Meeting Abstract
Many animal locomotion behaviors consist of repeating stereotyped body motions in a rhythmic fashion. When these rhythmic motions are recovered after the body encounters a disturbance, one may consider the characteristic motion to be a limit cycle of a stable nonlinear oscillator. We show, under the assumption that our data set consists of a collection of N trials each containing M cycles, that partitioning the data into cycles based on a distinguished event such as heel-strike (often used for human motion studies) and averaging the cycles produces statistically inferior model of typical motions to averaging based on an estimate of dynamical phase. The improved accuracy of the phase based model can enable effects to be detected that would otherwise require many more trials. Examples from several locomotion experimental datasets will be provided
