mlsanityData to 7 Oct 2026

In-context learning agents are asymmetric belief updaters

arXiv 2402.03969 ↗First posted 6 Feb 2024cs.LGcited by 13

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

Abstract (arXiv)

We study the in-context learning dynamics of large language models (LLMs) using three instrumental learning tasks adapted from cognitive psychology. We find that LLMs update their beliefs in an asymmetric manner and learn more from better-than-expected outcomes than from worse-than-expected ones. Furthermore, we show that this effect reverses when learning about counterfactual feedback and disappears when no agency is implied. We corroborate these findings by investigating idealized in-context learning agents derived through meta-reinforcement learning, where we observe similar patterns. Taken together, our results contribute to our understanding of how in-context learning works by highlighting that the framing of a problem significantly influences how learning occurs, a phenomenon also observed in human cognition.

Key facts. Cited by 13 papers in the map (data through 7 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 25 of its 57 references that are in the map. First posted on arXiv 6 Feb 2024.

Builds on

25 of 57 references.

References · 25

  1. 2017Attention Is All You Needcited by 59,932
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2023Sparks of Artificial General Intelligence: Early experiments with GPT-4cited by 2,579
  4. 2016Asynchronous Methods for Deep Reinforcement Learningcited by 2,218
  5. 2023Code Llama: Open Foundation Models for Codecited by 2,105
  6. 2023Voyager: An Open-Ended Embodied Agent with Large Language Modelscited by 2,081
  7. 2022Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?cited by 1,413
  8. 2021An Explanation of In-context Learning as Implicit Bayesian Inferencecited by 915
  9. 2022Transformers learn in-context by gradient descentcited by 714
  10. 2022What Can Transformers Learn In-Context? A Case Study of Simple Function Classescited by 700
  11. 2016RL2: Fast Reinforcement Learning via Slow Reinforcement Learningcited by 490
  12. 2023Large Language Models Are State-of-the-Art Evaluators of Translation Qualitycited by 416
13 more
  1. 2022Using cognitive psychology to understand GPT-3cited by 389
  2. 2023Larger language models do in-context learning differentlycited by 386
  3. 2016Learning to reinforcement learncited by 369
  4. 2022Data Distributional Properties Drive Emergent In-Context Learning in Transformerscited by 303
  5. 2023GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Modelscited by 174
  6. 2022Language models show human-like content effects on reasoning taskscited by 171
  7. 2019Explainable Machine Learning for Scientific Insights and Discoveriescited by 165
  8. 2019Meta-learning of Sequential Strategiescited by 68
  9. 2023Meta-in-context learning in large language modelscited by 59
  10. 2020Meta-trained agents implement Bayes-optimal agentscited by 37
  11. 2022Modeling Human Exploration Through Resource-Rational Reinforcement Learningcited by 13
  12. 2023In-Context Learning Dynamics with Random Binary Sequencescited by 13
  13. 2023How should the advent of large language models affect the practice of science?cited by 8

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 5; 2025: 6; 2026: 2.

13 papers.

Sort:

By area:

  1. 2024Can large language models explore in-context?cited by 75
  2. 2024Current state of LLM Risks and AI Guardrailscited by 44
  3. 2024Controlling Large Language Model Agents with Entropic Activation Steeringcited by 29
  4. 2024Sparse Autoencoders Reveal Temporal Difference Learning in Large Language Modelscited by 15
  5. 2025Generating Computational Cognitive Models using Large Language Modelscited by 12
  6. 2025Should You Use Your Large Language Model to Explore or Exploit?cited by 8
  7. 2025Stop Reducing Responsibility in LLM-Powered Multi-Agent Systems to Local Alignmentcited by 7
  8. 2024Large Language Models are Biased Reinforcement Learnerscited by 5
  9. 2025Cognitive models can reveal interpretable value trade-offs in language modelscited by 3
  10. 2025Large Language Models as Search Engines: Societal Challengescited by 3
  11. 2025Meta-learning ecological priors from large language models explains human learning and decision makingcited by 2
  12. 2026Reasoning aligns language models to human cognitioncited by 2
1 more
  1. 2026Conflict and Congruency Effects in Large Language Models: In-Weight and In-Context Competition in a Verbal Conflict Taskcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026In-Context Reinforcement Learning under Non-Stationarity: A Survey
  2. 2025Do Large Language Models Know What They Are Capable Of?
  3. 2025Reward Is Enough: LLMs Are In-Context Reinforcement Learners
  4. 2025Accumulating Context Changes the Beliefs of Language Models
  5. 2026Unrewarded Exploration in Large Language Models Reveals Latent Learning from Psychology
  6. 2026When Does Continual Learning Require Learning
  7. 2026Dynamic Theory of Mind as a Temporal Memory Problem: Evidence from Large Language Models
  8. 2025The Path of Self-Evolving Large Language Models: Achieving Data-Efficient Learning via Intrinsic Feedback