Multi-attribute, multi-alternative models of choice: Choice, reaction time, and process tracing.

Area of Science:

• Decision science
• Cognitive psychology
• Computational modeling

Background:

• Multi-alternative, multi-attribute choice is fundamental to decision-making.
• Existing computational models (utility, heuristic, dynamic) offer different explanations for choice behavior.
• Understanding how attention influences preference formation is crucial for refining these models.

Purpose of the Study:

• To compare the predictive accuracy of utility, heuristic, and dynamic computational models for multi-alternative, multi-attribute choice.
• To evaluate model predictions for choice probabilities and response times.
• To develop and test new models that integrate attentional processes into decision-making frameworks.

Main Methods:

• Contrasted choice predictions of utility, heuristic, and dynamic models using preferential and risky choice data.
• Assessed model fit using maximum likelihood and cross-validation.
• Tested response time predictions and analyzed process tracing data (eye/mouse tracking).
• Developed and validated the models of attentional sampling (MAS) framework.

Main Results:

• Dynamic models, particularly the multi-attribute linear ballistic accumulator (MLBA), outperformed heuristic models for risky choice.
• MLBA showed some accuracy in response time prediction but limited stimulus-level explanation.
• Process tracing data indicated models did not fully capture attention-preference interactions.
• A specific MAS variant demonstrated strong performance by integrating gaze patterns.

Conclusions:

• Dynamic models provide a better account of choice behavior than heuristic models.
• Current models inadequately represent the interplay between attention and preference formation.
• The developed models of attentional sampling (MAS) offer a promising direction for future research by incorporating empirical gaze data.