Fig. 4

A Illustration of the touchscreen apparatus used for paired associates learning (PAL) training and testing. B Timeline depicting the sequence of experiments in PAL. C Number of days to criterion for runners (Run) and sedentary controls (Sed). Data are shown as individual data points with mean (horizontal lines) and # Days to Criterion scale on the left Y-axis. Effect size (Cohen’s d) is shown as a black circle with vertical lines and gaussian distribution for the bootstrap 95% CI and Cohen’s d scale on the right y-axis. Both groups reached the initial criterion in PAL at the same rate before elevation of neurogenesis. D Mean (± SEM) % Accuracy during the final 4 days of PAL training before elevation of neurogenesis. Both groups performed with the same accuracy before elevation of neurogenesis. E % Accuracy in PAL training before (Pre) and after (Post) elevation of neurogenesis (% Accuracy is computed as an average across 4 days of testing for each swarm plot). Upper graph shows individual data points in swarm plots with means (± SD; gapped vertical line). Lower graph shows effect size (Cohen’s d) as a black circle with vertical lines and gaussian distributions for the bootstrap 95% CI. Both groups performed with the same % Accuracy before elevation of neurogenesis, but runners performed with significantly reduced accuracy after elevation of neurogenesis, indicating that they had forgotten the correct image-location pairings. F Mean (± SEM) percent accuracy in PAL following either running or sedentary control. Runners performed with significantly and persistently reduced accuracy relative to sedentary controls over 4 days of testing. G Mean (± SEM) % Accuracy in PAL during reversal learning. Both groups performed with the same accuracy during reversal learning, indicating that running did not enhance accuracy during PAL reversal learning