What aspects of music grab our attention more? An investigation of the effects of musical structure and performance asynchrony on the perception of leader-follower relations and quality during selective attention to a piano duet

Aims When listening to music, we can either listen to multiple parts as a whole, or selectively attend to one part. But what elements in music grab our attention and make us shift from holistic to selective listening? In this study, by manipulating stimulus complexity, we explore the neural correlates and relative influence musical hierarchical structure and temporal variations have on the assessment of musical quality and a leader-follower relationship between parts of a piano duet. Methods In both a behavioral and follow-up functional MRI experiment, healthy musicians were tested using a selective attention task. Auditory stimuli were variations of an excerpt of a Sicilian hunting song with a clear hierarchical structure: a melodic primo and an accompanying secondo part. Stimuli were either notation-based, computer generated samples or pre-recorded performances. To parametrically modulate the temporal leader-follower relationship i) metronomic notation samples were compared with either ii) original recorded performance stimuli (with natural temporal variance-lag0) or iii) performance stimuli with an additional synthetic lag (lag+ or lag-). Individuals were cued to attend to either of the two voices (primo or secondo) by means of a fade in of the non-attended part. The attended part would then either be leading or following temporally (lag+ or lag-) and/or structurally. Following each stimulus, participants were required to subjectively assess the relative leader-follower relationship, quality and difficulty. Results and Conclusions Behavioral data suggest that both structural and temporal elements are important for establishing and assessing a leader-follower relationship between parts. However, a dominance of hierarchical structure was observed. Specifically, irrespective of which part was objectively temporally leading, when the primo part was attended to, it was rated as leading. Furthermore, when attending to primo the overall rating of quality was higher. Across manipulations of temporal lag, attending to secondo was rated as more difficult but less so when the part had been manipulated to be objectively temporally leading (lag+). No effect of attention or temporal manipulation on quality ratings when using notation stimuli was observed suggesting the importance of natural variance. We suggest that by increasing the complexity of the stimuli used, a more extensive network of brain areas is activated by our attention task than seen in previous related studies. This activation, which should be observed in primary auditory and fronto-parietal cortices, should vary as a function of perceived quality