A Stochastic Version of General Recognition Theory

Area of Science:

• Cognitive psychology
• Perception
• Signal detection theory

Background:

• General Recognition Theory (GRT) traditionally offers a static framework for multivariate signal detection.
• Previous GRT models lacked dynamic process interpretations of perception.

Purpose of the Study:

• To present a stochastic version of GRT incorporating moment-by-moment perceptual fluctuations.
• To model the interplay between perceptual channels and decision-making processes.

Main Methods:

• Utilized a multivariate diffusion process to model perceptual channel output fluctuations.
• Incorporated a decision stage with linear or quadratic functions driving a univariate diffusion process for response determination.

Main Results:

• Established conditions for equivalence between stochastic and static GRT accuracy predictions.
• Demonstrated that decisional influences can corrupt traditional estimates of perceptual noise.

Conclusions:

• The stochastic GRT provides a more dynamic and process-oriented interpretation of perception.
• Highlights the need to consider decision-making when estimating perceptual noise in signal detection research.