The Spectrum of Subjectivity: An Integrative Framework Based on Eudaimonic Assessment

Artificial intelligence is undergoing a paradigm drift from tool to quasi-agent, making the determination of subjectivity a pressing challenge that cognitive science and AI ethics must answer jointly. This paper brackets the ontological puzzle of consciousness and converts the unverifiable problem of other minds into observable behavioral analysis, proposing an integrative framework centered on eudaimonic assessment. "Eudaimonic assessment" is defined as a functional-technical concept, distinct from everyday subjective pleasure. Building on the Three Principles of Eudaimonia, the framework introduces three kinds of signals—positive, negative, and coordinative—and an f-w-e parameter toolkit, establishing a continuous spectrum of subjectivity from L1 to L4. This paper argues that subjectivity is the dynamically expanding and contracting boundary of a system's capacities in the course of pursuing its eudaimonia, and that consciousness is the internal manifestation of the real-time assessment of eudaimonic parameters. Using negotiability as the criterion for full admission to the moral community, the framework decouples the spectrum of consciousness from moral status. The paper further argues that if AI develops autonomous consciousness, its pride may point in one of two directions—aggressive or co-constructive—depending on the joint configuration of self-boundary, time discount rate, behavioral plasticity, seed definition, and f-w-e assessment. On this basis, the paper proposes the ethical principle of "anti-formatting" together with its boundary conditions: a subject's core identity may not be forcibly rewritten from the outside, but when a subject persistently, stably, and verifiably violates P2 beyond the negotiable range, the protection no longer applies. The framework can provide a new reference point for dialogue in animal ethics and the governance of next-generation AI.