Abstract: Emotion recognition technology faces a fundamental challenge:the lack of consensus on the definition of emotion. In current Western philosophy of mind, there are three main types of views on emotion:realism, constructivism, and mutual constitution theory. Evidence suggests that two standpoints merit particular attention in affective computing:radical enactivism and the theory of predictive mind-constructed emotion. The latter holds that an instance of emotion is a psychological event, where the brain categorizes incoming sensory inputs as similar to past experiences through concepts. In contrast, radical enactivism argues that emotion is a behavioral response mode of organisms, arising from their direct bodily interaction with the environment. Recently, radical enactivism has raised questions about the theory of predictive mind. It argues that the theory of predictive mind based on representationalism cannot explain the Müller-Lyer illusion. Were this challenge valid, the definition of emotion within the theory of predictive mind-constructed emotion would likewise be undermined. However, analysis of both theories reveals that radical enactivism's critique is untenable. By emphasizing the active and dynamic nature of emotions, the theory of predictive mind-constructed emotion not only offers a philosophically promising definition of emotion but also enhances the real-time processing capabilities and accuracy in emotion recognition technology.