A computational approach that evaluates the value calculation process and the choice process in decision making

• Project/Area Number: 18K13366
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 10040:Experimental psychology-related
• Research Institution: Senshu University (2020); Nagoya University (2018-2019)
• Principal Investigator: Toyama Asako 専修大学, 文学研究科, 特別研究員 (10816549)
• Project Period (FY): 2018-04-01 – 2021-03-31
• Project Status: Completed (Fiscal Year 2020)
• Budget Amount: ¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000); Fiscal Year 2020: ¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000); Fiscal Year 2019: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000); Fiscal Year 2018: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
• Keywords: 計算論モデル / 強化学習モデル / 忘却過程 / 固執性 / 精神疾患 / シミュレーション / パラメータ推定のバイアス / 意思決定 / モデルフリー学習 / モデルベース学習 / 固執

Outline of Final Research Achievements

Decision-making is composed of a value calculation process that calculates the value of available options and a choice process that favors an option. In recent years, studies have characterized mental illness and personalities using model parameters. In this study, we pointed out the problems with the conventionally used reinforcement learning models and clarified the possibility that the model parameter estimation was biased due to model misspecification, which affected the conclusion of studies. In addition, we collected large-scale data using crowdsourcing and newly characterized mental illness with the proposed model.

Academic Significance and Societal Importance of the Research Achievements

本研究は従来使われてきた強化学習モデルのパラメータが，意思決定過程に関して誤った結論を導いていた可能性を指摘した。そのため，本研究の中で提案している新しいモデルを使うことで，これまでの理解が覆り，新たな人間像が浮かび上がってくる可能性がある。さらに，新しく提案したモデルのパラメータが抑うつや不安などの程度と対応していることが示唆された。そのため，近年世界的に推進されている計算論的精神医学に貢献し，精神疾患の診断や予測に役立つことが期待される。

Research Products

• [Journal Article] Revisiting the importance of model fitting for model-based fMRI: It does matter in computational psychiatry (2021)
  • Author(s): Katahira Kentaro, Toyama Asako
  • Journal Title: PLOS Computational Biology Volume: 17 Issue: 2 Pages: e1008738-e1008738
  • DOI: 10.1371/journal.pcbi.1008738
  • Peer Reviewed / Open Access
• [Journal Article] Reinforcement learning with parsimonious computation and a forgetting process (2019)
  • Author(s): Toyama Asako, Katahira Kentaro, Ohira Hideki
  • Journal Title: Frontiers in Human Neuroscience Volume: 13 Pages: 153-153
  • DOI: 10.3389/fnhum.2019.00153
  • Peer Reviewed / Open Access
• [Journal Article] Biases in estimating the balance between model-free and model-based learning systems due to model misspecification (2019)
  • Author(s): Toyama Asako, Katahira Kentaro, Ohira Hideki
  • Journal Title: Journal of Mathematical Psychology Volume: 91 Pages: 88-102
  • DOI: 10.1016/j.jmp.2019.03.007
  • Peer Reviewed / Open Access
• [Presentation] ATTENTION BIAS TO AFFECTIVE OUTCOMES (2020)
  • Author(s): Asako Toyama, Kentaro Katahira Hideki Ohira
  • Organizer: The 2020 SAS Annual Conference
  • Int'l Joint Research
• [Presentation] Forgetting Process in Model-Free and Model-Based Reinforcement Learning. (2019)
  • Author(s): Toyama, A., Katahira, K., & Ohira, H.
  • Organizer: The 4th Multi-disciplinary Conference on Reinforcement Learning and Decision Making
  • Int'l Joint Research
• [Presentation] モデルベース意思決定を特徴づける計算論モデルの 提案 モデルパラメータを用いた個人差の検討 (2019)
  • Author(s): 遠山朝子
  • Organizer: 日本心理学会第81回大会
• [Presentation] 忘却過程のある強化学習モデルの提案 (2019)
  • Author(s): 遠山朝子
  • Organizer: 第18回東海若手実験心理学研究会
• [Presentation] 忘却過程を含む強化学習モデルを用いたモデルフリー・モデルベースシステム比重の推定 (2019)
  • Author(s): 遠山朝子
  • Organizer: 第22回計算論的精神医学コロキウム