mlsanityData to 14 Oct 2026

Evidence of an Emergent "Self" in Continual Robot Learning

arXiv 2603.24350 ↗First posted 25 Mar 2026cs.RO cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

A key challenge to understanding self-awareness has been a principled way of quantifying whether an intelligent system has a concept of a "self", and if so how to differentiate the "self" from other cognitive structures. We propose that the "self" can be isolated by seeking the invariant portion of cognitive process that changes relatively little compared to more rapidly acquired cognitive skills - because our self is the most persistent aspect of our experiences. We used this principle to analyze the cognitive structure of robots under two conditions: One robot learns a constant task, while a second undergoes continual learning under variable tasks. We find that robots subjected to continual learning develop an invariant subnetwork that is significantly more stable (p < 0.001) compared to the control, and that this subnetwork is also functionally important: preserving it aids adaptation while damaging it impairs performance. We validate this pattern across three different robots spanning locomotion and manipulation.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 29 of its 53 references that are in the map. First posted on arXiv 25 Mar 2026.

Builds on

29 of 53 references.

References · 29

  1. 2018Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actorcited by 4,885
  2. 2016Overcoming catastrophic forgetting in neural networkscited by 4,787
  3. 2019Similarity of Neural Network Representations Revisitedcited by 1,755
  4. 2016Understanding deep learning requires rethinking generalizationcited by 1,674
  5. 2016Progressive Neural Networkscited by 1,361
  6. 2014How transferable are features in deep neural networks?cited by 1,225
  7. 2019Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learningcited by 1,204
  8. 2022Git Re-Basin: Merging Models modulo Permutation Symmetriescited by 471
  9. 2017PathNet: Evolution Channels Gradient Descent in Super Neural Networkscited by 364
  10. 2018Quantifying Generalization in Reinforcement Learningcited by 352
  11. 2021The Role of Permutation Invariance in Linear Mode Connectivity of Neural Networkscited by 291
  12. 2020Towards Continual Reinforcement Learning: A Review and Perspectivescited by 274
17 more
  1. 2020Understanding the Role of Individual Units in a Deep Neural Networkcited by 255
  2. 2018A Study on Overfitting in Deep Reinforcement Learningcited by 174
  3. 2020Explainable Reinforcement Learning: A Surveycited by 141
  4. 2021Geometry of the Loss Landscape in Overparameterized Neural Networks: Symmetries and Invariancescited by 127
  5. 2021Continual World: A Robotic Benchmark For Continual Reinforcement Learningcited by 116
  6. 2018On the importance of single directions for generalizationcited by 99
  7. 2024Sharing Knowledge in Multi-Task Deep Reinforcement Learningcited by 44
  8. 2022A Survey of Explainable Reinforcement Learningcited by 31
  9. 2016Learning Shared Representations in Multi-task Reinforcement Learningcited by 15
  10. 2021Full-Body Visual Self-Modeling of Robot Morphologiescited by 8
  11. 2026AI Must Embrace Specialization via Superhuman Adaptable Intelligencecited by 8
  12. 2024Distilling Reinforcement Learning Policies for Interpretable Robot Locomotion: Gradient Boosting Machines and Symbolic Regressioncited by 5
  13. 2025Exploring Human-AI Conceptual Alignment through the Prism of Chesscited by 2
  14. 2021Continual Learning with Neuron Activation Importancecited by 1
  15. 2023Modular Boundaries in Recurrent Neural Networkscited by 1
  16. 2024Robot Body Schema Learning from Full-body Extero/Proprioception Sensorscited by 1
  17. 2024Robot Metabolism: Towards machines that can grow by consuming other machinescited by 1

Built on by

Not cited yet.