Host: The Japanese Society for Artificial Intelligence
Name : The 38th Annual Conference of the Japanese Society for Artificial Intelligence
Number : 38
Location : [in Japanese]
Date : May 28, 2024 - May 31, 2024
In decision-making, cognitive biases like the anchoring effect can unconsciously skew judgments. This phenomenon involves using a presented number as an anchor, influencing subsequent responses. The responses vary based on the anchor, and the effect weakens when answers deviate significantly. The Anchoring-and-Adjustment concept suggests that initial responses are close to the anchor, adjusting over time to a plausible range. Previous research, inspired by the Bayesian Brain hypothesis, qualitatively replicated the mechanism through Bayesian updates. However, specific computational processes for subsequent responses need further exploration. This study employs variational Bayesian methods, assuming the brain iteratively considers anchors and expectations. It investigates how iteration influences the anchoring effect and examines the computational processes underlying responses with different iterations. Results show differences in responses based on iteration, providing insights into its influence. The study highlights methods individuals employ in generating responses through iterative consideration of anchors and expectations.