Emotion AI Faces Epistemic Limits, Underscoring Affective Sovereignty.

Emotion-sensing AI is increasingly integrated into various technologies, from vehicles to dialogue agents, creating what is termed the "Affectosphere" where emotions are computed at a societal scale. A critical, underexplored question in this domain is who holds the ultimate authority in defining the meaning of an individual's emotion. This research approaches the question from an epistemological perspective, focusing on the structural limits of measurement. It defines a "meaning distribution" for emotion labels and decomposes its uncertainty. The study demonstrates that while emotion AI can assign confident labels and discriminate aggregate differences, it cannot adequately estimate the irreducible component of meaning for individual instances, even with sufficient annotators. This systematic divergence is termed the "epistemic gap." The core finding is that a device's high confidence does not equate to having recovered irrecoverable meaning. Based on this, and the normative premise that systems unable to recover a quantity in principle should not be treated as authoritative, the study derives the norm of "affective sovereignty." This norm states that the final interpretive authority over one's emotion is procedurally reserved for the experiencing subject. These results suggest that the design, evaluation, and regulation of emotion AI should prioritize the allocation of interpretive authority over mere accuracy maximization.