A Two-Process Model for Control of Legato Articulatoin Across a Wide Range of Tempos During Piano Performance

Abstract

Prior reports indicated a non-linear increase in key overlap times (KOTs) as tempo slows for scales/arpeggios performed at internote intervals (INIs) of I00-1000 ms. Simulations illustrate that this function can be explained by a two-process model. An oscillating neural network based on dynamics of the vector-integration-to-endpoint model for central generation of voluntary actions, allows performers to compute an estimate of the time remaining before the oscillator's next cycle onset. At fixed successive threshold values of this estimate they first launch keystroke n+l and then lift keystroke n. As tempo slows, time required to pass between threshold crossings elongates, and KOT increases. If only this process prevailed, performers would produce longer than observed KOTs at the slowest tempo. The full data set is explicable if subjects lift keystroke n whenever they cross the second threshold or receive sensory feedback from stroke n+l, whichever comes earlier.