Identity, Cooperation and Framing Effects within Groups of Real and Simulated Humans

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Moon, Suhong, Kang, Minwoo, Suh, Joseph, Safdari, Mustafa, Canny, John
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915763913228288
author Moon, Suhong
Kang, Minwoo
Suh, Joseph
Safdari, Mustafa
Canny, John
author_facet Moon, Suhong
Kang, Minwoo
Suh, Joseph
Safdari, Mustafa
Canny, John
contents Humans act via a nuanced process that depends both on rational deliberation and also on identity and contextual factors. In this work, we study how large language models (LLMs) can simulate human action in the context of social dilemma games. While prior work has focused on "steering" (weak binding) of chat models to simulate personas, we analyze here how deep binding of base models with extended backstories leads to more faithful replication of identity-based behaviors. Our study has these findings: simulation fidelity vs human studies is improved by conditioning base LMs with rich context of narrative identities and checking consistency using instruction-tuned models. We show that LLMs can also model contextual factors such as time (year that a study was performed), question framing, and participant pool effects. LLMs, therefore, allow us to explore the details that affect human studies but which are often omitted from experiment descriptions, and which hamper accurate replication.
format Preprint
id arxiv_https___arxiv_org_abs_2601_16355
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Identity, Cooperation and Framing Effects within Groups of Real and Simulated Humans
Moon, Suhong
Kang, Minwoo
Suh, Joseph
Safdari, Mustafa
Canny, John
Computation and Language
Humans act via a nuanced process that depends both on rational deliberation and also on identity and contextual factors. In this work, we study how large language models (LLMs) can simulate human action in the context of social dilemma games. While prior work has focused on "steering" (weak binding) of chat models to simulate personas, we analyze here how deep binding of base models with extended backstories leads to more faithful replication of identity-based behaviors. Our study has these findings: simulation fidelity vs human studies is improved by conditioning base LMs with rich context of narrative identities and checking consistency using instruction-tuned models. We show that LLMs can also model contextual factors such as time (year that a study was performed), question framing, and participant pool effects. LLMs, therefore, allow us to explore the details that affect human studies but which are often omitted from experiment descriptions, and which hamper accurate replication.
title Identity, Cooperation and Framing Effects within Groups of Real and Simulated Humans
topic Computation and Language
url https://arxiv.org/abs/2601.16355