We introduce a formal framework in which an agent's value system is modelled as a coherence structure over a finite set of propositions. Coherence maximization is shown to be NP-hard (by reduction from MAX-SAT), providing a complexity-theoretic foundation for why stable worldviews are difficult to construct yet robust once attained. Using Lyapunov stability analysis and replicator dynamics, we prove that maximally coherent configurations act as emergent attractors in belief-revision dynamics. The framework is applied to process ontologies (Whitehead, Spinoza) and agent-relative value theory, bridging formal epistemology, computational complexity, and philosophy of religion.
Mathias Leonhardt (Thu,) studied this question.