Human behavior is intrinsically shaped by the continuous and reciprocal interplay between emotions and values. Recognizing this interconnection, we argue that the integration of both emotions and values into multi-agent systems can significantly enhance their reasoning and adaptability in complex social environments, particularly in human-agent interactions. This viewpoint paper presents a critical perspective on existing approaches to emotion-value integration in artificial agents, drawing on a selection of representative works to support the argument. Rather than providing a systematic review, we reflect on how most current models treat emotions and values in isolation or with limited interaction, which constrains the agents’ ability to simulate rich human-like behaviors. We identify key conceptual and technical limitations, and outline open challenges to guide future research—emphasizing the need for unified cognitive architectures, deeper functional integration, and learning mechanisms that jointly consider emotional experiences and value systems.