Separating transducer non-linearities and multiplicative noise in contrast discrimination

Area of Science:

• Visual Neuroscience
• Perceptual Psychology
• Computational Vision

Background:

• Contrast discrimination is fundamental to the visual system.
• Previous research debated whether saturating response functions or increasing noise limits contrast discrimination thresholds.

Purpose of the Study:

• Reanalyze existing datasets on contrast discrimination.
• Investigate the role of multiplicative noise versus saturating contrast response functions.
• Determine the factors limiting contrast discrimination.

Main Methods:

• Reanalyzed four datasets using novel fitting methods.
• Employed models incorporating multiplicative noise and flexible contrast response functions.
• Examined alternative parameterizations for intuitive interpretation.

Main Results:

• Two datasets were deemed too low in threshold for model discrimination.
• One dataset supported the multiplicative noise claim.
• The fourth dataset, despite a high chi-squared fit, strongly supported multiplicative noise, potentially trial-dependent.

Conclusions:

• Data support a generalized form of multiplicative noise limiting contrast discrimination.
• The precise nature of this noise (e.g., trial-dependent) requires further investigation.
• The jury remains out on the definitive factors limiting contrast discrimination.