Predictive uncertainty in auditory sequence processing

Whereas previous research has focussed on the confirmation and violation of musical expectations, our paper is the first that systematically studies the degree of uncertainty with which we make predictions about, for instance, the next note in a melody. Predictive uncertainty varies between different contexts and is a prospective property of our expectations before a musical event occurs rather than a retrospective reaction to it. Amongst other things, we demonstrate that predictive uncertainty can be understood in the framework of statistical learning. This principle explains how the human brain automatically acquires knowledge about the surrounding world and subsequently bases its predictions on this experience. One of our key findings is that musically trained individuals can predict the continuation of melodic phrases with greater degrees of certainty than those without musical training. Nevertheless, musical experience is only advantageous in specific cases where the musical style allows for predictability. Thus, capitalising on the potential for making high-certainty predictions about melodies is a key step in obtaining proficiency within a musical style.

Read the article on-line ...