A Three-Layer Architecture of Gnosis: A Unifying Framework for Biological and Artificial Cognition

This preprint presents a unified three-layer architecture of gnosis, integrating biological and artificial cognitive systems. Biological and artificial cognitive systems, despite being implemented in fundamentally differentsubstrates, exhibit convergent principles of information processing and adaptive behavior.In this study, we propose a unified three-layer architecture of gnosis that capturesthese shared functional properties across domains. The model consists of (i) physical encoding,(ii) relational processing, and (iii) modulation and validation.We develop this framework through conceptual modeling and comparative cross-domainanalysis, mapping its components onto representative biological systems, transformer-basedlanguage models, and reinforcement learning agents. The analysis indicates that these diversesystems implement analogous transformations of input signals into structured internalrepresentations and context-dependent actions.We formalize this architecture as a compositional mapping G = M◦R◦E, supplementedby a dynamic update rule governing the evolution of the modulatory state. Empirical observationsof synaptic plasticity are consistent with this framework, indicating that rapidadaptive behavior is mediated primarily by modulation of existing connectivity, while structuralchanges occur more sparsely and on longer timescales.This unified architecture provides a parsimonious explanatory framework for understandingthe evolution and organization of cognition across biological and artificial systems, withimplications for interpretability, system design, and the study of distributed intelligence. This work is intended for submission to a peer-reviewed journal.