Skip to content

Learning Dynamics of Continual Learning: A Unified View of Data Attribution, Forgetting, and Plasticity Loss

Sep 2026 · arXiv.org · Vol abs/2609.33620 · 1 citation · 55 references
Computer Science

TL;DR

This work derives a token- and layer-wise decomposition of how learning from one token changes another prediction, and connects data attribution, forgetting, and plasticity loss as distinct regimes of the same evolving learning dynamics.

Abstract

Modern language models are likely to be updated throughout their lifetime rather than trained once and frozen. Each update therefore participates in a recurring cycle: decide which experience to learn from, understand what that update changes, and remain capable of learning from what comes next. We show that these challenges are governed by the same evolving update--behavior interaction. We derive a token- and layer-wise decomposition of how learning from one token changes another prediction. By separating the softmax force, shared readout geometry, and residual connections, it exposes two interaction channels and yields a forward-computable approximation. Following this interaction through time reveals a unified picture of continual adaptation. Positive interaction identifies useful experience; negative interaction produces either concentrated collision or accumulated erosion; over longer horizons, updates reshape the shared geometry mediating future learning signals, reducing their transmission. These predictions lead to effective data selection, mechanism-specific controls for interference, and a readout-based diagnostic of future learnability whose degradation predicts the benefit of restoring the readout. Across models and training regimes, the same local interaction thus explains both what an update changes now and how learning today changes what can be learned tomorrow. This view connects data attribution, forgetting, and plasticity loss as distinct regimes of the same evolving learning dynamics.

View source

Similar papers

#artificial intelligence Preprint Sep 2026

Learning Under Forgetting: Statistical Support-Selective Retention in Stochastic Training Dynamics

Prior work has shown that neural networks exhibit implicit biases toward low-complexity structure (e.g., spectral bias), memorization dynamics, and compression-like effects during training, but a unified dynamical account of selective retention remains incomplete. We propose Repeated Reinforcement with Persistent Forge...

Fu-Jie Gao, Zu-Yue Zhang, Gang Sun · 0 citations
Preprint Aug 2026

Continual-learning rules shape representational drift

Together, these results link representational drift to the stability--plasticity trade-off: its magnitude is shaped by the mechanism that protects old knowledge, and suppressing it can restrict future learning.

Yi-Kai Si, Shanshan Qin · 1 citation
Open access Sep 2026

The Reuse-and-Append Memory Principle: Application to Latent Cause Inference

The latent cause theory of memory modification provides a computational account of how the brain decides whether to update existing memories or form new ones, but leaves unspecified the neural mechanisms implementing this inference. We propose Reuse-and-Append Memory (RAM), a set of mechanistic principles that achieve...

Z. Ben Houidi, S. Gershman · 0 citations
#machine learning Preprint Sep 2026

Intrinsic-Extrinsic Coupling in Learning Dynamics

Together, these results make intrinsic-extrinsic coupling operational by connecting executable state geometry to continuation-conditioned value, matched interaction identification, and closed-loop coordination, while separating identified coupling from complete-policy performance.

Qin-You Wang · 0 citations

Related blog posts

MIT News · Artificial Intelligence Sep 29, 2026

Who we become when we talk to machines

Professor Sherry Turkle’s new book, “Artificial Intimacy,” offers a withering critique of chatbots and the antisocial dynamics she believes they encourage.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.