Skip to main content

Open Access Using Probabilistic Choice Models to Investigate Auditory Unpleasantness

Download Article:
The potential of probabilistic choice models (the Bradley-Terry-Luce model, or preference trees) in scaling the perceived unpleasantness of sounds was evaluated. To that effect, 74 subjects made pair-wise comparisons of the unpleasantness of twelve binaurally-recorded, environmental sounds presented over headphones. The stimuli varied in their psychoacoustic characteristics and half of them were of technical, half of natural origin. A more sophisticated model than previously tested, namely a preference tree, was identified to account well for the structure underlying the data, indicating (1) that subjects changed criteria, when evaluating different sound pairs, and that (2) these criterion changes combined in a lawful way, so that it was possible to measure unpleasantness on a ratio-scale level across the entire set of sounds investigated. Contrary to expectation, sound origin (technical or natural) did not influence the unpleasantness judgments. Instead, the sounds could be grouped according to their (non-acoustical) intrusiveness, and loudness. A subsequent multiple-regression analysis showed that in the sub-groups of soft and loud sounds, a combination of sharpness (Smean ) and roughness (Rmean ), the latter differing in magnitude for the two groups, explained the unpleasantness-scale values very well (r 2 corr = 0.91). Direct magnitude estimates of unknown scale type, collected from the same listeners covered a much smaller range of ratios, and were roughly linear with the logarithm of the ratio scale derived from the preference tree. The advantage of the choice-theory modeling in providing information on the structure underlying the judgments is discussed.

Document Type: Research Article

Publication date: November 1, 2004

  • Access Key
  • Free content
  • Partial Free content
  • New content
  • Open access content
  • Partial Open access content
  • Subscribed content
  • Partial Subscribed content
  • Free trial content