Tears or Cheers? Benchmarking LLMs via Culturally Elicited Distinct Affective Responses

2026-01-19

AI Summary

This work addresses the limitation of current cultural alignment evaluations for large language models, which predominantly focus on declarative knowledge and fail to capture cross-cultural differences in affective interpretation. To bridge this gap, the authors propose CEDAR, a multimodal benchmark that uniquely targets culture-induced fine-grained emotional responses. They construct a dataset comprising 10,962 samples across seven languages and fourteen emotion categories, leveraging a novel pipeline that integrates generative pre-annotation by large language models with multilingual human validation. Through systematic evaluation of seventeen prominent multilingual models, the study reveals a significant disconnect between linguistic consistency and culturally aligned emotional understanding, thereby challenging conventional assessment paradigms in cross-cultural AI alignment.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Cognitive Modeling & Cognitive Systems: Affective Computing

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Culture serves as a fundamental determinant of human affective processing and profoundly shapes how individuals perceive and interpret emotional stimuli. Despite this intrinsic link extant evaluations regarding cultural alignment within Large Language Models primarily prioritize declarative knowledge such as geographical facts or established societal customs. These benchmarks remain insufficient to capture the subjective interpretative variance inherent to diverse sociocultural lenses. To address this limitation, we introduce CEDAR, a multimodal benchmark constructed entirely from scenarios capturing Culturally \underline{\textsc{E}}licited \underline{\textsc{D}}istinct \underline{\textsc{A}}ffective \underline{\textsc{R}}esponses. To construct CEDAR, we implement a novel pipeline that leverages LLM-generated provisional labels to isolate instances yielding cross-cultural emotional distinctions, and subsequently derives reliable ground-truth annotations through rigorous human evaluation. The resulting benchmark comprises 10,962 instances across seven languages and 14 fine-grained emotion categories, with each language including 400 multimodal and 1,166 text-only samples. Comprehensive evaluations of 17 representative multilingual models reveal a dissociation between language consistency and cultural alignment, demonstrating that culturally grounded affective understanding remains a significant challenge for current models.

Problem

• cultural alignment
• affective responses
• cross-cultural emotion
• subjective interpretation
• multilingual models

Innovation

• Culturally Elicited Affective Responses
• Multimodal Benchmark
• Cross-cultural Emotion Recognition
• LLM Cultural Alignment
• Human Evaluation Pipeline