The Dual Nature of Language: Metalanguage of Cognition and External Language of Meaning

This paper introduces a formal framework for understanding language in cognitive systems, distinguishing the Metalanguage of Cognition (MLC), an internal system of vector-based cognitive dynamics, from the External Language of Meaning (ELM), a symbolic system for communication. Rooted in Principia Cognitia (PC), which is grounded in a materialist ontology and distinguishing phenomena from their discrete signal representations, the framework defines MLC as a triple (S,R,O) of semions (vector representations of cognitive units), weighted relations, and operations, while ELM is a symbol set Σ linked via a mapping μ:S→Σ that transforms internal representations into external symbols, often losing structural detail. Grounded in the axiom that cognition is an activatable vector structure, this framework resolves philosophical challenges, such as Searle’s Chinese Room paradox, by showing that understanding requires MLC, not just ELM. Empirical evidence from transformer-based large language models (LLMs), where residual streams encode belief state geometries, validates semions as cognitive units. The MLC-ELM model unifies biological and artificial cognition, redefines rationality through metacognitive operations like detecting knowledge gaps (e.g., recognizing unfamiliar concepts), and informs AI design by prioritizing internal cognitive alignment. This substrate-neutral, testable framework advances cognitive science, neuroscience, AI, and philosophy.