Phenomenal unity is a key property of consciousness but often poorly defined. This paper introduces model unity: a system has model unity when a single relational model covering the whole system is optimal; otherwise, unity may exist only at a higher level. As part of expected float entropy minimisation theory, relational models interpret system states and may explain why visual experience is unified and why separate people do not share unified consciousness. Four example investigations are given, along with a postulate that distills the foundations of EFE minimisation into a clear statement for evaluating whether it identifies a self-evident fundamental property of consciousness.
A mathematical theory of consciousness based on Expected Float Entropy (EFE) minimisation is further investigated. EFE is a version of Shannon Entropy parameterised by relationships. For systems with learning-induced bias, certain relationship parameter choices yield much lower EFE values, thereby defining relationships. In this context, a brain state acquires meaning through the relational content of associated experience. EFE minimisation is tested as an association learning process, and results support a close connection between such learning and the emergence of consciousness. The theory may explain how the brain defines conscious content up to relationship isomorphism.