Fig. 4: Block sensitivity for wait times requires structure learning.

a Wait time adaptation ratio (average wait time for 20 μL in high/low blocks; purple) evolved over training, while trial initiation time ratio (average in high/low blocks; green) was below 1 on first session. b Linear regression coefficients for block (maroon) and reward (orange) gradually evolved over training for wait time. c Linear regression coefficient for previous reward was relatively stable across training for trial initiation time. d Overshoot in trial initiation time (difference between maximum z-scored trial initiation time and trial initiation time at trial 20 post-transition) was more prominent in last 10 sessions (red) compared to first 10 sessions (maroon). e Overshoot in trial initiation time dynamics evolved on a similar timescale as block sensitivity for wait times, smoothed with a moving average of 10 points. All error bars are mean ± S.E.M. Source data are provided as a Source Data file.