The psychometric function: I. Fitting, sampling, and goodness of fit.

Area of Science:

• Psychophysics
• Statistical modeling
• Perceptual science

Background:

• Psychometric functions model observer performance against stimulus variables.
• Accurate fitting is crucial for hypothesis testing in psychophysical research.
• Traditional methods struggle with observer errors and small datasets.

Purpose of the Study:

• To present an integrated approach for fitting psychometric functions.
• To develop methods for assessing goodness-of-fit.
• To provide tools for hypothesis testing in psychophysics.

Main Methods:

• Constrained maximum-likelihood estimation for parameter fitting.
• Development of novel goodness-of-fit tests.
• Monte Carlo simulations to address fitting challenges.

Main Results:

• Demonstrated bias in parameter estimates when stimulus-independent errors (lapses) are ignored.
• Showcased the limitations of traditional chi-squared tests with small psychophysical datasets.
• Advocated for Monte Carlo resampling techniques over asymptotic theory.

Conclusions:

• The proposed methods improve the accuracy of psychometric function fitting.
• Accounting for observer lapses is essential for reliable parameter estimation.
• Monte Carlo resampling offers a robust alternative for analyzing small psychophysical datasets.