Harmonicity: Behavioral and Neural Evidence for Functionality in Auditory Scene Analysis

ABSTRACT

This study examines sound perception and tests the functionality of harmonicity as a metric of auditory preference and regional brain activity. Most salient sounds contain harmonic overtones, a reliable acoustic natural regularity, which we model to computationally define harmonicity. In Experiment 1, listeners rated pairwise similarity of 20 intensity- and f₀-matched diverse acoustic stimuli. A multidimensional scaling solution revealed two principal timbre dimensions: tone color and harmonicity. These dimensions, respectively, correlated with the computational measures of spectral-envelope slope (r = 0.51) and RMSE deviation from a harmonic series (r = 0.78). The harmonicity dimension also correlated with rated preference (r = 0.48). This enabled creation of an ordered stimulus-preference continuum that ranged from positive (piano) to negative (hawk screech), used for fMRI tests in Experiment 2. Those results revealed a bilateral primary auditory cortical network for all sounds, but with inharmonic exemplars only producing activation in a fronto-insular cortical network, possibly associated with oddball detection. The results are consistent with dissonant sounds violating the regularity of acoustic-object harmonicity, thereby engaging brain regions associated with unexpected stimuli and attentional focus. The findings confirm harmonicity is an important auditory dimension related to preference, with promise to elicit systematic patterns of cortical network activation.

KEYWORDS:

• Consonance-Dissonance
• timbre
• multidimensional scaling
• fMRI
• auditory scene analysis
• auditory cortex
• spectral envelope
• harmonic overtones
• natural regularities

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Comparing positive stimuli to negative and negative to positive are not direct inverses of one another. The GIFT procedure is looking for regions that are coactivated by the first stimuli group that are not present in the second stimuli group. Assume areas A and B are coactivated for stimulus set 1. Areas C and D are coactivated for stimulus set 2. Subtracting set 2 from set 1 yields a network comprised of A and B, but does not indicate that a network of C and D was subtracted. Thus, these seeming inverse analyses will potentially reveal different results.