Improving parameter recovery for conflict drift-diffusion models

Area of Science:

• Cognitive psychology
• Computational neuroscience
• Decision-making models

Background:

• Drift-diffusion models (DDMs) are used to explain performance in conflict tasks.
• Existing DDMs feature varying architectures despite a common drift rate characteristic.
• Current comparative studies focus on model fit, neglecting parameter recovery, essential for valid interpretation.

Purpose of the Study:

• To introduce and evaluate an enhanced fitting procedure for drift-diffusion models.
• To improve both model fit and parameter recovery in cognitive task analysis.
• To address limitations in existing methods for assessing DDM performance.

Main Methods:

• Developed a novel fitting procedure utilizing a grid search for initial parameter values.
• Implemented a specific criterion for evaluating the goodness of fit.
• Conducted simulation studies to compare the new procedure against standard methods.

Main Results:

• The improved procedure significantly enhanced model fit compared to standard methods.
• Parameter recovery performance saw substantial improvements with the new procedure.
• The most complex drift-diffusion models exhibited the largest gains in fit and recovery.

Conclusions:

• The proposed fitting procedure offers a superior approach for analyzing drift-diffusion models.
• Enhanced parameter recovery validates the use of complex models in cognitive research.
• This method provides a more reliable framework for interpreting cognitive processes in conflict tasks.