Efficient adaptive procedures for threshold and concurrent slope estimates for psychophysics and speech...

Area of Science:

• Audiology
• Psychophysics
• Speech processing

Background:

• Adaptive procedures are crucial for efficient audiological testing.
• Estimating speech reception thresholds (SRT) and psychometric function slopes concurrently requires careful optimization.
• Binomial theory provides a framework for calculating minimum achievable standard deviations.

Purpose of the Study:

• To determine optimal target values for concurrent estimation of thresholds and slopes.
• To calculate minimum standard deviations as functions of stimulus level and track length.
• To introduce and evaluate an adaptive procedure based on optimal targets.

Main Methods:

• Calculations based on binomial theory for optimal target values (p1=0.19, p2=0.81).
• Development and Monte Carlo simulation of an adaptive procedure.
• Evaluation with normal-hearing and hearing-impaired listeners using German sentence intelligibility tests with varying 'j' factors.

Main Results:

• Optimal targets for the logistic psychometric function identified.
• Adaptive procedure efficiency increases with multiple Bernoulli trials per response.
• No significant advantage observed for higher 'j' factor sentence tests, possibly due to training effects.

Conclusions:

• At least 30 sentences are needed for reliable SRT and slope estimates with word scoring.
• The number of within-sentence trials (j factor) influences efficiency but may be confounded by training.
• Further research is needed to fully understand the impact of 'j' factor and training in adaptive audiological testing.