Fig. 1: Recording protocol and behavioral performance of mice on the AutoPI task.
a, Top, Recording protocol. During each recording session, mice performed a random foraging task before and after the AutoPI task. On the AutoPI task, mice alternated between trials with and without visible light (light and dark trials). In the last recording session, mice performed an additional random foraging trial in which the lever box was placed on the arena. Random foraging with the lever box placed on the arena consists of two recordings of 15 min each, and the lever is placed at a different location after the first 15-min recording. The red lines on the arena and floor of the home base are examples of the path of a mouse during a recording session. Bottom, The four phases of a trial on the AutoPI task. The mouse leaves the home base and searches for the lever box on the arena. Once at the lever box, the lever is pressed, triggering food reward delivery in the food magazine of the home base. The mouse returns to the home base (homing) to access the food reward. b, Search path length during dark and light trials (dark–light difference: N = 17 mice, two-sided Wilcoxon signed-rank test, statistic = 0.0, P = 1.53 × 10−5). c, Homing error at the periphery during dark and light trials. For definitions, see Extended Data Fig. 2a (N = 17 mice, two-sided Wilcoxon signed-rank test, statistic = 0.0, P = 1.53 × 10−5). d, Homing error as a function of search path length during dark trials (N = 17 mice, two-sided Friedman test, statistic = 45.6, P = 2.98 × 10−9). e, Pearson correlation coefficients between search path length and homing error during light and dark trials (N = 17 mice, difference between light and dark trials, two-sided Wilcoxon signed-rank test, statistic = 7.0, P = 2.90 × 10−4). ****P < 0.0001.
