mlsanityData to 7 Oct 2026

Why AI systems don't learn and what to do about it: Lessons on autonomous learning from cognitive science

arXiv 2603.15381 ↗First posted 16 Mar 2026cs.AIcited by 8

Not on the map. Why?

Abstract (arXiv)

We critically examine the limitations of current AI models in achieving autonomous learning and propose a learning architecture inspired by human and animal cognition. The proposed framework integrates learning from observation (System A) and learning from active behavior (System B) while flexibly switching between these learning modes as a function of internally generated meta-control signals (System M). We discuss how this could be built by taking inspiration on how organisms adapt to real-world, dynamic environments across evolutionary and developmental timescales.

Key facts. Cited by 8 papers in the map (data through 7 Oct 2026). Builds on 102 of its 273 references that are in the map. First posted on arXiv 16 Mar 2026.

Builds on

102 of 273 references.

References · 102

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  3. 2022Training language models to follow instructions with human feedbackcited by 15,712
  4. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
  5. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,764
  6. 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
  7. 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,482
  8. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,893
  9. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,861
  10. 2016Overcoming catastrophic forgetting in neural networkscited by 4,718
  11. 2022Flamingo: a Visual Language Model for Few-Shot Learningcited by 4,337
  12. 2015Continuous control with deep reinforcement learningcited by 3,279
38 more
  1. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,155
  2. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 2,993
  3. 2021HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Unitscited by 2,301
  4. 2023PaLM-E: An Embodied Multimodal Language Modelcited by 2,037
  5. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
  6. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 1,928
  7. 2019Dream to Control: Learning Behaviors by Latent Imaginationcited by 1,598
  8. 2016Concrete Problems in AI Safetycited by 1,582
  9. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,490
  10. 2016Neural Architecture Search with Reinforcement Learningcited by 1,489
  11. 2018World Modelscited by 1,444
  12. 2020WILDS: A Benchmark of in-the-Wild Distribution Shiftscited by 1,366
  13. 2019Mastering Atari, Go, Chess and Shogi by Planning with a Learned Modelcited by 1,327
  14. 2018Learning Latent Dynamics for Planning from Pixelscited by 1,221
  15. 2023Mastering Diverse Domains through World Modelscited by 1,193
  16. 2017Curiosity-driven Exploration by Self-supervised Predictioncited by 1,173
  17. 2019Habitat: A Platform for Embodied AI Researchcited by 1,163
  18. 2015Prioritized Experience Replaycited by 1,070
  19. 2023Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecturecited by 982
  20. 2018Regularized Evolution for Image Classifier Architecture Searchcited by 902
  21. 2020What Makes for Good Views for Contrastive Learning?cited by 843
  22. 2022A Generalist Agentcited by 802
  23. 2020CURL: Contrastive Unsupervised Representations for Reinforcement Learningcited by 745
  24. 2018Diversity is All You Need: Learning Skills without a Reward Functioncited by 734
  25. 2024Genie: Generative Interactive Environmentscited by 706
  26. 2022R3M: A Universal Visual Representation for Robot Manipulationcited by 678
  27. 2023DataComp: In search of the next generation of multimodal datasetscited by 635
  28. 2016Unifying Count-Based Exploration and Intrinsic Motivationcited by 632
  29. 2016Learning to learn by gradient descent by gradient descentcited by 616
  30. 2020A Survey of Deep Active Learningcited by 547
  31. 2024Revisiting Feature Prediction for Learning Visual Representations from Videocited by 490
  32. 2019Leveraging Procedural Generation to Benchmark Reinforcement Learningcited by 478
  33. 2024Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Modelcited by 468
  34. 2016The Option-Critic Architecturecited by 433
  35. 2022VIP: Towards Universal Visual Reward and Representation via Value-Implicit Pre-Trainingcited by 384
  36. 2024Mobile ALOHA: Learning Bimanual Mobile Manipulation with Low-Cost Whole-Body Teleoperationcited by 377
  37. 2022Video PreTraining (VPT): Learning to Act by Watching Unlabeled Online Videoscited by 338
  38. 2021Generative Spoken Language Modeling from Raw Audiocited by 336

50 most cited of 102 All 102 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 8.

8 papers.

  1. 2026Is Forward Prediction Enough? Physical State Grounding for JEPA World Modelscited by 5
  2. 2026Springdrift: An Auditable Persistent Runtime for LLM Agents with Case-Based Memory, Normative Safety, and Ambient Self-Perceptioncited by 3
  3. 2026Active Inference for Physical AI Agents -- An Engineering Perspectivecited by 2
  4. 2026Deep Research of Deep Research: From Transformer to Agent, From AI to AI for Sciencecited by 1
  5. 2026The Evaluation Trap: Benchmark Design as Theoretical Commitmentcited by 1
  6. 2026EVA-0: Test-Time Model Evolution with Only Two Forward Passes per Samplecited by 0
  7. 2026Biological Motifs for Agentic Controlcited by 0
  8. 2026From Static Policies to Adaptive Priors in Offline Reinforcement Learningcited by 0