Multiplication noise in the human visual system at threshold: 2. Probit estimation of parameters

Area of Science:

• Vision Science
• Mathematical Modeling
• Psychophysics

Background:

• Understanding visual detection requires models that link stimulus properties to perception.
• Existing models often struggle with noise and parameter estimation.

Purpose of the Study:

• To develop a novel mathematical technique for estimating visual detection model parameters.
• To apply this method to a threshold vision model incorporating Poisson noise.
• To quantify key visual parameters like ocular quantum efficiency and retinal dark counts.

Main Methods:

• A normalizing transform is applied to approximate Gaussian response statistics.
• Conventional probit analysis is used for parameter estimation.
• A frequency-of-seeing experiment with controlled stimulus energy levels and multiple trials was conducted.

Main Results:

• The technique successfully estimated parameters for individual subjects across varying stimulus energies and trial numbers.
• Estimated ocular quantum efficiency ranged from 12% to 23%.
• Average retinal dark counts were between 8 and 36, with threshold counts between 11 and 32.

Conclusions:

• The developed mathematical technique provides robust parameter estimation for visual detection models.
• The findings offer quantitative insights into human visual system efficiency and noise characteristics.
• The method's results align with established theories in the absence of specific noise sources.