Examining the cost of control over automatic credit assignment processes

Read full text Summarise with VIRAT
Authors: Ido Ben-Artzi, Noham Wolpe, Nitzan Shahar
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

When learning from reinforcement, individuals should assign credit selectively to features of their actions that causally determine outcomes. However, studies indicate that, despite explicit instructions and extensive experience clarifying which action features predict reward, participants nevertheless assign credit to outcome-irrelevant features, resulting in suboptimal performance and reduced monetary gains. One possible explanation for outcome-irrelevant learning comes from the “expected value of control” theory, which proposes that a cost-benefit tradeoff governs the engagement of cognitive control. Under this framework, inhibiting automatic credit assignment to outcome-irrelevant action features should occur when the extra offered payoff justifies the exertion of control. Here, we directly test this prediction by manipulating the stakes of a multi-armed bandit task within subjects. Using computational model simulations, we derive clear predictions about how higher stakes should promote greater control over automatic credit assignment. Establishing whether the value of control extends to the regulation of inherently false credit assignment is critical for evaluating the explanatory scope of the theory and for clarifying the mechanisms underlying outcome-irrelevant learning.

Fields of Study & Subjects

Neural and Behavioral Psychology Studies Reinforcement Learning in Robotics Decision-Making and Behavioral Economics

Publication details

Publication date
1 January 2026
Language
EN
Document type
Preprint