Elamur Research⌕
☰
home / papers / 2026 / ImplicitMemBench: Measuring Unconscious Behavioral Adaptation in Large Language Models

ImplicitMemBench: Measuring Unconscious Behavioral Adaptation in Large Language Models

Paper · 2026 · Chonghan Qin, Xiachong Feng, Weitao Ma et al.

Summary

Existing memory benchmarks for LLM agents, such as LoCoMo, LongMemEval and MemoryAgentBench, test whether a model can deliberately retrieve stored facts when asked. They do not test implicit memory, where past experience quietly changes later behavior without any prompt to recall it. The authors present ImplicitMemBench, a benchmark of 300 items built on the cognitive-science taxonomy of non-declarative memory. It covers three paradigms:

  • Procedural Memory: learning a skill from a single exposure that must survive distraction.
  • Priming: thematic bias, measured by comparing experimental and control versions of each item.
  • Classical Conditioning: CS–US associations that should shape the model's first choice.

Every item follows a Learning/Priming–Interfere–Test sequence, and only the model's first attempt is scored. Items are short, around 500 tokens. Scoring uses rule-based validators and LLM judges.

The authors tested 17 proprietary and open-source models. None scored above 66% overall, and all were well below human baselines. Models were much better at forming preferences than at learning to inhibit behavior. The benchmark matters because practical assistants need to adjust their behavior automatically, for example by avoiding tools that keep failing, rather than only recalling facts.

Key points

  • The best overall scores were DeepSeek-R1 at 65.3%, Qwen3-32B at 64.1% and GPT-5 at 63.0%. Procedural memory was the easiest paradigm and classical conditioning the hardest.
  • Inhibition tasks averaged 17.6%, compared with 75.0% for preference tasks. Jargon avoidance was solved only about 4% of the time.
  • Among the top-5 models, surface formatting tasks reached 93.8% accuracy, while deep multi-rule protocols reached only 60.0%. This suggests models learn individual rules but struggle to combine them.
  • Stronger priming correlated with more constraint violations (r=0.63). The authors interpret this as models copying style rather than transferring meaning.
  • Adding explicit memory modules to agents did not reliably raise scores. The authors conclude that implicit memory is more than storage plus retrieval.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Harbin Institute of TechnologyHarbin · China158 works · 28,841 citations🎓University of Hong KongPok Fu Lam · China89 works · 11,774 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksNeuroscience · 165 worksCognitive Neuroscience · 164 works

🧭 Topics 1Research subjects and languages the work is about.Language models & pretraining · 3,320 works

Cite

Cite as Chonghan Qin, Xiachong Feng, Weitao Ma, Xiaocheng Feng and Lingpeng Kong. 2026. ImplicitMemBench: Measuring Unconscious Behavioral Adaptation in Large Language Models. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.1301

@inproceedings{qin2026implicitmembench,
  title = {{ImplicitMemBench: Measuring Unconscious Behavioral Adaptation in Large Language Models}},
  author = {Chonghan Qin, Xiachong Feng, Weitao Ma, Xiaocheng Feng and Lingpeng Kong},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1301},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1301}
}

References

20 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Anthropic. 2025. Model report: Claude opus 4 & sonnet 4. Anthropic Transparency Hub. Last updated August 7, 2025. https://www.anthropic.com/transparency/model-report. ↗
  2. David Castillo-Bolado, Joseph Davidson, Finlay Gray, and Marek Rosa. 2024. Beyond prompts: Dynamic conversational benchmarking of large language models. In The Thirty-eight Conference on Neural Information Processing Systems Datasets and Benchmarks Track. find
  3. DeepSeek-AI, Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, Xiaokang Zhang, Xingkai Yu, Yu Wu, Z. F. Wu, Zhibin Gou, Zhihong Shao, Zhuoshu Li, Ziyi Gao, and 181 others. 2025. Deepseek-r1: Incentivizing reasoning capability in ll…
  4. Darshan Girish Deshpande, Varun Prashant Gangal, Hersh Mehta, J˛edrzej Rosłaniec, Anand Kannappan, Rebecca Qian, and Peng Wang. 2025. MEM-TRACK: Evaluating long-term memory and state tracking in multi-platform dynamic agent environments. In Workshop on Scaling Environments for Agents. find
  5. Junqing He, Liang Zhu, Rui Wang, Xi Wang, Gholamreza Haffari, and Jiaxing Zhang. 2025. MADial-bench: Towards real-world evaluation of memory-augmented dialogue generation. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: H… find
  6. Zihong He, Weizhe Lin, Hao Zheng, Fan Zhang, Matt W. Jones, Laurence Aitchison, Xuhai Xu, Miao Liu, Per Ola Kristensson, and Junxiao Shen. 2024. Human-inspired perspectives: A survey on ai long-term memory. find
  7. Yuanzhe Hu, Yu Wang, and Julian McAuley. 2026. Evaluating memory in LLM agents via incremental multi-turn interactions. In The Fourteenth International Conference on Learning Representations. find
  8. Adyasha Maharana, Dong-Ho Lee, Sergey Tulyakov, Mohit Bansal, Francesco Barbieri, and Yuwei Fang. 2024. Evaluating very long-term conversational memory of LLM agents. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2024, Bangkok… find
  9. OpenAI. 2024. Memory and new controls for chatgpt. OpenAI Blog. Accessed 2025-09. find
  10. OpenAI. 2025. Gpt-5 system card. https://openai.com/index/gpt-5-system-card/. Accessed 2025-09. ↗
  11. Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, and 25 others. 2024. Qwen2.5 technical report. find
  12. Larry R Squire. 2004. Memory systems of the brain: a brief history and current perspective. Neurobiology of learning and memory, 82(3):171–177. find
  13. Haoran Tan, Zeyu Zhang, Chen Ma, Xu Chen, Quanyu Dai, and Zhenhua Dong. 2025. MemBench: Towards more comprehensive evaluation on the memory of LLM-based agents. In Findings of the Association for Computational Linguistics: ACL 2025, pages 19336–19352, Vienna, Austria. Association for Computational L… find
  14. Di Wu, Hongwei Wang, Wenhao Yu, Yuwei Zhang, Kai-Wei Chang, and Dong Yu. 2025. Longmemeval: Benchmarking chat assistants on long-term interactive memory. In The Thirteenth International Conference on Learning Representations.
  15. Likang Wu, Zhi Zheng, Zhaopeng Qiu, Hao Wang, Hongchao Gu, Tingjia Shen, Chuan Qin, Chen Zhu, Hengshu Zhu, Qi Liu, Hui Xiong, and Enhong Chen. 2024. A survey on large language models for recommendation. World Wide Web (WWW), 27(5):60. find
  16. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, Chujie Zheng, Dayiheng Liu, Fan Zhou, Fei Huang, Feng Hu, Hao Ge, Haoran Wei, Huan Lin, Jialong Tang, and 41 others. 2025. Qwen3 technical report.
  17. Aohan Zeng, Xin Lv, Qinkai Zheng, Zhenyu Hou, Bin Chen, Chengxing Xie, Cunxiang Wang, Da Yin, Hao Zeng, Jiajie Zhang, Kedong Wang, Lucen Zhong, Mingdao Liu, Rui Lu, Shulin Cao, Xiaohan Zhang, Xuancheng Huang, Yao Wei, Yean Cheng, and 80 others. 2025. GLM-4.5: agentic, reasoning, and coding (ARC) fou…
  18. Zeyu Zhang, Quanyu Dai, Xiaohe Bo, Chen Ma, Rui Li, Xu Chen, Jieming Zhu, Zhenhua Dong, and Ji-Rong Wen. 2025a. A survey on the memory mechanism of large language model-based agents. ACM Trans. Inf. Syst., 43(6). find
  19. Zeyu Zhang, Quanyu Dai, Xiaohe Bo, Chen Ma, Rui Li, Xu Chen, Jieming Zhu, Zhenhua Dong, and Ji-Rong Wen. 2025b. A survey on the memory mechanism of large language model-based agents. ACM Trans. Inf. Syst., 43(6). find
  20. Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, Yifan Du, Chen Yang, Yushuo Chen, Zhipeng Chen, Jinhao Jiang, Ruiyang Ren, Yifan Li, Xinyu Tang, Zikang Liu, and 3 others. 2023. A survey of large language models.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/qin-2026-implicitmembench-measuring-unconscious-behavioral-adaptation/