Computational modeling of choice-induced preference change: A Reinforcement-Learning-based approach

Area of Science:

• Cognitive Science
• Neuroscience
• Decision Science

Background:

• Value learning is typically studied via externally guided decision-making (EDM), explained by reinforcement learning (RL).
• Internally guided decision-making (IDM) involves learning without external correctness, often attributed to choice-induced preference change (CIPC).
• The choice-based learning (CBL) model was proposed for CIPC but lacked empirical validation in IDM.

Purpose of the Study:

• To examine the validity and applicability of the CBL model in internally guided decision-making (IDM).
• To investigate the computational mechanisms underlying value updating during preference judgments.

Main Methods:

• Simulations using the CBL model.
• Conducting a behavioral experiment involving a preference judgment task with novel contour shapes.
• Applying computational model analyses to behavioral data to compare model fits.

Main Results:

• The CBL model, where both chosen and rejected item values are updated, provided a significantly better fit to IDM behavioral data than alternative models.
• This study provides the first computational evidence from choice behavior data that both chosen and rejected item values are updated in IDM.

Conclusions:

• The CBL model is a valid and effective framework for understanding value learning in internally guided decision-making (IDM).
• Demonstrates that choice-induced preference change (CIPC) in IDM involves updating the values of both selected and unselected options, contrary to previous findings based on rating data.