Quantal noise and decision rules in dynamic models of light adaptation

Area of Science:

• Computational Neuroscience
• Visual Psychophysics
• Sensory System Modeling

Background:

• Understanding light adaptation is crucial for explaining visual performance under varying light conditions.
• Probabilistic processes, such as quantal noise, are often included in models of visual system dynamics.
• The role of different noise sources and decision rules in visual sensitivity remains an active area of research.

Purpose of the Study:

• To evaluate the impact of incorporating probabilistic processes, like quantal noise, into computable models of light adaptation.
• To investigate how different model components, including noise stages and decision rules, influence visual sensitivity.
• To compare model predictions with empirical data and assess the contribution of quantal fluctuations to visual limitations.

Main Methods:

• Developed a general class of computable models for light adaptation dynamics with four distinct stages: early noise, deterministic filtering/gain change, late noise, and a decision rule.
• Analyzed two decision rules: an ideal (signal-known-exactly) detector and a peak-trough detector.
• Examined observer sensitivity as a function of mean luminance and temporal frequency under different model configurations.

Main Results:

• With an ideal detector and no late noise, model sensitivity was independent of filtering/gain stages, predicting a square-root luminance and flat temporal frequency function, contradicting data.
• A peak-trough detector model showed sensitivity reflecting the low-level filtering/gain stage.
• Late noise was necessary for the peak-trough model to exhibit both square-root and Weber regions in luminance-dependent sensitivity.

Conclusions:

• The ideal detector model, particularly without late noise, fails to account for empirical visual sensitivity data.
• The peak-trough detector, combined with late noise, provides a more plausible explanation for observed luminance and temporal frequency dependencies.
• Evidence suggests that quantal fluctuations may not be the primary limiting factor for visual sensitivity under typical conditions.