Bidirectional connectivity changes in the frontotemporal areas reflect figure-ground reversal in multivoiced music

Abstract

When listening to multivoiced music, it is possible to perceive distinct voices in differing pitch ranges independently. The melody that one focuses on becomes the figure, and other voices form the background. Shifting focus from one voice to another causes a figure-ground reversal, which may easily occur as a result of anticipation within repeated passages in a musical structure. We previously found that frontotemporal connectivity reflects changes in the melody of "Twinkle, Twinkle, Little Star" (TTLS) in Mozart's 12 Variations K. 265. We hypothesized that if frontotemporal connectivity remains unchanged in repeated passages, the melody of TTLS remains the figure; however, if connectivity changes, the melody shifts toward the background. Our findings show that frontotemporal connectivity only changed during the final, fourth repetition. This was accompanied by a bidirectional correlation in frontotemporal regions, implying both top-down and bottom-up processes. We captured a momentary figure-ground reversal in the multivoiced texture of continuously changing and developing music.