Isochronous versus non-isochronous pacing of gait: continuous coupling versus discrete error correction

Recent findings revealed that stride time series exhibit persistent fluctuations when a fractal metronome is used to pace walking. How can this phenomenon be modeled? So-called super central pattern generators (SCPGs) have been shown to reproduce persistent fluctuations of stride time series in self-paced walking and anti-persistent fluctuations of stride times as well as persistent fluctuations of asynchronies in isochronously paced walking. In the present study we investigated whether the SCPG approach is also applicable to non-isochronously paced walking. Using spectral analysis and auto-correlation functions, we found that experimental asynchrony series were qualitatively different between isochronously and non-isochronously paced walking, and that SCPGs provided no adequate account for the latter type of paced walking. As an alternative we proposed a model for stride time production with a correction mechanism based on the three preceding asynchronies with respect to the metronome. Spectral analysis, auto-correlation functions, and detrended cross-correlation analysis revealed strong similarities between experimental and simulated stride time series. This suggests that synchronization with non-isochronous metronomes is achieved on the basis of event-based timing, in contrast to synchronization with an isochronous metronome, which appears to be mediated by a continuous coupling between the limb dynamics and the rhythm of the metronome, i.e. as an instance of emergent timing. Our results further suggest that the presence of long-range correlations in inter-onset intervals of non-isochronous metronomes may enhance gait synchronization.