Further studies of a maximum-likelihood yes-no procedure

Area of Science:

• Auditory Psychophysics
• Signal Detection Theory
• Psychoacoustics

Background:

• Accurate estimation of auditory absolute thresholds is crucial for diagnosing hearing impairments.
• Traditional and adaptive procedures exist, but their sensitivity to response biases requires investigation.
• Understanding the influence of false-alarm rates on threshold estimation is essential for reliable auditory testing.

Purpose of the Study:

• To compare threshold estimates derived from a maximum-likelihood yes-no procedure and a traditional forced-choice adaptive procedure.
• To investigate the impact of false-alarm rates and response criteria (liberal vs. conservative) on auditory threshold estimates.
• To evaluate the effectiveness of a stimulus-power model in describing psychometric functions and informing maximum-likelihood procedures.

Main Methods:

• Absolute thresholds for pure tones (500, 1000, 2000 Hz) were measured using maximum-likelihood yes-no and three-down one-up forced-choice adaptive procedures.
• Listener response criteria (liberal/conservative) were manipulated in a second experiment using constant-stimulus yes-no and maximum-likelihood yes-no procedures.
• Psychometric functions were modeled using a stimulus-power model, and computer simulations assessed the effects of inattention and parameter mismatches.

Main Results:

• Threshold estimates from the two primary procedures were highly correlated.
• False-alarm rates significantly influenced threshold estimates, with estimates being 5-6 dB higher for low false-alarm rates.
• A stimulus-power model provided the best fit to data from the constant-stimulus method.

Conclusions:

• The maximum-likelihood yes-no procedure requires careful estimation of false-alarm probability for unbiased results.
• Listener response bias (criterion) demonstrably affects absolute threshold measurements in auditory tasks.
• The stimulus-power model offers a robust framework for understanding and modeling psychometric functions in auditory detection.