Fig. 4: Population prevalence of performance oscillations over the stride cycle.

a Subset of participants displaying significant oscillations in accuracy performance between 1.5 and 2.5 cycles per stride (cps). Each line is the Fourier fit strength to participant-level data between 0.2 and 10 cps (in steps of 0.2). The strength of the participant-level Fourier fit was compared to the 95% Confidence Interval of permuted data per participant (see Methods). For accuracy, 12/36 participants demonstrated significant oscillations at approximately 2 cps. b, c Display the same result for reaction time and response likelihood oscillations at 2 cps, respectively. d–f Independent subsets of participants with significant oscillations between 3.5 and 4.5 cps. g–i Remaining participants with significant oscillations either above 5 cps (n = 5, 6 and 7 out of 36 for accuracy, reaction time and responses, respectively), shown in red. Overlaid in dark grey are the fits for participants with no significant oscillation at any frequency (n = 6, 3, 3 for accuracy, reaction time and responses, respectively). j–l Bayesian estimates of population prevalence for oscillations at 2 cps (purple), 4 cps (pink), or any frequency in the range of 0.2–10 cps (black). Circular markers display the maximum a posteriori estimate of prevalence (MAP), and error bars display the 95% highest posterior density interval (HPDI) (see Methods).