Multiplication noise in the human visual system at threshold: 1. Quantum fluctuations and minimum detectable energy

Area of Science:

• Visual neuroscience
• Psychophysics
• Quantum optics

Background:

• The Hecht, Shlaer, and Pirenne experiments established the foundation for understanding human visual sensitivity at the quantum level.
• The visual system operates under threshold conditions where photon detection is crucial for visual perception.
• Signal-detection theory provides a framework for analyzing the trade-offs between sensitivity and reliability in perceptual tasks.

Purpose of the Study:

• To replicate and extend the frequency-of-seeing experiments of Hecht, Shlaer, and Pirenne using modern equipment.
• To quantify the trade-off between sensitivity (frequency of seeing) and reliability (false-positive rate) in human photon detection.
• To develop and validate a new neural-counting model that accounts for neural noise in visual detection.

Main Methods:

• Frequency-of-seeing experiments were conducted using an argon-ion laser (514.5 nm) as the light source.
• Subjects reported perceived stimuli, with instructions to include uncertain detections to explore the sensitivity-reliability trade-off.
• A novel neural-counting model was developed, incorporating Poisson stimulus fluctuations and additive/multiplicative neural noise within signal-detection theory.

Main Results:

• A clear trade-off was observed: detecting 147 photons at the cornea with 60% frequency of seeing resulted in a 1% false-positive rate (FPR).
• Conversely, detecting 34 photons at the cornea with 60% frequency of seeing yielded a 33% FPR.
• The developed neural-counting model accurately predicted experimental frequency-of-seeing data, indicating subjects can detect a single photon at the retina with 60% frequency of seeing at the cost of a 55% FPR.

Conclusions:

• Human visual perception at threshold involves a significant trade-off between detecting weak signals and avoiding false alarms.
• The new neural-counting model provides a robust framework for understanding photon detection in the visual system, accounting for neural noise.
• On average, subjects demonstrate the ability to detect single photons at the retina, albeit with a high probability of false positives.