Concepts Become Operational Only When Their Frameworks Are Activated: An Enactive Account of Conceptual Analysis in Large Language Model (LLM) Interaction

Abstract

This paper examines how conceptual coherence is achieved in interactions with large language models (LLMs). It argues that coherence does not arise from the presence of concepts alone, but depends on conceptual frameworks that coordinate two processes: the LLM’s generation of formal continuations and the user’s stabilization of interpretation. A case study shows that a framework can be developed within one interaction yet fail to be re-enacted in another, even when the same text is provided. The response remains coherent and preserves the relevant concepts, but reorganizes them under a different structure. This demonstrates that access to content is not sufficient for maintaining a line of thought. What is required is the activation of a framework that governs how that content is taken up and extended. The analysis develops an enactive method in which frameworks are not described from the outside but brought into operation within interaction. Conceptual frameworks are shown to function as sites of synchronization between formal continuity through recursion and interpretive stability through return. Concepts are accordingly redefined as constraints that become operative only within such frameworks. The operations of activation, stabilization, drift, and recalibration specify the conditions under which a framework is maintained or lost. The paper then addresses the problem of extending frameworks across contexts. It argues that frameworks do not persist between interactions, but can be reconstituted when their governing constraints are explicitly identified and articulated as invariants. Under these conditions, a framework becomes re-enterable: it can be re-instantiated across contexts while preserving its identity. The central result is that conceptual coherence in LLM-mediated interaction depends on the activation, stabilization, and re-entry of frameworks, rather than on content alone. A re-enterable framework functions as a concept at a higher level of abstraction, providing the basis for generalization across interactions.

Keywords