2023 Fiscal Year Final Research Report
Affective computing for mental states inference to solve social dilemma
| Project/Area Number |
21H03782
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 90030:Cognitive science-related
|
|---|
| Research Institution | Gifu University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
山田 誠二 国立情報学研究所, コンテンツ科学研究系, 教授 (50220380)
|
|---|
| Project Period (FY) |
2021-04-01 – 2024-03-31
|
| Keywords | 認知科学 / アフェクティブコンピューティング / 感情 / 心の理論 / 利害対立解消 / 協力行動 |
|---|
| Outline of Final Research Achievements |
This study verified that humans estimate the mental states of others and resolve conflicts of interest through Bayesian inference, using generative models (appraisal models) of mental states and emotional expressions as likelihood functions. This verification employed game-theoretical tasks such as the Prisoner's Dilemma, the Multi-Issue Ultimatum Game, and the Collective Risk Dilemma, utilizing both computer simulations and human experiments. The results demonstrated the validity of generative models based on Social Value Orientation (SVO) theory, multidimensional utility functions representing individual preferences and limits, and models based on emotions of multiple agents. Furthermore, we confirmed that humans actually resolve conflicts of interest through mental state inference using these generative models.
|
| Free Research Field |
人工知能
|
| Academic Significance and Societal Importance of the Research Achievements |
この研究成果の学術的意義は，感情表出と心的状態の関係を定量化し，ベイズ推論による意図推定プロセスを解明したことで，人の協力行動および利害対立解消を下支えする認知，感情メカニズムの計算論的理解とそれに基づく複雑な社会的相互作用のモデル化が可能となったことにある．社会的意義は，トレーニングプログラムによる対人スキルの向上，集団間の対立や社会的ジレンマの解決に対する人認知を考慮した新たなアプローチの設計，より自然で協力的な人-AI相互作用の設計に応用できる基礎的な知見を提供し，社会に実在する利害対立を実際に減らすことができる可能性を提示したことにある．
|
