Modeling reaction time and accuracy of multiple-alternative decisions.

Area of Science:

• Cognitive Psychology
• Computational Neuroscience
• Decision Science

Background:

• Sequential-sampling models are crucial for understanding decision-making processes.
• Racing diffusion models are frequently used to model evidence accumulation in perceptual tasks.

Purpose of the Study:

• To evaluate various sequential-sampling models with racing diffusion processes for multiple-alternative decisions.
• To identify the best-fitting model structures for perceptual discrimination data.

Main Methods:

• Data from two perceptual discrimination experiments were analyzed.
• Model structures varied in parameters like lateral inhibition, evidence decay, negative evidence, and starting point variability.
• A letter discrimination task manipulated stimulus difficulty, response alternative probability, and the number of alternatives.

Main Results:

• Model-fitting analyses successfully ruled out numerous model classes.
• The most successful models featured minimal to moderate evidence decay and no lateral inhibition.
• The best models indicated that adding response alternatives impacts non-decisional time, caution, or evidence quality.

Conclusions:

• The study refines understanding of evidence accumulation in multi-alternative decision tasks.
• Findings suggest specific architectural features of cognitive models are critical for explaining perceptual decision-making.
• The research provides a framework for future investigations into the neural and cognitive underpinnings of choice.