Action repetition biases choice in context-dependent decision-making.

Area of Science:

• Cognitive Science
• Neuroscience
• Behavioral Economics

Background:

• Decision biases make human behavior appear irrational.
• Context-dependent choices are influenced by past experiences.
• Existing computational models often need extensive environmental data.

Purpose of the Study:

• To test if decision biases are driven by context-specific action repetition.
• To develop and validate a computational model explaining these biases.

Main Methods:

• Conducted nine value-based decision-making tasks with 351 participants.
• Reanalyzed six prior datasets involving 350 participants.
• Developed a hierarchical Bayesian reinforcement learning model incorporating reward learning and action repetition.

Main Results:

• Increased within-context action repetition correlated with biased choices, including higher subjective valuation and lower uncertainty.
• The proposed model successfully explained biased choices in stable environments.
• Model comparison indicated superior performance over value normalization and goal-centric accounts.

Conclusions:

• Action repetition is a key driver of decision biases.
• A parsimonious model combining reward learning and action repetition sufficiently explains these biases.
• Findings offer insights into habit-like choice tendencies in decision-making.