The relation between the generalized matching law and signal-detection theory

Area of Science:

• Behavioral science
• Psychology
• Reinforcement learning

Background:

• The generalized matching law (GML) is a fundamental principle in behavioral science.
• Signal-detection theory (SDT) is widely used to analyze performance in discrimination tasks.

Purpose of the Study:

• To integrate the generalized matching law with signal-detection theory.
• To develop a unified framework for analyzing response patterns and reinforcement.
• To provide a quantitative measure of stimulus discrimination within the GML.

Main Methods:

• Applying the GML to a signal-detection matrix to derive two key equations.
• Analyzing the relationship between responding, reinforcement, and stimulus presence/absence.
• Utilizing logarithmic ratio and z-proportion transformations for bias measurement.
• Combining biases to derive a measure equivalent to SDT's d' and eta.

Main Results:

• The GML yields two equations relating response rates to reinforcement, with and without stimuli.
• Oppositely signed biases in these equations indicate evidence of stimulus discrimination.
• A combined measure of absolute logarithmic biases is equivalent to SDT's d' and eta.
• Eliminating stimulus-induced biases yields a GML statement on overall performance and reinforcement.

Conclusions:

• The generalized matching law offers a robust framework for understanding behavior in discrimination tasks.
• This integration provides a unified quantitative measure for both stimulus discrimination and overall performance.
• The findings bridge GML and SDT, offering a more comprehensive understanding of behavioral allocation and reinforcement.