AI Learning and Conceptual Transfer in the Game of Hidden Rules

隐藏规则游戏中的人工智能学习与概念迁移

Abstract: This report summarizes the work conducted on the Game of Hidden Rules (GOHR), focusing on reinforcement learning agents trained to infer hidden rules from trial-and-error feedback, representation design, rule difficulty analysis, transfer learning, generalization, and pseudo-bot-assisted human learning analysis. The report focuses on the Transformer-based A2C framework, Feature-Centric and Object-Centric representations, experimental findings, and classification of human learning data.

摘要： 本报告总结了在“隐藏规则游戏”（Game of Hidden Rules, GOHR）方面所开展的研究工作。研究重点包括：训练强化学习智能体通过试错反馈推断隐藏规则、表征设计、规则难度分析、迁移学习、泛化能力，以及伪机器人辅助的人类学习分析。报告重点探讨了基于 Transformer 的 A2C 框架、以特征为中心（Feature-Centric）和以对象为中心（Object-Centric）的表征方式、实验结果，以及人类学习数据的分类。

---

Paper Details:

• Authors: Christo Mathew, Wentian Wang, Jacob Feldman, Lazaros K. Gallos, Paul B. Kantor, Vladimir Menkov, Hao Wang
• Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
• Submission Date: 26 Jun 2026
• DOI: https://doi.org/10.48550/arXiv.2608.21372

论文详情：

• 作者： Christo Mathew, Wentian Wang, Jacob Feldman, Lazaros K. Gallos, Paul B. Kantor, Vladimir Menkov, Hao Wang
• 学科分类： 人工智能 (cs.AI)；机器学习 (cs.LG)
• 提交日期： 2026年6月26日
• DOI： https://doi.org/10.48550/arXiv.2608.21372