mlsanityData to 7 Oct 2026

Mind the gap: Challenges of deep learning approaches to Theory of Mind

arXiv 2203.16540 ↗First posted 30 Mar 2022cs.LG cs.AI q-bio.NCcited by 21

Not on the map. Why?

Abstract (arXiv)

Theory of Mind is an essential ability of humans to infer the mental states of others. Here we provide a coherent summary of the potential, current progress, and problems of deep learning approaches to Theory of Mind. We highlight that many current findings can be explained through shortcuts. These shortcuts arise because the tasks used to investigate Theory of Mind in deep learning systems have been too narrow. Thus, we encourage researchers to investigate Theory of Mind in complex open-ended environments. Furthermore, to inspire future deep learning systems we provide a concise overview of prior work done in humans. We further argue that when studying Theory of Mind with deep learning, the research's main focus and contribution ought to be opening up the network's representations. We recommend researchers use tools from the field of interpretability of AI to study the relationship between different network components and aspects of Theory of Mind.

Key facts. Cited by 21 papers in the map (data through 7 Oct 2026). Builds on 36 of its 116 references that are in the map. First posted on arXiv 30 Mar 2022.

Builds on

36 of 116 references.

References · 36

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2017Deep reinforcement learning from human preferencescited by 3,846
  4. 2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,014
  5. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
  6. 2017Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environmentscited by 1,519
  7. 2016Understanding intermediate layers using linear classifier probescited by 1,279
  8. 2018Scalable agent alignment via reward modeling: a research directioncited by 438
  9. 2021Recursively Summarizing Books with Human Feedbackcited by 287
  10. 2018Human-level performance in first-person multiplayer games with population-based deep reinforcement learningcited by 260
  11. 2018Techniques for Interpretable Machine Learningcited by 237
  12. 2018Machine Theory of Mindcited by 222
24 more
  1. 2017Learning with Opponent-Learning Awarenesscited by 218
  2. 2019MineRL: A Large-Scale Dataset of Minecraft Demonstrationscited by 210
  3. 2017Teacher-Student Curriculum Learningcited by 197
  4. 2019The Hanabi Challenge: A New Frontier for AI Researchcited by 194
  5. 2021Benchmarking the Spectrum of Agent Capabilitiescited by 193
  6. 2021Open-Ended Learning Leads to Generally Capable Agentscited by 177
  7. 2021Alignment of Language Agentscited by 167
  8. 2020Enhanced POET: Open-Ended Reinforcement Learning through Unbounded Invention of Learning Challenges and their Solutionscited by 103
  9. 2021MiniHack the Planet: A Sandbox for Open-Ended Reinforcement Learning Researchcited by 96
  10. 2014Deep Neural Networks Reveal a Gradient in the Complexity of Neural Representations across the Brain's Ventral Visual Pathwaycited by 93
  11. 2017Intrinsically Motivated Goal Exploration Processes with Automatic Curriculum Learningcited by 79
  12. 2018The Surprising Creativity of Digital Evolution: A Collection of Anecdotes from the Evolutionary Computation and Artificial Life Research Communitiescited by 78
  13. 2018Bayesian Action Decoder for Deep Multi-Agent Reinforcement Learningcited by 67
  14. 2019Ablation Studies in Artificial Neural Networkscited by 58
  15. 2020Imitating Interactive Intelligencecited by 55
  16. 2021Creating Multimodal Interactive Agents with Imitation and Self-Supervised Learningcited by 40
  17. 2021Tell me why! Explanations support learning relational and causal structurecited by 36
  18. 2020Emergent Reciprocity and Team Formation from Randomized Uncertain Social Preferencescited by 21
  19. 2021CX-ToM: Counterfactual Explanations with Theory-of-Mind for Enhancing Human Trust in Image Recognition Modelscited by 20
  20. 2021Learning Interpretation with Explainable Knowledge Distillationcited by 13
  21. 2021Theory of Mind for Deep Reinforcement Learning in Hanabicited by 11
  22. 2019Perspective Taking in Deep Reinforcement Learning Agentscited by 7
  23. 2019Modeling Theory of Mind in Multi-Agent Games Using Adaptive Feedback Controlcited by 3
  24. 2018Do deep reinforcement learning agents model intentions?cited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 1; 2023: 6; 2024: 8; 2025: 4; 2026: 2.

21 papers.

Sort:

By area:

  1. 2024NegotiationToM: A Benchmark for Stress-testing Machine Theory of Mind on Negotiation Surroundingcited by 49
  2. 2023Towards A Holistic Landscape of Situated Theory of Mind in Large Language Modelscited by 32
  3. 2023ToMChallenges: A Principle-Guided Dataset and Diverse Evaluation Tasks for Exploring Theory of Mindcited by 28
  4. 2025Re-evaluating Theory of Mind evaluation in large language modelscited by 23
  5. 2023Theory of Mind as Intrinsic Motivation for Multi-Agent Reinforcement Learningcited by 19
  6. 2023The feasibility of artificial consciousness through the lens of neurosciencecited by 18
  7. 2024A Notion of Complexity for Theory of Mind via Discrete World Modelscited by 16
  8. 2024Limits of Theory of Mind Modelling in Dialogue-Based Collaborative Plan Acquisitioncited by 10
  9. 2024Belief in the Machine: Investigating Epistemological Blind Spots of Language Modelscited by 5
  10. 2023Mixture of personality improved Spiking actor network for efficient multi-agent cooperationcited by 4
  11. 2024Enhancing Counterfactual Explanation Search with Diffusion Distance and Directional Coherencecited by 4
  12. 2024Improving Language Models for Emotion Analysis: Insights from Cognitive Sciencecited by 4
9 more
  1. 2025Towards Machine Theory of Mind with Large Language Model-Augmented Inverse Planningcited by 4
  2. 2025The Yokai Learning Environment: Tracking Beliefs Over Space and Timecited by 3
  3. 2022Robot Learning Theory of Mind through Self-Observation: Exploiting the Intentions-Beliefs Synergycited by 2
  4. 2024Self-correction is Not An Innate Capability in Language Modelscited by 2
  5. 2025EmoPerso: Enhancing Personality Detection with Self-Supervised Emotion-Aware Modellingcited by 2
  6. 2024Can AI and humans genuinely communicate?cited by 1
  7. 2026A formal definition and meta-model for a machine theory of mindcited by 1
  8. 2023From DDMs to DNNs: Using process data and models of decision-making to improve human-AI interactionscited by 0
  9. 2026Bias In, Bias Out? Finding Unbiased Subnetworks in Vanilla Modelscited by 0