in-context learning (ICL)

Test-time adaptation from prompt or retrieved context with no parameter updates.

Neighborhood — ranked by edge-count

paper

framework

Active Inference
supports
Foundational framework by Karl Friston; the paper extends it to three hierarchical levels for modeling meta-awareness.
Unified Contextual Control Theory (UCCT)
extends
A theory that pretrained latent patterns are bound to task targets via external semantic anchors; formalized by anchoring strength S.

claim

Consciousness requires evaluation in the service of behavioral policy modification
associated_with
Restriction that distinguishes conscious evaluators from mere homeostatic regulators like thermostats

method

few-shot prompting
implements
Providing k labeled examples in the prompt to steer model behavior.
base model probing
implements
Method of using base models (no post-training) to observe spontaneous self-referential behaviors without confound of memorized introspection language.

concept

Induction Heads
associated_withimplements
Mechanistic circuits in transformers documented by Olsson et al. 2022, cited as evidence for pattern-repository assumption
semantic anchoring
associated_with
The central idea that external structure binds latent patterns to desired targets.
autoregressive recurrence
associated_with
Transformers are recurrent through autoregression because the K/V stream provides horizontal information flow across positions, even though each forward pass is feedforward.
Skip-Trigram
associated_with
A three-token pattern of the form [source]...[destination][out] that one-layer attention heads implement; the paper's key characterization of one-layer transformer behavior
Mesa-Optimizer
associated_with
A learned optimizer running inside a base optimizer; transformers proposed as mesa-optimizers implementing gradient descent in-context
Associative Plasticity
supports
Learning mechanism: parameter updates resemble classical Hebbian learning with associative and decay terms.

artifact

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.

In-Context Learning of Representations (Park et al. 2025)framework0.812
Reports phase-like breakpoints and geometry changes as context scales; UCCT provides measurable predictor
In-Context Learning as Optimizationconcept0.810
Meta-Learning View of ICL (Dai et al. 2023)framework0.791
Views ICL as a form of meta-learning; UCCT sits alongside this account
In-Context Learning and Induction Heads (forthcoming paper)concept0.776
A follow-up paper extending the framework and induction head concept to larger more realistic models
Causal Representation Learning (CRL)framework0.776
Schölkopf et al.'s framework combining representation learning with causal inference.
in-context learning tasks with complex graph geometriesconcept0.746
Tasks involving graph-structured geometries for in-context learning, used to test manifold steering.
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 interactionclaim0.746
Extension of the thesis to deployed LLM inference via in-context learning
self-model through in-context learningconcept0.746
The thesis that transformers develop a self-model via ICL, not only from training data; base models bootstrap self-referential reasoning.