Generative AI in emotion and behavior analysis: a comparative study of public discourse on migration in Spain, Germany, and Russia (2024-2025)

Abstract

In this study, the authors examine the potential of generative artificial intelligence (AI) for advancing neurocommunication research through the comparative analysis of emotional and behavioral dynamics in migration discourse. Focusing on Spain, Germany, and Russia during the period 2024-2025, the research integrates large language models (LLMs) into a multimodal analytical framework to investigate how emotions and behavioral intentions are expressed, mediated, and culturally framed in public communication. The methodology combines two complementary tools: the GPT-5-based multimodal pipeline and the DeepSeek-V3 model. GPT-5 was employed for in-depth analysis of emotional tone, sentiment polarity, intensity, and narrative frames, as well as for scenario modeling of public opinion dynamics under varying rhetorical conditions. DeepSeek-V3, trained on data up to mid-2025, was used for verification, cross-linguistic comparison, and detection of cultural nuances in emotional framings. This dual approach enabled a more robust interpretation of data and reduced methodological bias. The datasets were compiled in Spanish, German, and Russian from social media, messengers, traditional and new media, forums, blogs, and thematic portals. They were preprocessed and normalized for analysis. Emotional reactions were examined across three respondent groups: citizens, migrants, and officials. Findings demonstrate that while similar emotional categories recur across contexts, the distribution and intensity of emotional responses vary significantly by country. In Spain, positive emotions dominate, reflecting humanitarian narratives and optimism about integration, with citizens expressing compassion and solidarity, migrants emphasizing gratitude and hope, and officials highlighting cooperation and international reputation. Germany is characterized by ambivalence: citizens reveal both fears of cultural erosion and recognition of migrants’ contributions, migrants oscillate between hope and experiences of discrimination, and officials combine pragmatic optimism with concerns over resources and bureaucratic obstacles. In Russia, neutral and cautious emotions prevail, with citizens showing high levels of anxiety about cultural identity and crime, migrants emphasizing gratitude alongside adaptation difficulties, and officials adopting a restrained, regulation-centered rhetoric. The comparative perspective confirms that emotions are not merely individual reactions but function as communicative mechanisms shaping public opinion and behavior. By situating emotional and behavioral patterns within cultural and institutional contexts, generative AI contributes to neurocommunication research by modeling how discourse influences perception, decision-making, and social practices.