Separating decision and encoding noise in signal detection tasks.

Area of Science:

• Cognitive Psychology
• Psychophysics
• Computational Neuroscience

Background:

• Signal detection theory (SDT) is a framework for understanding perception and decision-making.
• Internal noise is a key factor influencing response variability in SDT models.
• Existing methods often struggle to disentangle noise sources.

Purpose of the Study:

• To develop and validate an extension to SDT for separately estimating representational and decision noise.
• To provide a computational framework for quantifying distinct sources of internal noise.
• To enhance the precision of SDT models in explaining response variability.

Main Methods:

• Developed a novel SDT framework constraining decision noise models.
• Combined a multipass external noise paradigm with confidence ratings.
• Utilized simulation studies to test estimation accuracy across various noise configurations and decision rules.

Main Results:

• Successfully demonstrated the separate estimation of representational and decision noise.
• Showed robustness of the method across different decision rules and resilience to miss-specification.
• Applied the framework to a visual detection task with varying response categories.

Conclusions:

• The proposed SDT extension effectively separates internal noise into representational and decision components.
• This framework offers a more nuanced understanding of response variability in perceptual tasks.
• The method advances research on quantifying underlying sources of variability in signal detection.