Confirmation Bias Trumps Performance Optimization in Overt Active Learning
- Yul Kang, Department of Engineering, University of Cambridge, Cambridge, United Kingdom
- Daniel Wolpert, Mortimer B. Zuckerman Mind Brain Behavior Institute, Columbia University, New York, New York, United States
- Mate Lengyel, Department of Engineering, University of Cambridge, Cambridge, United Kingdom
AbstractWhen gathering information, different sources typically have distinct levels of informativeness. Therefore, it is optimal to actively select the source of information to learn from (i.e., perform “active learning”). It has been debated whether humans optimize task performance in active learning or use a simple heuristic of seeking information that confirms their beliefs. Critically, depending on one’s subjective beliefs, confirmation bias can in fact be optimal. Thus, without measuring subjective beliefs, previous approaches were unable to distinguish between these alternatives. Using a perceptual decision-making task, we measured participants’ subjective beliefs before and after a new piece of information was presented. We then characterized confirmation-based and performance optimizing strategies with respect to these subjective beliefs. We found that participants’ strategy was dominated by confirmation bias, modulated only weakly by the performance optimization. We discuss potential reasons that may limit performance optimization in active learning.
Return to previous page