as a Source Data file.