Effects of test procedure on individual loudness functions

To determine if measurement procedure effects occur for loudness perception data measured using a categorical rating scale.

Loudness data were obtained from 40 normal-hearing adult volunteers, using 30 levels of pure tones at four frequencies (500, 1000, 2000, and 4000 Hz), with judgments made on a 9-point categorical scale. Two presentation orders, random and sequential, were compared within subjects. Subjects were divided into two groups: one group heard only a single tone on every trial, whereas the other group was presented with a maximum level reference tone at the start of each trial.

A significant difference was found between loudness function exponents measured with the random and sequential presentation order of stimulus level. A significant difference was found between loudness function exponents measured when a high-level referent was presented at the start of each trial. The sequential presentation order was further subdivided into ascending and descending runs, and the loudness function exponents for each run were examined separately. The results showed a significant interaction between sequence (ascending versus descending) and group.

For normal-hearing listeners, the procedure used to measure loudness has an effect on the loudness function exponent obtained. These results appear to be related to stimulus context effects. Loudness function exponents are smaller when the stimulus is preceded by a stimulus level greater than the level of the test tone. This occurred when a high-level referent was presented at the start of each trial or when the stimulus level from the previous trial was greater than the test level, as in a descending run. It seems likely that the difference between loudness function exponents obtained with a random and sequential presentation of level can be explained by the same phenomenon. The significance of these results for hearing aid fittings in which loudness normalization is the goal is discussed.