Cross-Cultural Simulation of Citizen Emotional Responses to Bureaucratic Red Tape Using LLM Agents

Wanchun Ni; Jiugeng Sun; Yixian Liu; Mennatallah El-Assady

cs.AIApr 14, 2026

Cross-Cultural Simulation of Citizen Emotional Responses to Bureaucratic Red Tape Using LLM Agents

Authors: Wanchun Ni, Jiugeng Sun, Yixian Liu, Mennatallah El-Assady

Organizations: ETH Zurich Zurich, Switzerland · City University of Hong Kong Hong Kong SAR, China

Abstract

Improving policymaking is a central concern in public administration. Prior human subject studies reveal substantial cross-cultural differences in citizens' emotional responses to red tape during policy implementation. While LLM agents offer opportunities to simulate human-like responses and reduce experimental costs, their ability to generate culturally appropriate emotional responses to red tape remains unverified. To address this gap, we propose an evaluation framework for assessing LLMs' emotional responses to red tape across diverse cultural contexts. As a pilot study, we apply this framework to a single red-tape scenario. Our results show that all models exhibit limited alignment with human emotional responses, with notably weaker performance in Eastern cultures. Cultural prompting strategies prove largely ineffective in improving alignment. We further introduce RAMO, an interactive interface for simulating citizens' emotional responses to red tape and for collecting human data to improve models. The interface is publicly available at https://ramo-chi.ivia.ch.

Figures & tables

Model	Default		Cultural-aware
	Overlap@3	SAS	Overlap@3	SAS $\uparrow$
GPT-4o	-	-0.12	-	-0.25
GPT-5-2025-08-07	-	-0.50	-	-0.12
Gemini-3-pro	-	-0.11	-	0.00
Qwen3-max-2026-01-23	-	-0.13	-	-0.25

Table 1. Alignment between LLM-simulated and human emotions in Hong Kong SAR. For culture-aware agents, higher SAS means more alignment with human results in significance.

Model	Default		Cultural-aware
	Overlap@3	SAS	Overlap@3 $\uparrow$	SAS $\uparrow$
GPT-4o	0.16	$-0.0038$	0.18	-0.0038
GPT-5-2025-08-07	0.20	$-0.1338$	0.20	$-0.1263$
Gemini-3-pro	0.42	$-0.1263$	0.10	$-0.2463$
Qwen3-max-2026-01-23	0.20	$-0.1338$	0.20	$-0.1263$

Table 2. Alignment between LLM-predicted and human emotions in Mainland China. Higher Overlap@3 and higher Significance Alignment Score (SAS) indicate better alignment with human results. (For Mainland China, SAS favors an intermediate significance outcome; see Appendix A.2 .)

Model	Default		Cultural-aware
	Overlap@3	SAS	Overlap@3 $\uparrow$	SAS $\uparrow$
GPT-4o	1.00	0.22	1.00	0.22
GPT-5-2025-08-07	0.93	0.56	1.00	0.56
Gemini-3-pro	0.25	0.00	0.50	0.00
Qwen3-max-2026-01-23	0.50	0.33	0.50	0.11

Table 3. Alignment between LLM-predicted and human emotions in Germany. Higher Overlap@3 and

SAS

indicate better alignment.

Figure 1. Overview of RAMO interface design, detailed in Section 5 . Major components are enlarged for visibility and labelled with circled letters for reference. The interface is designed for policymakers to simulate potential emotional responses to various red-tape scenarios across cultures.

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Figure 2. Cultural factor scores of Mainland China, Germany and Hong Kong SAR. Bar chart (or table-style graphic) comparing Hofstede cultural factor scores for Mainland China, Germany, and Hong Kong SAR across multiple dimensions.

Large Language Model-Based Social Simulation

Large language model-based social simulation studies how language-model agents represent people and interact to produce social behavior.

See all 257 papers

Expressing Social Emotions: Misalignment Between LLMs and Human Cultural Emotion Norms
Apr 18, 2026Sree Bhattacharyya, Manas Mehta, Leona Chen +4LLM Evaluation Cultural Bias in Language Models
The Pennsylvania State University · Duke University · Northeastern University +1
Improving Cross-Cultural Survey Simulation with Calibrated Value Personas
May 15, 2026Axel Abels, Elias Fernandez Domingos, Apurva Shah +1Language Model Calibration Cultural Bias in Language Models
1Machine Learning Group, Universit´e Libre de Bruxelles · 2AI Lab, Vrije Universiteit Brussel · 3FARI Institute, Universit´e Libre de Bruxelles - Vrije Universiteit Brussel +2
LLMs Infer Cultural Context but Fail to Apply It When Responding
Jun 16, 2026Yisong Miao, Jian Zhu, Vered ShwartzPragmatic Reasoning in Language Models Cultural Bias in Language Models
National University of Singapore · University of British Columbia · Canada CIFAR AI Chair, Vector Institute

Apr 18, 2026J/K move · Enter open · S save

Expressing Social Emotions: Misalignment Between LLMs and Human Cultural Emotion Norms

Sree Bhattacharyya, Manas Mehta, Leona Chen, Cristina Salvador+3

The Pennsylvania State University · Duke University · Northeastern University+1

LLM Evaluation Cultural Bias in Language Models Language Model Bias Evaluation

Open paper AskPDF arXiv

The expression of emotions that serve social purposes, such as asserting independence or fostering interdependence, is central to human interactions and varies systematically across cultures. As LLMs are increasingly used to simulate human behavior in culturally nuanced interactions, it is important to understand whether they faithfully capture human patterns of social emotion expression. When LLM responses are not culturally aligned, their utility is compromised -- particularly when users assume they are interacting with a culturally attuned interlocutor, and may act on advice that proves inappropriate in their cultural context. We present a psychologically informed evaluation framework of cross-cultural social emotion expression in LLMs. Using a human study comparing European American and Latin American participants' expression of engaging and disengaging emotions, we evaluate six frontier LLMs on their ability to reflect culturally differentiated patterns for expressing social emotions. We find systematic misalignment between model and human behavior: all models express engaging emotions more than disengaging ones, with particularly stark differences observed for the generally well-represented European American persona. We further highlight that LLM responses are highly concentrated and deterministic, failing to capture the diversity of human responses in expressing social emotions. Our ablation analyses reveal that these patterns are robust to sampling temperatures, partially sensitive to prompt language, and dependent on the response elicitation format. Together, our findings highlight limitations in how current LLMs represent the interaction of cultural and emotional axes, particularly when expressing social emotions, with direct implications for their deployment in cross-cultural affective contexts.

Abstract

Large Language Model-Based Social Simulation

Explore similar work

Expressing Social Emotions: Misalignment Between LLMs and Human Cultural Emotion Norms

Improving Cross-Cultural Survey Simulation with Calibrated Value Personas

LLMs Infer Cultural Context but Fail to Apply It When Responding