HumanLLM: Benchmarking and Improving LLM Anthropomorphism via Human Cognitive Patterns
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866911599940337664 |
|---|---|
| author | Wang, Xintao Yang, Jian Li, Weiyuan Xie, Rui Huang, Jen-tse Gao, Jun Huang, Shuai Kang, Yueping Gou, Yuanli Feng, Hongwei Xiao, Yanghua |
| author_facet | Wang, Xintao Yang, Jian Li, Weiyuan Xie, Rui Huang, Jen-tse Gao, Jun Huang, Shuai Kang, Yueping Gou, Yuanli Feng, Hongwei Xiao, Yanghua |
| contents | Large Language Models (LLMs) have demonstrated remarkable capabilities in reasoning and generation, serving as the foundation for advanced persona simulation and Role-Playing Language Agents (RPLAs). However, achieving authentic alignment with human cognitive and behavioral patterns remains a critical challenge for these agents. We present HumanLLM, a framework treating psychological patterns as interacting causal forces. We construct 244 patterns from $\sim$12,000 academic papers and synthesize 11,359 scenarios where 2-5 patterns reinforce, conflict, or modulate each other, with multi-turn conversations expressing inner thoughts, actions, and dialogue. Our dual-level checklists evaluate both individual pattern fidelity and emergent multi-pattern dynamics, achieving strong human alignment ($r=0.90$) while revealing that holistic metrics conflate simulation accuracy with social desirability. HumanLLM-8B outperforms Qwen3-32B on multi-pattern dynamics despite 4$\times$ fewer parameters, demonstrating that authentic anthropomorphism requires cognitive modeling -- simulating not just what humans do, but the psychological processes generating those behaviors. Our dataset, code, and model are available at:https://github.com/YJGoodbye2024/HumanLLM |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_10198 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | HumanLLM: Benchmarking and Improving LLM Anthropomorphism via Human Cognitive Patterns Wang, Xintao Yang, Jian Li, Weiyuan Xie, Rui Huang, Jen-tse Gao, Jun Huang, Shuai Kang, Yueping Gou, Yuanli Feng, Hongwei Xiao, Yanghua Computation and Language I.2.7 Large Language Models (LLMs) have demonstrated remarkable capabilities in reasoning and generation, serving as the foundation for advanced persona simulation and Role-Playing Language Agents (RPLAs). However, achieving authentic alignment with human cognitive and behavioral patterns remains a critical challenge for these agents. We present HumanLLM, a framework treating psychological patterns as interacting causal forces. We construct 244 patterns from $\sim$12,000 academic papers and synthesize 11,359 scenarios where 2-5 patterns reinforce, conflict, or modulate each other, with multi-turn conversations expressing inner thoughts, actions, and dialogue. Our dual-level checklists evaluate both individual pattern fidelity and emergent multi-pattern dynamics, achieving strong human alignment ($r=0.90$) while revealing that holistic metrics conflate simulation accuracy with social desirability. HumanLLM-8B outperforms Qwen3-32B on multi-pattern dynamics despite 4$\times$ fewer parameters, demonstrating that authentic anthropomorphism requires cognitive modeling -- simulating not just what humans do, but the psychological processes generating those behaviors. Our dataset, code, and model are available at:https://github.com/YJGoodbye2024/HumanLLM |
| title | HumanLLM: Benchmarking and Improving LLM Anthropomorphism via Human Cognitive Patterns |
| topic | Computation and Language I.2.7 |
| url | https://arxiv.org/abs/2601.10198 |