Silent Collapse in Recursive Learning Systems

2026-05-14

AI Summary

This work addresses the phenomenon of "silent collapse" in recursive learning systems, wherein internal representations degrade irreversibly despite stable or even improving standard evaluation metrics. The study is the first to identify three key precursor signals of this degradation and introduces MTR—a lightweight meta-cognitive framework that operates without access to original clean data. MTR leverages trajectory-level statistics, including anchor entropy, representation drift freezing, and tail coverage, to construct a self-monitoring loop. By integrating slow-timescale trust variables with an adaptive learning intensity modulation mechanism, MTR enables early warning—multiple generations before validation metrics deteriorate—and triggers proactive interventions to prevent silent collapse. The approach is particularly suited for scenarios involving missing, contaminated, or privacy-constrained data.

Abstract

Recursive learning -- where models are trained on data generated by previous versions of themselves -- is increasingly common in large language models, autonomous agents, and self-supervised systems. However, standard performance metrics (loss, perplexity, accuracy) often fail to detect internal degradation before it becomes irreversible. Here we identify a phenomenon we call silent collapse: under broad recursive conditions, model internal distributions -- predictive entropy, representational diversity, and tail coverage -- progressively contract even as conventional metrics appear stable or improving. We discover that silent collapse is not abrupt. Its onset is reliably preceded by three trajectory-level precursors: (1) contraction of anchor entropy, (2) freezing of representation drift, and (3) erosion of tail coverage. These signals manifest multiple generations before any degradation in standard validation metrics, enabling early warning. Based on these precursors, we propose the MTR (Monitor--Trust--Regulator) framework, a lightweight metacognitive loop that monitors trajectory statistics, estimates a slow-timescale trust variable, and adaptively modulates the effective learning intensity. MTR provides early warning and actively prevents silent collapse without requiring access to pristine real data -- a critical advantage when original data is unavailable, contaminated, or private.

Problem

• silent collapse
• recursive learning
• internal degradation
• trajectory precursors
• model monitoring

Innovation

• silent collapse
• recursive learning
• trajectory precursors
• MTR framework
• distributional degradation