Multi-Voiced Music Bypasses Attentional Limitations in the Brain

Abstract

Attentional limits make it difficult to comprehend concurrent speech streams. However, multiple musical streams are processed comparatively easily. Coherence may be a key difference between music and stimuli like speech, which does not rely on the integration of multiple streams for comprehension. The musical organization between melodies in a composition may provide a cognitive scaffold to overcome attentional limitations when perceiving multiple lines of music concurrently. We investigated how listeners attend to multi-voiced music, examining biological indices associated with processing structured versus unstructured music. We predicted that musical structure provides coherence across distinct musical lines, allowing listeners to attend to simultaneous melodies, and that a lack of organization causes simultaneous melodies to be heard as separate streams. Musician participants attended to melodies in a Coherent music condition featuring flute duets and a Jumbled condition where those duets were manipulated to eliminate coherence between the parts. Auditory-evoked cortical potentials were collected to a tone probe. Analysis focused on the N100 response which is primarily generated within the auditory cortex and is larger for attended versus ignored stimuli. Results suggest that participants did not attend to one line over the other when listening to Coherent music, instead perceptually integrating the streams. Yet, for the Jumbled music, effects indicate that participants attended to one line while ignoring the other, abandoning their integration. Our findings lend support for the theory that musical organization aids attention when perceiving multi-voiced music.

Keywords: N100 response; attention; auditory scene analysis; counterpoint; electroencephalography; multivoiced music; polyphony.

FIGURE 1

Sample portion of Musical Excerpts. (A) shows sample of Coherent excerpt. Music in Coherent excerpts were presented as written by the composers. (B) shows sample of corresponding Jumbled excerpt. The score was scrambled and pulled apart by taking large portions of one line (see highlighted box for example) and inserting it in a different location. Additionally, the tempo (i.e., speed) of bottom line was slower than tempo of top line. Coherence between the two lines has been destroyed, although each line sounded relatively intact when played alone because chunks were excerpted at natural stopping points (ex. cadences or other structural pauses). Excerpt sounds as if the two melodies are playing completely independently and separately.

FIGURE 2

EEG waveforms. (A) shows the grand average waveforms highlighting the N100 component (box) for the four regions of interest [prefrontal, central, left, and right regions, see (B)]. (C) depicts the topographic maps of the N100 component for the four conditions as well as the difference between attend and ignore conditions, generated by subtracting the topographic map of the attend from the ignore conditions. Topographic maps zoom in at 115 ms where the greatest negativity occurs in the grand average waveforms.

FIGURE 3

N100 amplitudes for Attend and Ignore responses in the Jumbled Condition (A) and Coherent Condition (B). Bar graphs comparing N100 group-mean amplitudes in both music conditions (mean ± S.E.) for Prefrontal and Central scalp regions of interest. An effect of attention on the N100 response magnitude (i.e., N100 attention effect) is seen in the Jumbled music condition (cacophonous compositions, A) but not the Coherent music condition (normal music compositions, B). *p < 0.05.