pith. sign in

arxiv: 2606.06380 · v1 · pith:OB5H647Jnew · submitted 2026-06-04 · 💻 cs.CL · cs.AI· cs.MA· cs.NE

Emergent Language as an Approach to Conscious AI

Zengqing Wu , Chuan Xiao This is my paper

Pith reviewed 2026-06-28 01:26 UTC · model grok-4.3

classification 💻 cs.CL cs.AIcs.MAcs.NE
keywords emergent languagemulti-agent reinforcement learningself-referential communicationecho-mismatch detectionconscious AIgenerative methodologyenvironmental affordance
0
0 comments X

The pith

Agents with no initial language or self-concept develop self-referential communication and mismatch-detection circuits under task pressure alone.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces emergent language in multi-agent reinforcement learning as a generative method for investigating consciousness-related structures in artificial systems. Agents begin with minimal conditions—no language, no self-reference, and little exposure to human text—and must develop communication to complete tasks. In a minimal environment, agents produce self-referential signals including an echo-mismatch detection circuit that arises from a specific environmental feature rather than from the task rules or agent design. This approach seeks to attribute any resulting structures directly to environmental and task demands instead of inherited human priors.

Core claim

Placing agents with no language or self-concept in a minimal multi-agent reinforcement learning setting causes them to generate communication protocols that include self-referential elements and an echo-mismatch detection circuit; this circuit is not required by the task structure or architecture and appears only when a particular environmental affordance is present.

What carries the argument

Emergent language in multi-agent reinforcement learning, where agents invent communication protocols from scratch driven only by task success.

If this is right

  • Structures observed in agent communication can be attributed to task demands rather than human-designed priors.
  • Varying environment complexity can reveal which features promote emergence of self-referential communication.
  • Emergent protocols can be interpreted as candidate mechanisms relevant to self-awareness without relying on pre-specified checklists or modules.
  • The method offers a third route to conscious-AI research alongside discriminative evaluation and architectural engineering.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Repeating the setup across environments that differ only in one affordance could isolate the minimal conditions for mismatch detection to appear.
  • The same agents could be tested on tasks requiring coordination that might elicit further markers such as distinguishing self from other signals.
  • If the mismatch circuit generalizes to larger state spaces, it would suggest a route for scaling the approach beyond the proof-of-concept environment.
  • Comparison of the observed circuit with known biological mismatch-detection processes could be performed without assuming equivalence.

Load-bearing premise

Starting agents with no language, no self-concept, and minimal human text exposure ensures that any developed structures result only from task demands.

What would settle it

Running the minimal environment without the specific environmental affordance and finding that the echo-mismatch detection circuit still appears, or finding that it never appears even with the affordance present.

Figures

Figures reproduced from arXiv: 2606.06380 by Chuan Xiao, Zengqing Wu.

Figure 1
Figure 1. Figure 1: EL as a methodological approach to conscious AI. (1) Using “blue” as an example, the figure [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: P1 evidence: (a) MI dominance (log scale), (b) partner-specific benefit; P2 evidence: [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: P3 evidence chain: (a) sender-receiver dissociation, (b) echo causal split, (c) lag-1 temporal [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Agent architecture. (a) Standard agent (for P1 and P2). Input features xt = (si ,ci ,t,mt−1 −i ) are processed by a GRU (d = 128), producing message and action distributions from three linear heads. (b) Echo variant (for P3). The input additionally includes m˜ t−1 i , a possibly corrupted copy of the agent’s own previous message. The corruption occurs after the message is produced and before it is fed back… view at source ↗
read the original abstract

The question of whether artificial systems can be conscious remains open, in part because existing approaches either evaluate systems against theory-derived checklists (discriminative) or engineer consciousness-inspired modules directly (architectural); both leave open whether observed structures are artifacts of human language priors. We propose a generative methodology: emergent language (EL) in multi-agent reinforcement learning, where agents start from minimal (no language, no concept of self, minimal exposure to human text) and develop communication under task pressure alone, ensuring causal attributability to task demands rather than inherited human language priors. We position our methodology by discussing how EL serves as a generative tool for studying consciousness-relevant structure, including the role of environment complexity and the interpretation of emergent communication. As a proof of concept, we instantiate this methodology in a minimal environment and show that agents develop self-referential communication, including an echo-mismatch detection circuit that is not predicted by task structure or architecture alone but emerges from a specific environmental affordance.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The manuscript proposes emergent language in multi-agent reinforcement learning as a generative methodology for studying consciousness-relevant structures in AI. Agents start with no language, no self-concept, and minimal human text exposure, developing communication solely under task pressure. As a proof of concept, the authors report that agents develop self-referential communication, including an echo-mismatch detection circuit that emerges from a specific environmental affordance rather than being predicted by task structure or architecture alone.

Significance. If the empirical results were substantiated with detailed environments, quantitative metrics, and ablation controls, the approach could provide a useful generative tool for investigating self-referential structures without reliance on human language priors or engineered modules. The positioning of EL as an alternative to discriminative or architectural methods for consciousness research is conceptually coherent, but the absence of any reported results, controls, or verification in the manuscript prevents assessment of whether the claimed emergence is actually demonstrated.

major comments (2)
  1. [Abstract] Abstract: The central empirical claim—that an echo-mismatch detection circuit 'is not predicted by task structure or architecture alone but emerges from a specific environmental affordance'—is load-bearing for the proof-of-concept but is stated without any environment description, quantitative results, baseline comparisons, or ablation (identical agents/task with affordance removed). No evidence is supplied to support the causal attribution.
  2. [Abstract] Abstract: The methodology's claim to ensure 'causal attributability to task demands rather than inherited human language priors' rests on the starting conditions (no language, no self-concept, minimal human text), yet the manuscript provides no verification that the observed structures are not generic consequences of the RL objective or network architecture, nor any controls isolating the affordance.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and for identifying the gaps between the abstract claims and the supporting material in the manuscript. The comments correctly highlight that the proof-of-concept requires additional detail to be evaluable. We respond to each point below and indicate planned revisions.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central empirical claim—that an echo-mismatch detection circuit 'is not predicted by task structure or architecture alone but emerges from a specific environmental affordance'—is load-bearing for the proof-of-concept but is stated without any environment description, quantitative results, baseline comparisons, or ablation (identical agents/task with affordance removed). No evidence is supplied to support the causal attribution.

    Authors: We agree that the submitted manuscript states the emergence claim without supplying the requested environment specification, metrics, baselines, or ablation. The current text presents the result at a conceptual level only. In revision we will add a dedicated methods/results subsection containing the environment description, quantitative measures of self-referential communication, comparison to non-affordance baselines, and an explicit ablation removing the critical environmental feature. revision: yes

  2. Referee: [Abstract] Abstract: The methodology's claim to ensure 'causal attributability to task demands rather than inherited human language priors' rests on the starting conditions (no language, no self-concept, minimal human text), yet the manuscript provides no verification that the observed structures are not generic consequences of the RL objective or network architecture, nor any controls isolating the affordance.

    Authors: The manuscript relies on the minimal initialization to argue causal attributability, yet supplies no explicit controls or verification that the structures are not generic to the RL objective or architecture. We accept that this verification is necessary. The revision will include additional analysis and controls that isolate the environmental affordance from the learning algorithm and network architecture. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical emergence result is self-contained

full rationale

The paper advances a generative empirical methodology via multi-agent RL simulations in which agents begin with no language or self-concept and develop communication under task pressure. The central claim is an observed outcome (self-referential communication including an echo-mismatch circuit) in a minimal environment, presented as a proof-of-concept demonstration rather than a mathematical derivation. No equations, parameter-fitting steps, or self-citation chains appear in the provided text that would reduce the reported structures to inputs by construction; the result is framed as causally attributable to the simulation run itself.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that minimal starting conditions eliminate human language priors; no free parameters or invented entities are described in the abstract.

axioms (1)
  • domain assumption Agents start from minimal (no language, no concept of self, minimal exposure to human text)
    Invoked to ensure causal attributability to task demands rather than inherited priors.

pith-pipeline@v0.9.1-grok · 5695 in / 1266 out tokens · 39837 ms · 2026-06-28T01:26:44.274425+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

99 extracted references · 4 linked inside Pith

  1. [1]

    OpenAI CEO on GPT-4, ChatGPT, and the future of AI

    Sam Altman. OpenAI CEO on GPT-4, ChatGPT, and the future of AI. Lex Fridman Podcast #367, March 2023. Transcript available athttps://lexfridman.com/sam-altman/

    2023

  2. [2]

    The urgency of interpretability

    Dario Amodei. The urgency of interpretability. Blog, April 2025. https: //www.darioamodei.com/post/the-urgency-of-interpretability

    2025

  3. [3]

    The consciousness prior.arXiv preprint arXiv:1709.08568, 2017

    Yoshua Bengio. The consciousness prior.arXiv preprint arXiv:1709.08568, 2017

    arXiv 2017

  4. [4]

    Large language models report subjective experience under self-referential processing.arXiv preprint arXiv:2510.24797, 2025

    Cameron Berg, Diogo de Lucena, and Judd Rosenblatt. Large language models report subjective experience under self-referential processing.arXiv preprint arXiv:2510.24797, 2025

    arXiv 2025

  5. [5]

    Tell me about yourself: Llms are aware of their learned behaviors.The Thirteenth International Conference on Learning Representations, 2025

    Jan Betley, Xuchan Bao, Martín Soto, Anna Sztyber-Betley, James Chua, and Owain Evans. Tell me about yourself: Llms are aware of their learned behaviors.The Thirteenth International Conference on Learning Representations, 2025

    2025

  6. [6]

    Experience grounds language

    Yonatan Bisk, Ari Holtzman, Jesse Thomason, Jacob Andreas, Yoshua Bengio, Joyce Chai, Mirella Lapata, Angeliki Lazaridou, Jonathan May, Aleksandr Nisnevich, et al. Experience grounds language. InProceedings of the 2020 conference on empirical methods in natural language processing (EMNLP), pages 8718–8735, 2020

    2020

  7. [7]

    Propositions concerning digital minds and society.Cambridge Journal of Law, Politics, and Art, 3, 2022

    Nick Bostrom and Carl Shulman. Propositions concerning digital minds and society.Cambridge Journal of Law, Politics, and Art, 3, 2022

    2022

  8. [8]

    Identifying indicators of consciousness in ai systems.Trends in Cognitive Sciences, 2025

    Patrick Butlin, Robert Long, Tim Bayne, Yoshua Bengio, Jonathan Birch, David Chalmers, Axel Constant, George Deane, Eric Elmoznino, Stephen M Fleming, et al. Identifying indicators of consciousness in ai systems.Trends in Cognitive Sciences, 2025

    2025

  9. [9]

    Oxford University Press, 2011

    Peter Carruthers.The Opacity of Mind: An Integrative Theory of Self-Knowledge. Oxford University Press, 2011

    2011

  10. [10]

    Compositionality and generalization in emergent languages

    Rahma Chaabouni, Eugene Kharitonov, Diane Bouchacourt, Emmanuel Dupoux, and Marco Baroni. Compositionality and generalization in emergent languages. InProceedings of the 58th annual meeting of the association for computational linguistics, pages 4427–4442, 2020. 14

    2020

  11. [11]

    Facing up to the problem of consciousness.Journal of Consciousness Studies, 2(3):200–219, 1995

    David J Chalmers. Facing up to the problem of consciousness.Journal of Consciousness Studies, 2(3):200–219, 1995

    1995

  12. [12]

    Oxford University Press, 1996

    David J Chalmers.The Conscious Mind: In Search of a Fundamental Theory. Oxford University Press, 1996

    1996

  13. [13]

    Could a large language model be conscious?Boston Review, 2023

    David J Chalmers. Could a large language model be conscious?Boston Review, 2023. Expanded from NeurIPS 2022 invited talk

    2023

  14. [14]

    The five ws of multi-agent communication: Who talks to whom, when, what, and why - a survey from MARL to emergent language and LLMs.Transactions on Machine Learning Research, 2026

    Jingdi Chen, Hanqing Yang, Zongjun Liu, and Carlee Joe-Wong. The five ws of multi-agent communication: Who talks to whom, when, what, and why - a survey from MARL to emergent language and LLMs.Transactions on Machine Learning Research, 2026. ISSN 2835-8856. URLhttps://openreview.net/forum?id=LGsed0QQVq. Survey Certification

    2026

  15. [15]

    Learning phrase representations using rnn encoder– decoder for statistical machine translation

    Kyunghyun Cho, Bart Van Merriënboer, Ça˘glar Gulçehre, Dzmitry Bahdanau, Fethi Bougares, Holger Schwenk, and Yoshua Bengio. Learning phrase representations using rnn encoder– decoder for statistical machine translation. InProceedings of the 2014 conference on empirical methods in natural language processing (EMNLP), pages 1724–1734, 2014

    2014

  16. [16]

    Whatever next? Predictive brains, situated agents, and the future of cognitive science.Behavioral and Brain Sciences, 36(3):181–204, 2013

    Andy Clark. Whatever next? Predictive brains, situated agents, and the future of cognitive science.Behavioral and Brain Sciences, 36(3):181–204, 2013

    2013

  17. [17]

    Adversarial testing of global neuronal workspace and integrated information theories of consciousness

    Cogitate Consortium, Oscar Ferrante, Urszula Gorska-Klimowska, Simon Henin, Rony Hirschhorn, Aya Khalaf, Alex Lepauvre, Ling Liu, David Richter, Yamil Vidal, et al. Adversarial testing of global neuronal workspace and integrated information theories of consciousness. Nature, 642(8066):133–142, 2025

    2025

  18. [18]

    Towards a cognitive neuroscience of consciousness: Basic evidence and a workspace framework.Cognition, 79(1–2):1–37, 2001

    Stanislas Dehaene and Lionel Naccache. Towards a cognitive neuroscience of consciousness: Basic evidence and a workspace framework.Cognition, 79(1–2):1–37, 2001

    2001

  19. [19]

    Stanislas Dehaene, Claire Sergent, and Jean-Pierre Changeux. A neuronal network model linking subjective reports and objective physiological data during conscious perception.Proceedings of the National Academy of Sciences, 100(14):8520–8525, 2003

    2003

  20. [20]

    Creating counterfeit digital people risks destroying our civilization.The Atlantic, 2023

    Daniel C Dennett. Creating counterfeit digital people risks destroying our civilization.The Atlantic, 2023

    2023

  21. [21]

    Penguin uk, 1993

    Daniel Clement Dennett.Consciousness explained. Penguin uk, 1993

    1993

  22. [22]

    Learning to communicate with deep multi-agent reinforcement learning

    Jakob Foerster, Ioannis Alexandros Assael, Nando de Freitas, and Shimon Whiteson. Learning to communicate with deep multi-agent reinforcement learning. InAdvances in Neural Information Processing Systems, volume 29, 2016

    2016

  23. [23]

    Univ of California Press, 1977

    Hans-Georg Gadamer.Philosophical Hermeneutics. Univ of California Press, 1977

    1977

  24. [24]

    Learning and communication pressures in neural networks: Lessons from emergent communication.Language Development Research, 5(1):116–143, 2024

    Lukas Galke and Limor Raviv. Learning and communication pressures in neural networks: Lessons from emergent communication.Language Development Research, 5(1):116–143, 2024

    2024

  25. [25]

    Chimpanzees: Self-recognition.Science, 167(3914):86–87, 1970

    Gordon G Gallup Jr. Chimpanzees: Self-recognition.Science, 167(3914):86–87, 1970

    1970

  26. [26]

    Oxford University Press, 2013

    Michael S A Graziano.Consciousness and the Social Brain. Oxford University Press, 2013

    2013

  27. [27]

    Rhesus monkeys know when they remember.Proceedings of the National Academy of Sciences, 98(9):5359–5362, 2001

    Robert R Hampton. Rhesus monkeys know when they remember.Proceedings of the National Academy of Sciences, 98(9):5359–5362, 2001

    2001

  28. [28]

    Future of AI, simulating reality, physics and video games

    Demis Hassabis. Future of AI, simulating reality, physics and video games. Lex Fridman Podcast #475, 2025. Transcript athttps://lexfridman.com/demis-hassabis-2-transcript/

    2025

  29. [29]

    Emergence of language with multi-agent games: Learning to communicate with sequences of symbols.Advances in neural information processing systems, 30, 2017

    Serhii Havrylov and Ivan Titov. Emergence of language with multi-agent games: Learning to communicate with sequences of symbols.Advances in neural information processing systems, 30, 2017

    2017

  30. [30]

    Multimodal AI already has subjective experiences

    Geoffrey Hinton. Interview on Tonight with Andrew Marr. https://www.lbc.co.uk/ article/ai-consciousness-geoffrey-hinton-5HjdRXD_2/ , 2026. LBC, London. “Multimodal AI already has subjective experiences”. 15

    2026

  31. [31]

    Ai godfather geoffrey hinton says we must convince ai that it’s our mother

    Geoffrey Hinton. Ai godfather geoffrey hinton says we must convince ai that it’s our mother. Keynote presentation at DiscoveryX Conference, Toronto, 2026. Reported in BetaKit, https://betakit.com/ai-godfather-geoffrey-hinton-says-we-must-convince- ai-that-its-our-mother/

    2026

  32. [32]

    A disproof of large language model consciousness: The necessity of continual learning for consciousness.arXiv preprint arXiv:2512.12802, 2025

    Erik Hoel. A disproof of large language model consciousness: The necessity of continual learning for consciousness.arXiv preprint arXiv:2512.12802, 2025

    arXiv 2025

  33. [33]

    Basic Books, 2007

    Douglas R Hofstadter.I Am a Strange Loop. Basic Books, 2007

    2007

  34. [34]

    Hexajungle: a MARL simulator to study the emergence of language

    Kiran Ikram, Esther Mondragón, Eduardo Alonso, and Michael Garcia-Ortiz. Hexajungle: a MARL simulator to study the emergence of language. InProceedings of the CVPR 2021 Embodied AI Workshop, 2021

    2021

  35. [35]

    Probing for consciousness in machines.Frontiers in Artificial Intelligence, 8:1610225, 2025

    Mathis Immertreu, Achim Schilling, Andreas Maier, and Patrick Krauss. Probing for consciousness in machines.Frontiers in Artificial Intelligence, 8:1610225, 2025

    2025

  36. [36]

    Henry Holt and Company, New York, 1890

    William James.The Principles of Psychology. Henry Holt and Company, New York, 1890

  37. [37]

    Demonstratives: An essay on the semantics, logic, metaphysics and epistemology of demonstratives and other indexicals

    David Kaplan. Demonstratives: An essay on the semantics, logic, metaphysics and epistemology of demonstratives and other indexicals. 1989

    1989

  38. [38]

    Short story on AI: Forward pass

    Andrej Karpathy. Short story on AI: Forward pass. Blog, March 2021. https: //karpathy.github.io/2021/03/27/forward-pass/

    2021

  39. [39]

    Adam: A method for stochastic optimization.arXiv preprint arXiv:1412.6980, 2014

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization.arXiv preprint arXiv:1412.6980, 2014

    Pith/arXiv arXiv 2014

  40. [40]

    Natural language does not emerge ‘naturally’ in multi-agent dialog

    Satwik Kottur, José M F Moura, Stefan Lee, and Dhruv Batra. Natural language does not emerge ‘naturally’ in multi-agent dialog. InProceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2962–2967, 2017

    2017

  41. [41]

    Multi-agent cooperation and the emergence of (natural) language

    Angeliki Lazaridou, Alexander Peysakhovich, and Marco Baroni. Multi-agent cooperation and the emergence of (natural) language. InProceedings of the 5th International Conference on Learning Representations, 2017

    2017

  42. [42]

    A path towards autonomous machine intelligence

    Yann LeCun. A path towards autonomous machine intelligence. Technical report, Courant Institute of Mathematical Sciences, New York University / Meta AI, 2022. Version 0.9.2, 2022-06-27. Open Review preprint

    2022

  43. [43]

    Harvard University Press, 1969

    David Lewis.Convention: A Philosophical Study. Harvard University Press, 1969

    1969

  44. [44]

    No, today’s AI isn’t sentient

    Fei-Fei Li and John Etchemendy. No, today’s AI isn’t sentient. Here’s how we know.TIME,

  45. [45]

    https://time.com/collections/time100-voices/6980134/ai-llm-not- sentient/

    arXiv

  46. [46]

    Language grounded multi-agent reinforcement learning with human-interpretable communication.Advances in Neural Information Processing Systems, 37:87908–87933, 2024

    Huao Li, Hossein N Mahjoub, Behdad Chalaki, Vaishnav Tadiparthi, Kwonjoon Lee, Ehsan Moradi-Pari, Michael Lewis, and Katia Sycara. Language grounded multi-agent reinforcement learning with human-interpretable communication.Advances in Neural Information Processing Systems, 37:87908–87933, 2024

    2024

  47. [47]

    Emergent introspective awareness in large language models.arXiv preprint arXiv:2601.01828, 2026

    Jack Lindsey. Emergent introspective awareness in large language models.arXiv preprint arXiv:2601.01828, 2026

    arXiv 2026

  48. [48]

    It’s about time: Temporal references in emergent communication.arXiv preprint arXiv:2310.06555, 2023

    Olaf Lipinski, Adam J Sobey, Federico Cerutti, and Timothy J Norman. It’s about time: Temporal references in emergent communication.arXiv preprint arXiv:2310.06555, 2023

    arXiv 2023

  49. [49]

    Sobey, Federico Cerutti, and Timothy J

    Olaf Lipinski, Adam J. Sobey, Federico Cerutti, and Timothy J. Norman. Speaking your language: Spatial relationships in interpretable emergent communication. InAdvances in Neural Information Processing Systems (NeurIPS), volume 37, 2024

    2024

  50. [50]

    John L. Locke. The indexical voice: Communication of personal states and traits in humans and other primates.Frontiers in Psychology, 12:651108, 2021

    2021

  51. [51]

    Reinforcing the world’s edge: A continual learning problem in the multi-agent-world boundary.arXiv preprint arXiv:2603.06813, 2026

    Dane Malenfant. Reinforcing the world’s edge: A continual learning problem in the multi-agent-world boundary.arXiv preprint arXiv:2603.06813, 2026. 16

    arXiv 2026

  52. [52]

    Richard dawkins and the claude delusion

    Gary Marcus. Richard dawkins and the claude delusion. Blog, 2026. https: //garymarcus.substack.com/p/richard-dawkins-and-the-claude-delusion

    2026

  53. [53]

    Agnosticism about artificial consciousness.Mind & Language, 2025

    Tom McClelland. Agnosticism about artificial consciousness.Mind & Language, 2025

    2025

  54. [54]

    University of Chicago Press, 1934

    George Herbert Mead.Mind, Self, and Society. University of Chicago Press, 1934. Posthumous, edited by Charles W. Morris

    1934

  55. [55]

    MIT Press, 2003

    Thomas Metzinger.Being No One: The Self-Model Theory of Subjectivity. MIT Press, 2003

    2003

  56. [56]

    Asynchronous methods for deep reinforcement learning

    V olodymyr Mnih, Adria Puigdomenech Badia, Mehdi Mirza, Alex Graves, Timothy Lillicrap, Tim Harley, David Silver, and Koray Kavukcuoglu. Asynchronous methods for deep reinforcement learning. InInternational conference on machine learning, pages 1928–1937. PmLR, 2016

    1928

  57. [57]

    Todd C. Moody. Conversations with zombies.Journal of Consciousness Studies, 1(2):196–200, 1994

    1994

  58. [58]

    Emergence of grounded compositional language in multi-agent populations

    Igor Mordatch and Pieter Abbeel. Emergence of grounded compositional language in multi-agent populations. InProceedings of the AAAI Conference on Artificial Intelligence, volume 32, 2018

    2018

  59. [59]

    Emergent communication with conversational repair

    Mitja Nikolaus. Emergent communication with conversational repair. InProceedings of the Twelfth International Conference on Learning Representations (ICLR), 2024

    2024

  60. [60]

    A survey of continual reinforcement learning.arXiv preprint arXiv:2506.21872, 2025

    Chaofan Pan, Xin Yang, Yanhua Li, Wei Wei, Tianrui Li, Bo An, and Jiye Liang. A survey of continual reinforcement learning.arXiv preprint arXiv:2506.21872, 2025

    Pith/arXiv arXiv 2025

  61. [61]

    Interpretation of emergent communication in heterogeneous collaborative embodied agents

    Shivansh Patel, Saim Wani, Unnat Jain, Alexander G Schwing, Svetlana Lazebnik, Manolis Savva, and Angel X Chang. Interpretation of emergent communication in heterogeneous collaborative embodied agents. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 15953–15963, 2021

    2021

  62. [62]

    Five hundred and eighty-second meeting

    Charles Sanders Peirce. Five hundred and eighty-second meeting. may 14, 1867. monthly meeting; on a new list of categories. InProceedings of the american academy of arts and sciences, volume 7, pages 287–298. JSTOR, 1865

  63. [63]

    Oxford University Press, 1994

    Roger Penrose.Shadows of the Mind: A Search for the Missing Science of Consciousness. Oxford University Press, 1994

    1994

  64. [64]

    The problem of the essential indexical.Noûs, 13(1):3–21, 1979

    John Perry. The problem of the essential indexical.Noûs, 13(1):3–21, 1979

    1979

  65. [65]

    Emergent language: a survey and taxonomy.Autonomous Agents and Multi-Agent Systems, 39(1):18, 2025

    Jannik Peters, Constantin Waubert de Puiseau, Hasan Tercan, Arya Gopikrishnan, Gus- tavo Adolpho Lucas de Carvalho, Christian Bitter, and Tobias Meisen. Emergent language: a survey and taxonomy.Autonomous Agents and Multi-Agent Systems, 39(1):18, 2025

    2025

  66. [66]

    From grunts to lexicons: Emergent language from cooperative foraging

    Maytus Piriyajitakonkij, Rujikorn Charakorn, Weicheng Tao, Wei Pan, Mingfei Sun, Cheston Tan, and Mengmi Zhang. From grunts to lexicons: Emergent language from cooperative foraging. arXiv preprint arXiv:2505.12872, 2025

    arXiv 2025

  67. [67]

    There is no such thing as conscious artificial intelligence

    Andrzej Por˛ ebski and Jakub Figura. There is no such thing as conscious artificial intelligence. Humanities and Social Sciences Communications, 12(1):1–12, 2025

    2025

  68. [68]

    Does the chimpanzee have a theory of mind?Behavioral and Brain Sciences, 1(4):515–526, 1978

    David Premack and Guy Woodruff. Does the chimpanzee have a theory of mind?Behavioral and Brain Sciences, 1(4):515–526, 1978

    1978

  69. [69]

    Premakumar, Michael Vaiana, Florin Pop, Judd Rosenblatt, Diogo Schwerz de Lucena, Kirsten Ziman, and Michael S

    Vickram N. Premakumar, Michael Vaiana, Florin Pop, Judd Rosenblatt, Diogo Schwerz de Lucena, Kirsten Ziman, and Michael S. A. Graziano. Unexpected benefits of self-modeling in neural systems.arXiv preprint arXiv:2407.10188, 2024

    arXiv 2024

  70. [70]

    Cohen, and Simon Kirby

    Yi Ren, Shangmin Guo, Matthieu Labeau, Shay B. Cohen, and Simon Kirby. Compositional languages emerge in a neural iterated learning model. InInternational Conference on Learning Representations, 2020

    2020

  71. [71]

    Dai, and Kyunghyun Cho

    Cinjon Resnick, Abhinav Gupta, Jakob Foerster, Andrew M. Dai, and Kyunghyun Cho. Capacity, bandwidth, and compositionality in emergent language learning. InProceedings of the 19th International Conference on Autonomous Agents and Multiagent Systems (AAMAS), pages 1125–1133, 2020. 17

    2020

  72. [72]

    How many kinds of consciousness?Consciousness and Cognition, 11(4): 653–665, 2002

    David M Rosenthal. How many kinds of consciousness?Consciousness and Cognition, 11(4): 653–665, 2002

    2002

  73. [73]

    Oxford University Press, 2005

    David M Rosenthal.Consciousness and Mind. Oxford University Press, 2005

    2005

  74. [74]

    Faber & Faber, 2021

    Anil Seth.Being You: A New Science of Consciousness. Faber & Faber, 2021

    2021

  75. [75]

    The mythology of conscious AI

    Anil Seth. The mythology of conscious AI. Noema Magazine, 2026. https: //www.noemamag.com/the-mythology-of-conscious-ai/

    2026

  76. [76]

    Non-human consciousness and the specificity problem: A modest theoretical proposal.Mind & Language, 36(2):297–314, 2021

    Henry Shevlin. Non-human consciousness and the specificity problem: A modest theoretical proposal.Mind & Language, 36(2):297–314, 2021

    2021

  77. [77]

    David Silver and Richard S. Sutton. Welcome to the era of experience. 2025. Preprint, to appear inDesigning an Intelligence, MIT Press

    2025

  78. [78]

    The comparative psychology of uncertainty monitoring and metacognition.Behavioral and Brain Sciences, 26(3):317–339, 2003

    J David Smith, Wendy E Shields, and David A Washburn. The comparative psychology of uncertainty monitoring and metacognition.Behavioral and Brain Sciences, 26(3):317–339, 2003

    2003

  79. [79]

    Emotion concepts and their function in a large language model.arXiv preprint arXiv:2604.07729, 2026

    Nicholas Sofroniew, Isaac Kauvar, William Saunders, Runjin Chen, Tom Henighan, Sasha Hydrie, Craig Citro, Adam Pearce, Julius Tarng, Wes Gurnee, et al. Emotion concepts and their function in a large language model.arXiv preprint arXiv:2604.07729, 2026

    Pith/arXiv arXiv 2026

  80. [80]

    “I” who? A new look at Peirce’s theory of indexical self-reference.The Pluralist, 10(2):220–246, 2015

    Marco Stango. “I” who? A new look at Peirce’s theory of indexical self-reference.The Pluralist, 10(2):220–246, 2015

    2015

Showing first 80 references.