Fractal dynamics characterize healthy human gait, with stride-interval fluctuations exhibiting long-range correlations generated by intrinsic locomotor control. These correlations diminish when gait is synchronized to an external metronome, often accompanied by curvature in detrended fluctuation analysis (DFA) scaling, but whether this reflects suppression of intrinsic structure or a scale-dependent reorganization of control remains unclear. Here, we identified a delayed-correction mechanism capable of reproducing the characteristic curvature of DFA scaling during metronome walking. We reanalyzed treadmill data from twelve healthy men (1.1 m/s, 20 min) under free walking (FW) and metronome walking (MW). Stride intervals were extracted from foot-switch measurements and evaluated using DFA. As expected, FW showed a near-linear DFA profile, whereas MW exhibited pronounced curvature, indicating reorganization of temporal structure across scales. To test the mechanism underlying this curvature, we constructed difference series from FW stride intervals and compared their DFA profiles with those of MW. Guided by simulation results predicting curvature emergence under delayed error correction, we systematically varied the differencing lag (j = 1-30). Both empirical and simulated analyses showed that MW-like curvature emerged when the current stride was corrected using information from approximately six earlier strides, with statistical analyses supporting an optimal range of 4-8 strides. These findings indicate that metronomic cueing does not simply eliminate fractal (long-range temporal) structure of gait but reorganizes gait dynamics through the superposition of two control processes: long-range fractal structure and a short-range, multi-stride delayed error correction loop.
Hirata et al. (2026) studied this question.