The Semantic Sphere: A Meta-Structural Framework for Unified Theories of Meaning in Artificial Intelligence

ABSTRACT This article introduces the Semantic Sphere, a meta-structural conceptual framework designed to unify symbolic, subsymbolic, and dynamical approaches to semantic emergence in artificial intelligence. Current models of meaning—ranging from symbolic representations to deep learning, predictive processing, and formal ontologies—remain fragmented and lack a shared semantic geometry. The Semantic Sphere addresses this gap by defining a triadic conceptual topology in which meaning arises from structural coherence among admissible transformations, without relying on internal states or representational content. The framework integrates existing theories by interpreting them as local projections, constraints, or trajectories within a unified semantic space. This meta-structural perspective provides a coherent foundation for understanding emergent representations, grounding, multimodal integration, and stateless cognitive architectures. The Semantic Sphere thus offers a generalisable and unifying account of semantic organization, with implications for AGI design, explainability, and the theoretical foundations of artificial intelligence. The framework aims to provide a principled foundation for future research on semantic emergence and unified cognitive architectures.