Event-related potentials to changes of rhythmic unit: differences between musicians and nonmusicians.

We investigated whether the expected differences between musicians and nonmusicians in their ability to detect a rhythm change were reflected in their event-related potentials (ERPs) and, if reflected, how these ERP differences associated with behavioral indices. Stimuli were three consecutive and equally spaced drum beats followed by a rest period to form a rhythmic unit (RU). By using three different inter-beat periods, three RUs were produced. Combinations of these RUs served as the "target/standard" pairs of an oddball sequence. In four different experiments, we tried two RU-change types each with two levels of detection difficulty. ERPs were recorded from the F3, Fz, F4, Cz and Pz scalp sites of 12 musicians and 12 nonmusicians. RT, hit and false-alarm rates were also measured. The data have shown with high statistical confidence that, associated with the musicians' better detection performance and shorter RTs, their ERP P3 to rhythm changes peaked significantly earlier and was significantly larger compared to nonmusicians. Intergroup ERP differences allowed above 90% correct classification. This study has also showed that not only violations of relatively complex musical regularities, but very simple rhythmic unit alterations could lead to significant P3 differences between musicians and nonmusicians. The high accuracy of the musician/nonmusician classification based only on their P3 data strongly supported the hypothesis that sensory and/or cognitive advantage of musicians in detecting rhythm changes does reflect in their P3.