Reproduce “Enhancing LLMs with Cognitive-Affective Personality Inference for Simulating Human Social-Psychological Behavior” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhibo Deng, Dongyuan Li, Shuwen Ge, Ziqing Zhang, Ying Zhang, Renhe Jiang
SPIN simulates participants by separating a reusable core from a condition-specific state that informs each decision. Across two reconstructed study families and four base LLMs, it is evaluated by comparing outputs with human behavioral summaries. SPIN ranks first on both family headline scores for all four models; DeepSeek V3.2 averages 0.869 versus 0.770 for its best baseline. These scores measure aggregate alignment, not individual psychology.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.