Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Theory-of-mind social reasoning
Models mental states and social intelligence in agents for human interaction and grounded language.
69 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Human-AI and goal-conditioned, keyword cluster “communication · ai · collaboration”
State History Tried Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
16 papers in 2023–24 and 18 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 5.4% to 4.7%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Instructable Intelligent Personal Agent (AAAI 2016 · 74 citations)
- 2017 · Learning through Dialogue Interactions by Asking Questions (ICLR 2017 · 81 citations)
- 2018 · Hybrid Retrieval-Generation Reinforced Agent for Medical Image Report Generation (NeurIPS 2018 · 228 citations)
- 2019 · BabyAI: A Platform to Study the Sample Efficiency of Grounded Language Learning (ICLR 2019 · 66 citations)
- 2020 · That and There: Judging the Intent of Pointing Actions with Robotic Arms (AAAI 2020 · 10 citations)
- 2021 · Watch-And-Help: A Challenge for Social Perception and Human-AI Collaboration (ICLR 2021 · 43 citations)
- 2022 · Fusing Task-Oriented and Open-Domain Dialogues in Conversational Agents (AAAI 2022 · 51 citations)
- 2023 · Help Me Heal: A Reinforced Polite and Empathetic Mental Health and Legal Counseling Dialogue System for Crime Victims (AAAI 2023 · 16 citations)
- 2024 · Cooper: Coordinating Specialized Agents towards a Complex Dialogue Goal (AAAI 2024 · 8 citations)
- 2025 · MuMA-ToM: Multi-modal Multi-Agent Theory of Mind (AAAI 2025 · 7 citations)
- 2026 · Dimension-Aware Active Annotation for Aesthetic Perception via Multi-Agent Human–AI Collaboration (AAAI 2026 · 1 citation)
Earliest papers here
- Instructable Intelligent Personal Agent (AAAI 2016 · 74 citations)
- Learning through Dialogue Interactions by Asking Questions (ICLR 2017 · 81 citations)
- Dialogue Learning With Human-in-the-Loop (ICLR 2017 · 46 citations)
What has been tried
51 papers here report 467 measured changes (a component added, removed or swapped, compared with a baseline): helped 69%, no clear effect 9%, hurt 18%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add global coordination to rank and filter topic candidates | ESConv · BL-1 | 19.73 -> 22.76 | Helped | Cooper: Coordinating Specialized Agents towards a Complex Dialogue Goal (AAAI 2024 · 8 citations) |
| add inverse planning | MuMA-ToM all questions · accuracy | 55.3 -> 76.6 | Helped | MuMA-ToM: Multi-modal Multi-Agent Theory of Mind (AAAI 2025 · 7 citations) |
| replace uniform proposals with neural recognition-network data-driven proposals (amortized inference) | Construction · goal inference accuracy | Helped | Neural Amortized Inference for Nested Multi-Agent Reasoning (AAAI 2024 · 4 citations) | |
| Add Stage 1 mental-state reasoning | Social cognition tasks · accuracy (AVG.) | 77.9 -> 80.5 | Helped | MetaMind: Modeling Human Social Thoughts with Metacognitive Multi-Agent Systems (NeurIPS 2025 · 3 citations) |
| learn participant-specific prospect-theory parameters instead of using shared parameters | Human model approximation, Rounds 2–4 · Spearman rank correlation distribution in [0.75, 1] | Round 2: 0.353 -> 0.375; Round 3: 0.359 -> 0.391; Round 4: 0.337 -> 0.408 | Helped | Does Your AI Agent Get You? A Personalizable Framework for Approximating Human Models from Argumentation-based Dialogue Traces (AAAI 2025 · 2 citations) |
| Remove the BDI module | Daily normalized index price series · correlation | 0.77 -> 0.34 | Hurt | TwinMarket: A Scalable Behavioral and Social Simulation for Financial Markets (NeurIPS 2025 · 2 citations) |
| Use random rather than fixed training agent compositions | VAP Easy and Hard · wAcc. | GNN: 0.53 -> 0.54 (Easy), 0.23 -> 0.28 (Hard); NLM: 0.49 -> 0.56 (Easy), 0.24 -> 0.36 (Hard) | Helped | LogiCity: Advancing Neuro-Symbolic AI with Abstract Urban Simulation (NeurIPS 2024 · 2 citations) |
| replace generated talking-face videos with retrieved pre-recorded clips | Prototype 2 · generation speed | Helped | SPASCA: Social Presence and Support with Conversational Agent for Persons Living with Dementia (AAAI 2025 · 1 citation) | |
| Add perspective-taking during communication | Dirt collection · task completion rate | 54% -> 67% | Helped | MindForge: Empowering Embodied Agents with Theory of Mind for Lifelong Cultural Learning (NeurIPS 2025 · 1 citation) |
| add reciprocal altruism to collective collaboration | CIFAR10 · average accuracy | 63.60 -> 87.22 | Helped | Socialized Learning: Making Each Other Better Through Multi-Agent Collaboration (ICML 2024) |
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 34 papers; from OpenAlex and the papers' affiliations).
- Carnegie Mellon University · 3 papers
- Massachusetts Institute of Technology · 2 papers
- The Hong Kong Polytechnic University · 2 papers
- Google · 2 papers
- Johns Hopkins University · 2 papers
- University of Virginia · 2 papers
- Meta · 2 papers
- The Chinese University of Hong Kong · 2 papers
- Google DeepMind · 2 papers
Use it
Download BibTeX of 40 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Human-AI coordination agents (125 papers)
- Collaborative multi-agent perception (86 papers)
- Boundedly rational agent modeling (87 papers)
- Emergent communication in referential games (50 papers)
Elsewhere in the topic
- World-model-based agents (110 papers)
- Imitation and skill learning (100 papers)
- Language-grounded reinforcement agents (96 papers)
- Embodied agent learning (93 papers)
- Intrinsically motivated exploration (86 papers)
- Interactive multi-agent trajectory prediction (75 papers)
- Learning-guided MAPF search (73 papers)
- Vision-language navigation agents (66 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).