Related Experiment Videos
Maximum likelihood difference scales represent perceptual magnitudes and predict appearance matches.
Christiane B Wiebel1, Guillermo Aguilar2, Marianne Maertens1
1Modeling of Cognitive Processes, Department of Software Engineering and Theoretical Computer Science, Technische Universität Berlin, Berlin, Germany.
Journal of Vision
|April 7, 2017
Summary
This study introduces maximum likelihood difference scaling to measure perceptual lightness scales directly. This new method accurately predicts lightness perception and outperforms traditional adjustment methods.
Area of Science:
- Visual Perception
- Psychophysics
- Computational Neuroscience
Background:
- Understanding the link between internal experience and physical variables is crucial in perception research.
- Traditional methods like the method of adjustment have limitations, including indirect measurement of perceptual scales and the need for cross-context comparisons.
Purpose of the Study:
- To develop a more direct method for measuring perceptual scales of surface lightness.
- To overcome the shortcomings of the method of adjustment in perception research.
Main Methods:
- Utilized maximum likelihood difference scaling to measure perceptual lightness scales.
- Observers compared surface lightness within the same context, avoiding cross-condition comparisons.
- Collected data on perceptual matches using a conventional adjustment experiment.
Main Results:
- Perceptual scales derived from maximum likelihood difference scaling directly measured surface lightness.
- These scales accurately predicted perceptual matches from adjustment experiments, both qualitatively and quantitatively.
- A contrast-based model of lightness perception explained a significant portion of the variance in both scaling (98%) and matching (88%) data.
Conclusions:
- Maximum likelihood difference scaling offers a more direct and accurate approach to measuring perceptual scales.
- The findings support a contrast-based model for lightness perception.
- Perceptual scales may be closer to the underlying variables of interest than direct matching data.