Analysis of timing variability in human movements by aligning parameter curves in time.

The analysis of timing in human movements requires a reference with which timing can be quantified. In reactive movements this reference is given by the stimulus. However, many movements do not respond to such an external event. In throwing, for instance, the hand opening for release has to be timed to an acceleration of the throwing arm. A common approach to analyzing release-timing variability is to choose a landmark in the movement that is supposed to have a fixed temporal relation to the release. Such distinct landmarks, however, are not always well definable. Therefore, the present article describes an alternative approach analyzing timing variability on the basis of the alignment of different trials relative to their kinematic shape, by shifting the trials in the time domain. The basic assumption behind this approach is that single throwing movements are one instance of an acquired movement template, and thus show a considerable similarity. In contrast, the location of the temporal moment of release varies from trial to trial, generating imprecision regarding the release timing. In trials synchronized with respect to the release, this variability can be assessed by shifting the kinematic profiles of the throwing movements in time such that they superimpose as closely as possible. As a result, the corresponding time shifts for all trials represent a measure of the release time deviations across trials, and the standard deviation of these deviations represents the timing variability. Aside from timing analyses in such movements as throwing, the approach can be applied to very different tasks with timing demands-for example, to neurophysiological signals.