Tapping to Bach: Resonance-based modeling of pulse

We studied the process of pulse finding by using mechanical performances of composed music in which it is fairly challenging to find a pulse. Human participants synchronized to short musical excerpts that varied in the amount of musical information present, the amount of syncopation, and the metrical position of the first note. We quantified the period and phase of tapping, the amount of nonsynchronized tapping, the time to start tapping, the variability of the intertap interval, and the deviation of taps from the intended musical events. The amount of musical information and the amount of syncopation were predictors of pulse-finding ability (metrical appropriateness of tapping, time to start tapping, variability of tapping, and deviations from the beat), and the metrical position of the first note of excerpts biased participants to tap with the corresponding phase. In addition, the degree to which participants tapped in consensus correlated positively with pulse-finding ability. We modeled participants' behavior by using a resonance equation to calculate the strength of all metrically appropriate periodicities. The summed activation strength across the periodicities correlated with pulse-finding ability. Our results demonstrate that musical pulse finding is a useful behavioral paradigm for modeling the influences of stimulus features on complex sensorimotor synchronization.