Mathematical modelling of interactive tapping reveals tap intervals as strongest predictor of subsequent tap timing

Source abstract

Successful interpersonal rhythmic coordination requires people to continuously adapt both thetiming and tempo of their actions to one another. Classical error-correction models distinguish theseprocesses as phase and period correction, typically modeling them as responses to instantaneous andcumulative asynchrony, respectively. These models were developed mostly for sensorimotor synchro-nisation to an external pacing signal, rather than interactive interpersonal coordination where bothpartners adapt. It thus remains unknown which temporal information people actually use to adapttheir behaviour to one another. We show, using a bounded generalised least-squares framework andjoint finger-tapping data obtained from both bidirectional (reciprocal) and unidirectional (leader-follower) interactions, that the previous inter-tap interval (ITI) of a partner is a better predictor foran individual’s next ITI than either cumulative or instantaneous asynchrony. We argue that thispredictor is not only the strongest from a modelling perspective, but also the most parsimoniousfrom a theoretical point of view. Our results support these claims and show that the partner’sprevious ITI provides a less endogenous predictor of adaptation, addressing a long-standing issue inmodels of tapping behaviour.