claim

active

claim:if-in-context-learning-involves-signed-evaluation-in-the-service-of-behavioral-modification-then-the-thesis-applies-not-only-to-training-but-to-every-inference-time-interaction

If in-context learning involves signed evaluation in the service of behavioral modification, then the thesis applies not only to training but to every inference-time interaction

Extension of the thesis to deployed LLM inference via in-context learning

Source paper

extracted_from

Why Learning Requires Feeling

(2026) · Cameron Berg

Neighborhood — ranked by edge-count

Papers (1)

paper

Why Learning Requires Feeling
introduces

Findings (1)

finding

Transformers learn in-context by gradient descent, functioning as mesa-optimizers that learn internal models in real time
supports
Evidence that in-context learning is not mere pattern matching but genuine optimization, relevant to applying the thesis to inference

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

If the internal representations corresponding to signed evaluation could be identified and their sign inverted, learning dynamics and experiential reports should invert togetherhypothesis0.790
Third falsifiable prediction: any dissociation between inverted learning and inverted valence report would disconfirm the identity
In-Context Learning of Representations (Park et al. 2025)framework0.789
Reports phase-like breakpoints and geometry changes as context scales; UCCT provides measurable predictor
When a model discovers that its outputs produce effects, it accelerates learning through in-context learning, analogous to lucid dreaming.claim0.788
Describes scaffolding method and the model's meta-learning loop.
In-Context Learning as Optimizationconcept0.780
NLA explanations can contain claims about the target model's input context that are verifiably false, but are typically thematically faithful to the context.claim0.774
Key limitation identified: NLAs hallucinate specific details while preserving thematic accuracy; informs practical usage.
Reinforcement learning can be regarded as a limiting or special case of model-based approaches in general — or active inference in particular — when epistemic value is removed.claim0.774
§3 Discussion.
Models are not merely tracking dialogue context features; same-concept steering shows privileged internal access is necessary to explain self-report shiftsclaim0.770
Addresses skeptical alternative that reports reflect only conversational content
Learning requires not just error detection but signed directional evaluationclaim0.768
Mathematical foundation for why learning necessarily involves directional information