Comparing sensitivity estimates from MLDS and forced-choice methods in a slant-from-texture experiment.

Area of Science:

• Psychophysics
• Perceptual science
• Visual perception

Background:

• Maximum Likelihood Difference Scaling (MLDS) estimates perceptual scales from difference judgments.
• MLDS has been proposed for estimating near-threshold discrimination performance.
• MLDS may require less data and be preferred by observers over traditional methods.

Purpose of the Study:

• To compare the sensitivity estimation capabilities of MLDS and two-interval forced-choice (2-IFC) methods.
• To assess the theoretical equivalence and empirical agreement between MLDS and 2-IFC.
• To evaluate the performance of MLDS, particularly its confidence interval coverage.

Main Methods:

• Simulations were used to examine the theoretical equivalence between MLDS and 2-IFC.
• Empirical comparison involved a slant-from-texture task using both MLDS and 2-IFC.
• Sensitivity estimation was performed assuming an underlying signal-detection model.

Main Results:

• MLDS and 2-IFC showed agreement in sensitivity estimation, except at low sensitivity or when MLDS assumptions were violated.
• MLDS confidence intervals exhibited low coverage, indicating they were often too narrow.
• Empirical results showed substantial observer-dependent agreement between MLDS and 2-IFC for slant perception.

Conclusions:

• Both MLDS and 2-IFC can be used to estimate sensitivity to differences in slant.
• MLDS offers potential benefits in efficiency and observer preference.
• The limitations of MLDS, particularly concerning confidence interval coverage and assumption violations, warrant careful consideration.