Moral Value Transfer via Cross-Domain Cognitive Analogies

Current work (e.g., Thapa et al., 2025; Lazzari et al., 2024) grounds moral classification mainly in established ethical frameworks like MFT or classical moral theories. But what if we borrowed the conceptual richness of domains that routinely handle context-dependent, multi-factor classification—such as medical triage, legal risk scoring, or even ecological impact assessment? This research would design analogy-driven hybrid frameworks, mapping the “urgency,” “scope,” or “consequences” axes from such fields onto moral classification in LLMs. For example, a moral scenario could be scored not just on deontology or utilitarianism, but also on “moral urgency” (analogous to triage), “systemic impact,” or “ambiguity tolerance.” This approach could yield more nuanced, real-world-relevant moral reasoning, breaking free from the sometimes rigid boundaries of existing moral taxonomies. It synthesizes the “transferring conceptualizations analogously” heuristic with insights from both NLP and applied ethics.

References:

1. Evaluation of Ethical Decision Making in Large Language Models Across Classical Moral Frameworks. Bishal Thapa, Alden Duarte-Vasquez, Sahar Hooshmand, Heena Rathore (2025). International Conference on Artificial Intelligence Testing.
2. Explainable Moral Values: a neuro-symbolic approach to value classification. Nicolas Lazzari, S. D. Giorgis, Aldo Gangemi, V. Presutti (2024). ESWC Satellite Events.