A dynamic, stochastic, computational model of preference reversal phenomena

Area of Science:

• Decision Science
• Cognitive Psychology
• Behavioral Economics

Background:

• Preference orderings can vary based on how they are elicited (e.g., choice tasks vs. pricing tasks).
• These preference reversals pose a significant challenge to traditional decision theories that assume stable preferences.
• Prior models attempting to explain these reversals often modify decision weights, attribute values, or combination rules, yet none have fully succeeded.

Purpose of the Study:

• To present a novel computational model that explains preference reversals across different elicitation methods.
• To account for empirical trends in preference ordering without altering fundamental decision-making parameters like weights or values.
• To make new predictions about response distributions and response times.

Main Methods:

• Development of a new computational model focusing on dynamic evaluation and response processes.
• Testing the model's ability to predict preference orderings across six distinct elicitation methods.
• Analyzing the model's capacity to maintain stable evaluations of options irrespective of the elicitation method.

Main Results:

• The proposed model successfully predicts preference orderings across six different elicitation methods.
• The model demonstrates stable attribute evaluations across various methods, aligning with empirical observations.
• The model generates novel, testable predictions concerning response distributions and response times.

Conclusions:

• A dynamic evaluation and response process offers a more robust explanation for preference reversals than models altering decision weights or values.
• This new computational framework advances our understanding of decision-making under different elicitation conditions.
• The model's success in predicting both preference orderings and response characteristics provides a unified account of preference reversal phenomena.