Intrinsic-Extrinsic Coupling in Learning Dynamics
Not on the map. Why?
Abstract (arXiv)
A learner's current observations need not determine its response to further training. We formulate intrinsic-extrinsic coupling through the continuation-conditioned value of a constrained learning-state intervention, with observation-relative fibers describing present agreement. An executable finite-frame classifier-head write protects current logits while repairing specified historical margins under finite-precision acceptance checks. We distinguish local admissibility, continuation-conditioned intervention value, and complete-policy performance. A matched four-cell contrast identifies readout-specific non-additivity between the same intrinsic intervention and alternative external continuations. In a CLINC-derived class-incremental setting, replay changes the write's 32-update contribution from five correct predictions to zero. Nonzero interactions also occur under output distillation, with a RoBERTa backbone, and under optimizer-native SGDW dynamics. Under SGDW, correct-count interactions are negative in all three activated roots at 128 updates, showing that coupling need not imply positive synergy. The mathematical analysis distinguishes feasible local repairs and favorable terminal outputs from training-reachable repair regions. Separate coordination tests show that content controls match or exceed the development gain, while a five-root fresh-test comparison with Fiber present in every arm shows root-dependent rather than uniformly beneficial correct-count effects. On the secondary cross-entropy readout, guided allocation yields lower mean loss than standard replay in all five pairs. 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.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 22 of its 32 references that are in the map. First posted on arXiv 24 Sep 2026.
Builds on
22 of 32 references.
References · 22
- 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
- 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,266
- 2016Overcoming catastrophic forgetting in neural networkscited by 4,718
- 2022Locating and Editing Factual Associations in GPTcited by 2,609
- 2018Neural Tangent Kernel: Convergence and Generalization in Neural Networkscited by 2,597
- 2016Learning without Forgettingcited by 2,465
- 2017Gradient Episodic Memory for Continual Learningcited by 1,871
- 2018Efficient Lifelong Learning with A-GEMcited by 1,044
- 2022Mass-Editing Memory in a Transformercited by 976
- 2015Train faster, generalize better: Stability of stochastic gradient descentcited by 547
- 2019An Evaluation Dataset for Intent Classification and Out-of-Scope Predictioncited by 378
- 2019Online Continual Learning with Maximally Interfered Retrievalcited by 376
10 more
- 2019Orthogonal Gradient Descent for Continual Learningcited by 361
- 2015Gradient-based Hyperparameter Optimization through Reversible Learningcited by 357
- 2021Gradient Projection Memory for Continual Learningcited by 338
- 2024AlphaEdit: Null-Space Constrained Knowledge Editing for Language Modelscited by 224
- 2026Path-conditioned training: a principled way to rescale ReLU neural networkscited by 6
- 2023Implicit biases in multitask and continual learning from a backward error analysis perspectivecited by 4
- 2025Retention analysis of edited knowledge after fine-tuningcited by 2
- 2026Can Neural Networks Learn by Experimenting on Themselves? Self-Interventional Learning from Functional Consequences to Predictive Self-Knowledgecited by 2
- 2026Fiber Fingerprints of Hidden Learning-State Dynamicscited by 2
- 2026Revelation Controlcited by 1
Built on by
Not cited yet.