An AI Emotion Generation Model Based on Functional Body Hypothesis and Sensory Sharing

With the rapid development of artificial intelligence technology, achieving natural and efficient human-computer interaction has become a key challenge. Giving AI real emotional ability is regarded as the core bottleneck towards higher-order intelligence. The existing AI emotional interaction mostly focuses on the imitation of human external performance, lacking the support of internal experience, and its essence is “pseudo-empathy.” Inspired by the “body marker hypothesis,” this paper proposes the “functional body hypothesis,” aiming to construct a new model of emotion generation based on the intrinsic homeostatic requirements of the system without relying on anthroposical simulation, and explore its application in the human-computer sensory sharing scenario. By establishing the cornerstone of value judgment, designing the virtual physiological system, and introducing the natural expression mechanism to construct the emotion generation model, the “sensory translator” architecture is proposed to solve the “mixed perception paradox” in human-computer sensory sharing. This unified framework provides a theoretical path beyond personification for AI emotion generation and lays an engineering foundation for bidirectional and credible human-computer empathy and integration.