Modeling confidence and response time in recognition memory.

Area of Science:

• Cognitive Psychology
• Computational Neuroscience
• Psychophysics

Background:

• Recognition memory confidence judgments are crucial for understanding memory retrieval.
• Existing models often rely on signal detection theory, which may oversimplify the underlying processes.
• The interpretation of receiver operating characteristic (z-ROC) functions in recognition memory is debated.

Purpose of the Study:

• To introduce a novel computational model for confidence judgments in recognition memory.
• To test the model's ability to account for empirical data on confidence and response times.
• To re-evaluate the standard signal detection interpretation of z-ROC functions.

Main Methods:

• Developed a diffusion model where confidence criteria influence drift rates in racing Ornstein-Uhlenbeck processes.
• Fit the model to confidence judgments and quantile response times from two recognition memory experiments.
• Manipulated factors such as word frequency and speed-accuracy trade-offs.

Main Results:

• The proposed model successfully captured confidence judgments and response time distributions.
• Demonstrated that the standard signal detection interpretation of z-ROC functions is inadequate.
• Explained sequential effects in z-ROC function slopes, showing a ~10% change based on prior responses.

Conclusions:

• The new diffusion model provides a more accurate account of confidence judgments in recognition memory.
• Findings challenge the prevailing signal detection theory framework for interpreting z-ROC data.
• The model offers insights into the dynamic processes underlying memory retrieval and confidence assessment.