Monitor-Generate-Verify (MGV): Formalising Metacognitive Theory for Language Model Reasoning

TL;DR

The paper identifies a prefix dominance trap in current test-time reasoning architectures and proposes Monitor-Generate-Verify (MGV), a formal framework that adds explicit metacognitive monitoring prior to generation. Grounded in Flavell's metacognition and Nelson and Narens' metamemory, MGV preserves psychological structure while translating it into algorithmic constructs, including monitoring signals, thresholds, and memory evolution. Although no empirical validation is provided, the framework offers a vocabulary for diagnosing component failures and suggests concrete architectural directions, potentially harmonizing with resource-rational analyses. The work further aligns metacognitive theory with rational meta-reasoning literature, outlining future pathways to ground MGV in normative principles and to integrate it with meta-reasoning and meta-learning in language models. Overall, MGV serves as a diagnostic blueprint for augmenting LLM reasoning with pre-generation monitoring and iterative refinement guided by verification feedback.

Abstract

Test-time reasoning architectures such as those following the Generate-Verify paradigm, where a model iteratively refines or verifies its own generated outputs, prioritise generation and verification but exclude the monitoring processes that determine when and how reasoning should begin. This omission may contribute to the prefix dominance trap, in which models commit early to suboptimal reasoning paths and seldom recover, yielding roughly 20% accuracy loss. We address this architectural gap by proposing the Monitor-Generate-Verify (MGV) framework, a computational translation of Flavell's and Nelson and Narens' metacognitive theories that preserves their psychological detail. MGV extends the Generate-Verify paradigm by adding explicit monitoring that captures metacognitive experiences (from difficulty assessments to confidence judgements) before generation begins and refines future monitoring through verification feedback. Though we present no empirical validation, MGV provides a vocabulary for diagnosing component-level failures in reasoning systems, suggests specific architectural interventions for future designs, and identifies connections to resource-rational analysis that may ground its mechanisms in normative principles.