| Paper Abstract and Keywords |
| Presentation |
2026-07-02 15:25
A Computational Model of Anchoring and Adjustment Processes via Gibbs Sampling Isao Ozawa, Airi Ono, Fumiya Komatsu, Tomoaki Hamada, Yoshihito Yasaki, Kazuki Takahashi, Takashi Takekawa (Kogakuin Univ.) NC2026-5 IBISML2026-5 |
| Abstract |
(in Japanese) |
(See Japanese page) |
| (in English) |
The anchoring effect can be interpreted as Bayesian updating by treating prior knowledge as a prior distribution and the anchor as uncertain data. Conversely, a widely recognized qualitative hypothesis known as "Anchoring and Adjustment" suggests that individuals initially accept the anchor and then sequentially adjust it based on their own knowledge. In this presentation, we establish a foundational model that derives the joint posterior distribution of the true value $mu$ and the anchor's reliability $gamma$, and propose a computational hypothesis assuming that $mu$ and $gamma$ are iteratively updated through Gibbs sampling. Under this hypothesis, the dynamic transition from the initial state, where $mu$ is strongly biased toward the anchor, to the convergence toward the posterior distribution through iteration, elegantly maps onto the adjustment process. Furthermore, by introducing a stopping rule based on convergence criteria, the model naturally accounts for the variability in response times observed in empirical data. In this study, we construct a hierarchical Bayesian model incorporating the proposed mechanism and estimate its parameters using MCMC to comprehensively reproduce the cognitive process from anchor presentation to response and response time. |
| Keyword |
(in Japanese) |
(See Japanese page) |
| (in English) |
Anchoring and Adjustment / Gibbs Sampling / Dynamics / Response Time / Hierarchical Bayesian Model / / / |
| Reference Info. |
IEICE Tech. Rep., vol. 126, no. 90, NC2026-5, pp. 18-18, July 2026. |
| Paper # |
NC2026-5 |
| Date of Issue |
2026-06-25 (NC, IBISML) |
| ISSN |
Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
| Download PDF |
NC2026-5 IBISML2026-5 |