Rapid Profiling of Loudness Among Older Adults

Efficient assessment of loudness may be an important step to enable custom signal-processing schemes in hearing devices to meet individual listeners' unique hearing profiles. A Bayesian adaptive procedure, the quick-Categorical Loudness Scaling (qCLS) procedure, to estimate loudness growth across a wide frequency range was evaluated in the present study in terms of its short-term test-retest reliability (in experiment 1) and agreement with existing loudness assessment methods (in experiment 2).

In experiment 1, 41 older adults were recruited, from whom the loudness growth across the frequency range from 250 to 8000 Hz was estimated using the qCLS procedure. During the procedure, the participants provided categorical loudness ratings to pure-tone stimuli presented at various frequencies and levels. The collected responses were used to update the estimates of equal-loudness-level contours at 20, 40, 60, 80, and 100 phons, and the test stimulus for the subsequent trial was selected randomly along 1 of these interim estimated equal-loudness contours. Each run of the procedure took 100 test trials to complete, approximately 5 min. Two runs of the qCLS procedure (test and retest) were conducted in the same test session. In experiment 2, 10 adults were recruited, and the loudness growth was assessed at 500, 1000, 2000, and 4000 Hz, using both the qCLS procedure (100 trials) and a widely adopted adaptive procedure, that is, Adaptive Categorical Loudness Scaling (ACALOS) procedure (300 trials).

The test-retest reliability was evaluated in experiment 1, for the estimated thresholds that separated the 11 response categories used for loudness rating, that is, the category boundaries. The mean absolute deviation between test and retest, averaged across participants, was 6.3 dB for the category boundaries. Results from experiment 2 showed that the one run of the qCLS procedure (100 trials) and the ACALOS procedure (300 trials) led to a mean absolute deviation, averaged across participants, of 5.1 dB.

Satisfactory reliability may be achieved using the qCLS procedure within 100 trials of testing, which takes approximately 5 min to complete. The estimated loudness growth function also agrees well with those obtained using the ACALOS procedure.