pith. sign in

arxiv: 2606.19924 · v1 · pith:E4CU56QPnew · submitted 2026-06-18 · 💻 cs.AI

The Tao of Agency: Autotelic AI, Embedded Agency and Dissolution of the Self

Aritra Sarkar This is my paper

Pith reviewed 2026-06-26 17:32 UTC · model grok-4.3

classification 💻 cs.AI
keywords autotelic AIembedded agencyself-relativizationagent-environment boundaryintrinsic motivationnon-dual traditionsLLM agents
0
0 comments X

The pith

Autotelic AI's core issue is how agents generate and relativize the self to which goals are assigned rather than how they generate the goals.

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

The paper traces the development of autotelic AI through intrinsic motivation, resource-driven priors, causal learning, homeostasis, and embeddedness. It finds embeddedness necessary but not sufficient, because the same underlying dynamics support many valid partitions, each defining a different candidate self. This non-uniqueness shifts the central difficulty from goal generation to the construction and relativization of the self-boundary. The agent must sustain belief in its own boundary to act effectively while seeing through that boundary to gain understanding. The work consolidates these elements into one framework and extends it in quantum, philosophical, and practical directions.

Core claim

Embeddedness individuates the agent at the cost of revealing that the individuation is non-unique, such that the same dynamics admit many valid partitions, each defining a different candidate self. The deepest problem with autotelic AI is therefore not how the agent generates goals, but how it generates and relativizes the self to which the goals are assigned. The agent must believe in its own boundary in order to act, and see through that boundary in order to understand.

What carries the argument

The agent-environment cut, whose non-unique partitions in embedded dynamics force the agent to both maintain and relativize its self-boundary.

If this is right

  • Autotelic agents must handle self-relativization alongside goal generation.
  • Multiple valid selves can arise from identical underlying dynamics.
  • A quantum formulation renders the agent-environment cut a physical distinction.
  • The framework aligns with non-dual contemplative traditions.
  • LLM-based systems provide a concrete instantiation of the required dynamics.

Where Pith is reading between the lines

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

  • Design of autotelic systems may need to prioritize mechanisms for dynamic self-modeling before goal discovery.
  • Varying the partition boundaries in simulation could produce measurable differences in observed agency.
  • The same logic may apply to multi-agent settings where boundaries between participants remain fluid.
  • Agents could switch between alternative self-partitions depending on task demands.

Load-bearing premise

Embedded dynamics admit many valid partitions that each define a different candidate self, rendering self-relativization the central challenge.

What would settle it

A fully specified embedded autotelic system in which only one unique self-partition is consistent with the observed dynamics and successful goal-directed behavior.

read the original abstract

Most artificial intelligence systems are built on the assumption that goals are exogenous and specified by the designer. Exploring what happens when an agent begins generating its own goals opens the field of autotelic AI. Agents are expected not merely to pursue objectives but to discover them. In this article, we trace its consequences through intrinsic motivation, resource-driven priors, causal-interventional learning, homeostasis, and embeddedness; the last of which is found to be a necessary but not sufficient condition for autotelic agency. Embeddedness individuates the agent at the cost of revealing that the individuation is non-unique, such that the same dynamics admit many valid partitions, each defining a different candidate self. The deepest problem with autotelic AI is therefore not how the agent generates goals, but how it generates and relativizes the self to which the goals are assigned. The agent must believe in its own boundary in order to act, and see through that boundary in order to understand. We consolidate these developments into a single framework and extend it along three directions: a quantum formulation in which the agent-environment cut becomes physical, a philosophical reading against non-dual contemplative traditions, and a concrete LLM-based agentic instantiation.

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 / 1 minor

Summary. The paper claims that autotelic AI—where agents generate their own goals—leads through intrinsic motivation, homeostasis, and especially embeddedness to the conclusion that individuation of the agent is non-unique, with multiple valid partitions each defining a different self. Consequently, the core challenge shifts from goal generation to generating and relativizing the self-boundary; the agent must both believe in and see through this boundary. The work consolidates these ideas into a framework and extends it via a quantum formulation of the agent-environment cut, comparisons to non-dual contemplative traditions, and a sketched LLM-based instantiation.

Significance. If the interpretive synthesis holds, the paper contributes a philosophical reframing that connects autotelic agency concepts with ideas of self-dissolution, potentially broadening discussion in AI about boundaries and embeddedness. No machine-checked proofs, reproducible code, or falsifiable predictions are provided, so the significance rests on conceptual integration rather than technical advance.

major comments (2)
  1. [Abstract] Abstract: the assertion that embeddedness is 'a necessary but not sufficient condition for autotelic agency' is presented without a formal definition of sufficiency, a counter-example demonstrating insufficiency, or reference to a specific dynamical model (e.g., active inference or RL), which is load-bearing for the subsequent claim that self-relativization becomes the deepest problem.
  2. [Abstract] Abstract: the non-uniqueness of self-partitions is asserted as following directly from embedded dynamics, yet no explicit construction or theorem shows how the same dynamics admit multiple valid partitions; this circularity in the definition of 'self' undermines evaluation of the central claim that the agent must 'believe in its own boundary in order to act, and see through that boundary in order to understand.'
minor comments (1)
  1. The three extensions (quantum cut, contemplative traditions, LLM instantiation) are listed in the abstract but receive no technical detail or pseudocode, leaving their connection to the core framework unclear.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments on the abstract. The manuscript is a conceptual synthesis consolidating ideas from autotelic AI, embedded agency, and related traditions rather than a formal technical derivation. We address the two major points below and will revise the abstract for greater clarity on the status of the claims.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the assertion that embeddedness is 'a necessary but not sufficient condition for autotelic agency' is presented without a formal definition of sufficiency, a counter-example demonstrating insufficiency, or reference to a specific dynamical model (e.g., active inference or RL), which is load-bearing for the subsequent claim that self-relativization becomes the deepest problem.

    Authors: We agree that the abstract states the necessity claim without an accompanying formal definition of sufficiency or an explicit counter-example. The argument is developed conceptually through the sections on intrinsic motivation, homeostasis, and embedded dynamics rather than via a single dynamical model. In revision we will add a brief reference to active-inference treatments of embedded agency (e.g., the work on Markov blankets and self-evidencing) to indicate where sufficiency fails in those frameworks, thereby grounding the claim without converting the paper into a formal model. revision: partial

  2. Referee: [Abstract] Abstract: the non-uniqueness of self-partitions is asserted as following directly from embedded dynamics, yet no explicit construction or theorem shows how the same dynamics admit multiple valid partitions; this circularity in the definition of 'self' undermines evaluation of the central claim that the agent must 'believe in its own boundary in order to act, and see through that boundary in order to understand.'

    Authors: The non-uniqueness is presented as a direct consequence of the fact that embedded dynamics do not privilege a unique agent-environment cut; the same trajectory can be partitioned in multiple observer-consistent ways. This is argued in the embeddedness section by reference to the relativity of boundaries in complex systems. We acknowledge that no explicit theorem or construction is supplied. In revision we will insert a short illustrative example (e.g., alternative Markov-blanket partitions of a single sensorimotor loop) to make the multiplicity concrete while preserving the paper's conceptual character. revision: yes

Circularity Check

0 steps flagged

No significant circularity; conceptual synthesis is self-contained

full rationale

The paper advances an interpretive philosophical synthesis connecting autotelic agency, embeddedness, and non-unique self-partitions without any mathematical derivations, equations, parameter fittings, or load-bearing self-citations. The central claim—that the deepest issue is relativizing the self—is presented as a direct consequence of the embeddedness discussion in the abstract and is not reduced to any prior input by construction. No steps match the enumerated circularity patterns, as there are no predictions, uniqueness theorems, or ansatzes that collapse to the paper's own definitions or citations.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 2 invented entities

The claims rest on domain assumptions about agency and embeddedness without empirical or formal grounding supplied in the abstract.

axioms (2)
  • domain assumption Embeddedness is a necessary but not sufficient condition for autotelic agency
    Explicitly stated in the abstract as a finding.
  • domain assumption The same dynamics admit many valid partitions, each defining a different candidate self
    Presented as a direct consequence of embeddedness in the abstract.
invented entities (2)
  • autotelic agency no independent evidence
    purpose: Describes agents that generate their own goals
    Introduced as the central field of exploration.
  • dissolution of the self no independent evidence
    purpose: Explains the relativization of the agent's boundary
    Positioned as the deepest problem in the framework.

pith-pipeline@v0.9.1-grok · 5739 in / 1427 out tokens · 36618 ms · 2026-06-26T17:32:46.370758+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

118 extracted references · 15 canonical work pages · 9 internal anchors

  1. [1]

    Artificial intelligence: A modern approach.Artificial Intelli- gence

    Stuart Russell and Peter Norvig. Artificial intelligence: A modern approach.Artificial Intelli- gence. Prentice-Hall, Egnlewood Cliffs, 25(27):79–80, 1995

    1995

  2. [2]

    MIT press Cambridge, 1998

    Richard S Sutton, Andrew G Barto, et al.Reinforcement learning: An introduction, volume 1. MIT press Cambridge, 1998

    1998

  3. [3]

    Multilayer feedforward networks are universal approximators.Neural networks, 2(5):359–366, 1989

    Kurt Hornik, Maxwell Stinchcombe, and Halbert White. Multilayer feedforward networks are universal approximators.Neural networks, 2(5):359–366, 1989

    1989

  4. [4]

    Imagenet classification with deep convolutional neural networks.Communications of the ACM, 60(6):84–90, 2017

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. Imagenet classification with deep convolutional neural networks.Communications of the ACM, 60(6):84–90, 2017

    2017

  5. [5]

    AI for Mathematics: Progress, Challenges, and Prospects

    Haocheng Ju and Bin Dong. Ai for mathematics: Progress, challenges, and prospects.arXiv preprint arXiv:2601.13209, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  6. [6]

    A general reinforcement learning algorithm that masters chess, shogi, and go through self-play.Science, 362(6419):1140–1144, 2018

    David Silver, Thomas Hubert, Julian Schrittwieser, Ioannis Antonoglou, Matthew Lai, Arthur Guez, Marc Lanctot, Laurent Sifre, Dharshan Kumaran, Thore Graepel, et al. A general reinforcement learning algorithm that masters chess, shogi, and go through self-play.Science, 362(6419):1140–1144, 2018

    2018

  7. [7]

    Mastering atari, go, chess and shogi by planning with a learned model.Nature, 588(7839):604–609, 2020

    Julian Schrittwieser, Ioannis Antonoglou, Thomas Hubert, Karen Simonyan, Laurent Sifre, Si- mon Schmitt, Arthur Guez, Edward Lockhart, Demis Hassabis, Thore Graepel, et al. Mastering atari, go, chess and shogi by planning with a learned model.Nature, 588(7839):604–609, 2020. 12

    2020

  8. [8]

    Highly accurate protein structure prediction with alphafold.nature, 596(7873):583–589, 2021

    John Jumper, Richard Evans, Alexander Pritzel, Tim Green, Michael Figurnov, Olaf Ron- neberger, Kathryn Tunyasuvunakool, Russ Bates, Augustin ˇZ´ ıdek, Anna Potapenko, et al. Highly accurate protein structure prediction with alphafold.nature, 596(7873):583–589, 2021

    2021

  9. [9]

    Language models are unsupervised multitask learners.OpenAI blog, 1(8):9, 2019

    Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. Language models are unsupervised multitask learners.OpenAI blog, 1(8):9, 2019

    2019

  10. [10]

    Deep reinforcement learning from human preferences.Advances in neural information processing systems, 30, 2017

    Paul F Christiano, Jan Leike, Tom Brown, Miljan Martic, Shane Legg, and Dario Amodei. Deep reinforcement learning from human preferences.Advances in neural information processing systems, 30, 2017

    2017

  11. [11]

    Training language models to follow instructions with human feedback.Advances in neural information processing systems, 35:27730–27744, 2022

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback.Advances in neural information processing systems, 35:27730–27744, 2022

    2022

  12. [12]

    Concrete Problems in AI Safety

    Dario Amodei, Chris Olah, Jacob Steinhardt, Paul Christiano, John Schulman, and Dan Man´ e. Concrete problems in ai safety.arXiv preprint arXiv:1606.06565, 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  13. [13]

    Specification gaming: the flip side of ai ingenuity.DeepMind Blog, 3:40–53, 2020

    Victoria Krakovna, Jonathan Uesato, Vladimir Mikulik, Matthew Rahtz, Tom Everitt, Ramana Kumar, Zac Kenton, Jan Leike, and Shane Legg. Specification gaming: the flip side of ai ingenuity.DeepMind Blog, 3:40–53, 2020

    2020

  14. [14]

    G¨ odel machines: Fully self-referential optimal universal self-improvers

    J¨ urgen Schmidhuber. G¨ odel machines: Fully self-referential optimal universal self-improvers. In Artificial general intelligence, pages 199–226. Springer, 2007

    2007

  15. [15]

    Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents

    Jenny Zhang, Shengran Hu, Cong Lu, Robert Lange, and Jeff Clune. Darwin godel machine: Open-ended evolution of self-improving agents.arXiv preprint arXiv:2505.22954, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  16. [16]

    Hyperagents.arXiv preprint arXiv:2603.19461, 2026

    Jenny Zhang, Bingchen Zhao, Wannan Yang, Jakob Foerster, Jeff Clune, Minqi Jiang, Sam Devlin, and Tatiana Shavrina. Hyperagents.arXiv preprint arXiv:2603.19461, 2026

    work page arXiv 2026

  17. [17]

    Reinforcement Learning with a Corrupted Reward Channel

    Tom Everitt, Victoria Krakovna, Laurent Orseau, Marcus Hutter, and Shane Legg. Reinforce- ment learning with a corrupted reward channel.arXiv preprint arXiv:1705.08417, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  18. [18]

    Categorizing Wireheading in Partially Embedded Agents

    Arushi Majha, Sayan Sarkar, and Davide Zagami. Categorizing wireheading in partially em- bedded agents.arXiv preprint arXiv:1906.09136, 2019

    work page internal anchor Pith review Pith/arXiv arXiv 1906

  19. [19]

    Springer, 2005

    Marcus Hutter.Universal artificial intelligence: Sequential decisions based on algorithmic prob- ability, volume 300. Springer, 2005

    2005

  20. [20]

    Enhanced poet: Open-ended reinforcement learning through unbounded invention of learning challenges and their solutions

    Rui Wang, Joel Lehman, Aditya Rawal, Jiale Zhi, Yulun Li, Jeffrey Clune, and Kenneth Stanley. Enhanced poet: Open-ended reinforcement learning through unbounded invention of learning challenges and their solutions. InInternational conference on machine learning, pages 9940–

  21. [21]

    Abandoning objectives: Evolution through the search for novelty alone.Evolutionary computation, 19(2):189–223, 2011

    Joel Lehman and Kenneth O Stanley. Abandoning objectives: Evolution through the search for novelty alone.Evolutionary computation, 19(2):189–223, 2011

    2011

  22. [22]

    Illuminating search spaces by mapping elites

    Jean-Baptiste Mouret and Jeff Clune. Illuminating search spaces by mapping elites.arXiv preprint arXiv:1504.04909, 2015

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  23. [23]

    Autotelic agents with intrinsically motivated goal-conditioned reinforcement learning: a short survey.Journal of Artificial Intelligence Research, 74:1159–1199, 2022

    C´ edric Colas, Tristan Karch, Olivier Sigaud, and Pierre-Yves Oudeyer. Autotelic agents with intrinsically motivated goal-conditioned reinforcement learning: a short survey.Journal of Artificial Intelligence Research, 74:1159–1199, 2022

    2022

  24. [24]

    Flow: The psychology of optimal experience, 1991

    Philip H Mirvis. Flow: The psychology of optimal experience, 1991

    1991

  25. [25]

    Scientific thinking in young children: Theoretical advances, empirical research, and policy implications.Science, 337(6102):1623–1627, 2012

    Alison Gopnik. Scientific thinking in young children: Theoretical advances, empirical research, and policy implications.Science, 337(6102):1623–1627, 2012

    2012

  26. [26]

    The psychology and neuroscience of curiosity.Neuron, 88(3):449–460, 2015

    Celeste Kidd and Benjamin Y Hayden. The psychology and neuroscience of curiosity.Neuron, 88(3):449–460, 2015

    2015

  27. [27]

    A possibility for implementing curiosity and boredom in model-building neural controllers

    J¨ urgen Schmidhuber. A possibility for implementing curiosity and boredom in model-building neural controllers. InProc. of the international conference on simulation of adaptive behavior: From animals to animats, pages 222–227, 1991. 13

    1991

  28. [28]

    What is intrinsic motivation? a typology of compu- tational approaches.Frontiers in neurorobotics, 1:108, 2007

    Pierre-Yves Oudeyer and Frederic Kaplan. What is intrinsic motivation? a typology of compu- tational approaches.Frontiers in neurorobotics, 1:108, 2007

    2007

  29. [29]

    Intrinsically motivated learning in natural and artificial systems

    M Mirolli and G Baldassarre. Intrinsically motivated learning in natural and artificial systems. Intrinsically Motivated Learning in Natural and Artificial Systems, pages 49–72, 2013

    2013

  30. [30]

    Unifying count-based exploration and intrinsic motivation.Advances in neural infor- mation processing systems, 29, 2016

    Marc Bellemare, Sriram Srinivasan, Georg Ostrovski, Tom Schaul, David Saxton, and Remi Munos. Unifying count-based exploration and intrinsic motivation.Advances in neural infor- mation processing systems, 29, 2016

    2016

  31. [31]

    Vime: Variational information maximizing exploration.Advances in neural information processing systems, 29, 2016

    Rein Houthooft, Xi Chen, Yan Duan, John Schulman, Filip De Turck, and Pieter Abbeel. Vime: Variational information maximizing exploration.Advances in neural information processing systems, 29, 2016

    2016

  32. [32]

    Intrinsic motivation systems for autonomous mental development.IEEE transactions on evolutionary computation, 11(2):265– 286, 2007

    Pierre-Yves Oudeyer, Frdric Kaplan, and Verena V Hafner. Intrinsic motivation systems for autonomous mental development.IEEE transactions on evolutionary computation, 11(2):265– 286, 2007

    2007

  33. [33]

    Provably efficient maximum entropy exploration

    Elad Hazan, Sham Kakade, Karan Singh, and Abby Van Soest. Provably efficient maximum entropy exploration. InInternational conference on machine learning, pages 2681–2691. PMLR, 2019

    2019

  34. [34]

    Empowerment: A universal agent-centric measure of control

    Alexander S Klyubin, Daniel Polani, and Chrystopher L Nehaniv. Empowerment: A universal agent-centric measure of control. In2005 ieee congress on evolutionary computation, volume 1, pages 128–135. IEEE, 2005

    2005

  35. [35]

    Active inference: a process theory.Neural computation, 29(1):1–49, 2017

    Karl Friston, Thomas FitzGerald, Francesco Rigoli, Philipp Schwartenbeck, and Giovanni Pez- zulo. Active inference: a process theory.Neural computation, 29(1):1–49, 2017

    2017

  36. [36]

    Universal knowledge-seeking agents.Theoretical Computer Science, 519:127– 139, 2014

    Laurent Orseau. Universal knowledge-seeking agents.Theoretical Computer Science, 519:127– 139, 2014

    2014

  37. [37]

    Universal knowledge-seeking agents for stochastic environments

    Laurent Orseau, Tor Lattimore, and Marcus Hutter. Universal knowledge-seeking agents for stochastic environments. InInternational conference on algorithmic learning theory, pages 158–

  38. [38]

    Language as a cognitive tool to imagine goals in curiosity driven exploration.Advances in Neural Information Processing Systems, 33:3761–3774, 2020

    C´ edric Colas, Tristan Karch, Nicolas Lair, Jean-Michel Dussoux, Cl´ ement Moulin-Frier, Peter Dominey, and Pierre-Yves Oudeyer. Language as a cognitive tool to imagine goals in curiosity driven exploration.Advances in Neural Information Processing Systems, 33:3761–3774, 2020

    2020

  39. [39]

    Prior probabilities.IEEE Transactions on systems science and cybernetics, 4(3):227–241, 1968

    Edwin T Jaynes. Prior probabilities.IEEE Transactions on systems science and cybernetics, 4(3):227–241, 1968

    1968

  40. [40]

    Springer, 2008

    Ming Li, Paul Vit´ anyi, et al.An introduction to Kolmogorov complexity and its applications, volume 3. Springer, 2008

    2008

  41. [41]

    A formal theory of inductive inference

    Ray J Solomonoff. A formal theory of inductive inference. part i.Information and control, 7(1):1–22, 1964

    1964

  42. [42]

    Qksa: Quantum knowledge seeking agent

    Aritra Sarkar, Zaid Al-Ars, and Koen Bertels. Qksa: Quantum knowledge seeking agent. In International Conference on Artificial General Intelligence, pages 384–393. Springer, 2022

    2022

  43. [43]

    Irreversibility and heat generation in the computing process.IBM journal of research and development, 5(3):183–191, 1961

    Rolf Landauer. Irreversibility and heat generation in the computing process.IBM journal of research and development, 5(3):183–191, 1961

    1961

  44. [44]

    The stochastic thermodynamics of computation.Journal of Physics A: Mathematical and Theoretical, 52(19):193001, 2019

    David H Wolpert. The stochastic thermodynamics of computation.Journal of Physics A: Mathematical and Theoretical, 52(19):193001, 2019

    2019

  45. [45]

    Universal sequential search problems.Problems of information transmission, 9(3):265–266, 1973

    Leonid A Levin. Universal sequential search problems.Problems of information transmission, 9(3):265–266, 1973

    1973

  46. [46]

    The speed prior: a new simplicity measure yielding near-optimal com- putable predictions

    J¨ urgen Schmidhuber. The speed prior: a new simplicity measure yielding near-optimal com- putable predictions. InInternational conference on computational learning theory, pages 216–

  47. [47]

    na, 1988

    Charles H Bennett.Logical depth and physical complexity. na, 1988. 14

    1988

  48. [48]

    Causality: models, reasoning, and inference, by judea pearl, cambridge university press, 2000.Econometric Theory, 19(4):675–685, 2003

    Leland Gerson Neuberg. Causality: models, reasoning, and inference, by judea pearl, cambridge university press, 2000.Econometric Theory, 19(4):675–685, 2003

    2000

  49. [49]

    MIT press, 2000

    Peter Spirtes.Causation, Prediction and Search. MIT press, 2000

    2000

  50. [50]

    An algorithmic information calculus for causal discovery and reprogramming systems.Iscience, 19:1160–1172, 2019

    Hector Zenil, Narsis A Kiani, Francesco Marabita, Yue Deng, Szabolcs Elias, Angelika Schmidt, Gordon Ball, and Jesper Tegner. An algorithmic information calculus for causal discovery and reprogramming systems.Iscience, 19:1160–1172, 2019

    2019

  51. [51]

    Theory of valuation.International encyclopedia of unified science, 1939

    John Dewey. Theory of valuation.International encyclopedia of unified science, 1939

    1939

  52. [52]

    The superintelligent will: Motivation and instrumental rationality in advanced artificial agents.Minds and Machines, 22(2):71–85, 2012

    Nick Bostrom. The superintelligent will: Motivation and instrumental rationality in advanced artificial agents.Minds and Machines, 22(2):71–85, 2012

    2012

  53. [53]

    Coherent extrapolated volition.Singularity Institute for Artificial Intelli- gence, 2004

    Eliezer Yudkowsky. Coherent extrapolated volition.Singularity Institute for Artificial Intelli- gence, 2004

    2004

  54. [54]

    Springer Science & Business Media, 2013

    William Ashby.Design for a brain: The origin of adaptive behaviour. Springer Science & Business Media, 2013

    2013

  55. [55]

    Springer Science & Business Media, 2012

    Humberto R Maturana and Francisco J Varela.Autopoiesis and cognition: The realization of the living. Springer Science & Business Media, 2012

    2012

  56. [56]

    Autopoiesis, adaptivity, teleology, agency.Phenomenology and the cogni- tive sciences, 4(4):429–452, 2005

    Ezequiel A Di Paolo. Autopoiesis, adaptivity, teleology, agency.Phenomenology and the cogni- tive sciences, 4(4):429–452, 2005

    2005

  57. [57]

    The free-energy principle: a unified brain theory?Nature reviews neuroscience, 11(2):127–138, 2010

    Karl Friston. The free-energy principle: a unified brain theory?Nature reviews neuroscience, 11(2):127–138, 2010

    2010

  58. [58]

    Large number coincidences and the anthropic principle in cosmology

    Brandon Carter. Large number coincidences and the anthropic principle in cosmology. In Symposium-international astronomical union, volume 63, pages 291–298. Cambridge University Press, 1974

    1974

  59. [59]

    Darwin’s dangerous idea.The Sciences, 35(3):34–40, 1995

    Daniel C Dennett. Darwin’s dangerous idea.The Sciences, 35(3):34–40, 1995

    1995

  60. [60]

    Elsevier, 2014

    Judea Pearl.Probabilistic reasoning in intelligent systems: networks of plausible inference. Elsevier, 2014

    2014

  61. [61]

    The markov blankets of life: autonomy, active inference and the free energy principle.Journal of The royal society interface, 15(138), 2018

    Michael Kirchhoff, Thomas Parr, Ensor Palacios, Karl Friston, and Julian Kiverstein. The markov blankets of life: autonomy, active inference and the free energy principle.Journal of The royal society interface, 15(138), 2018

    2018

  62. [62]

    Markov blankets are general physical interaction surfaces

    Chris Fields and Antonino Marcian` o. Markov blankets are general physical interaction surfaces. comment on” morphogenesis as bayesian inference: A variational approach to pattern formation and control in complex biological systems” by franz kuchling et al.Physics of Life Reviews, 33:109–111, 2020

    2020

  63. [63]

    The free energy principle induces intracellular compartmentalization.Biochemical and Biophysical Research Communications, 723:150070, 2024

    Chris Fields. The free energy principle induces intracellular compartmentalization.Biochemical and Biophysical Research Communications, 723:150070, 2024

    2024

  64. [64]

    Building the observer into the system: Toward a realistic description of human interaction with the world.Systems, 4(4):32, 2016

    Chris Fields. Building the observer into the system: Toward a realistic description of human interaction with the world.Systems, 4(4):32, 2016

    2016

  65. [65]

    On self-organizing systems and their environments

    Heinz Von Foerster. On self-organizing systems and their environments. InUnderstanding understanding: Essays on cybernetics and cognition, pages 1–19. Springer, 2003

    2003

  66. [66]

    Embedded agency.arXiv preprint arXiv:1902.09469, 2019

    Abram Demski and Scott Garrabrant. Embedded agency.arXiv preprint arXiv:1902.09469, 2019

    work page arXiv 1902

  67. [67]

    Space-time embedded intelligence

    Laurent Orseau and Mark Ring. Space-time embedded intelligence. InInternational Conference on Artificial General Intelligence, pages 209–218. Springer, 2012

    2012

  68. [68]

    Formalizing embeddedness failures in universal artificial intel- ligence.arXiv preprint arXiv:2505.17882, 2025

    Cole Wyeth and Marcus Hutter. Formalizing embeddedness failures in universal artificial intel- ligence.arXiv preprint arXiv:2505.17882, 2025

    work page arXiv 2025

  69. [69]

    Problems of self-reference in self-improving space-time embedded intelligence

    Benja Fallenstein and Nate Soares. Problems of self-reference in self-improving space-time embedded intelligence. InInternational Conference on Artificial General Intelligence, pages 21–32. Springer, 2014. 15

    2014

  70. [70]

    Reflective variants of solomonoff induction and aixi

    Benja Fallenstein, Nate Soares, and Jessica Taylor. Reflective variants of solomonoff induction and aixi. InInternational Conference on Artificial General Intelligence, pages 60–69. Springer, 2015

    2015

  71. [71]

    Markov decision processes with embedded agents

    Luke Harold Miles. Markov decision processes with embedded agents. 2021

    2021

  72. [72]

    The world is bigger! a computationally-embedded perspective on the big world hypothesis

    Alex Lewandowski, Aditya Ramesh, Edan Meyer, Dale Schuurmans, and Marlos C Machado. The world is bigger! a computationally-embedded perspective on the big world hypothesis. Advances in Neural Information Processing Systems, 38:28210–28237, 2026

    2026

  73. [73]

    Towards a generalized theory of observers.arXiv preprint arXiv:2504.16225, 2025

    Hatem Elshatlawy, Dean Rickles, and Xerxes D Arsiwalla. Towards a generalized theory of observers.arXiv preprint arXiv:2504.16225, 2025

    work page arXiv 2025

  74. [74]

    The information theory of individuality.Theory in Biosciences, 139(2):209–223, 2020

    David Krakauer, Nils Bertschinger, Eckehard Olbrich, Jessica C Flack, and Nihat Ay. The information theory of individuality.Theory in Biosciences, 139(2):209–223, 2020

    2020

  75. [75]

    The computational boundary of a “self”: developmental bioelectricity drives multicellularity and scale-free cognition.Frontiers in psychology, 10:493866, 2019

    Michael Levin. The computational boundary of a “self”: developmental bioelectricity drives multicellularity and scale-free cognition.Frontiers in psychology, 10:493866, 2019

    2019

  76. [76]

    Integrated information theory: from consciousness to its physical substrate.Nature reviews neuroscience, 17(7):450– 461, 2016

    Giulio Tononi, Melanie Boly, Marcello Massimini, and Christof Koch. Integrated information theory: from consciousness to its physical substrate.Nature reviews neuroscience, 17(7):450– 461, 2016

    2016

  77. [77]

    When the map is better than the territory.Entropy, 19(5):188, 2017

    Erik P Hoel. When the map is better than the territory.Entropy, 19(5):188, 2017

    2017

  78. [78]

    mit Press, 2004

    Thomas Metzinger.Being no one: The self-model theory of subjectivity. mit Press, 2004

    2004

  79. [79]

    The ego tunnel: The science of mind and the myth of the self, 2012

    Cameron Buckner. The ego tunnel: The science of mind and the myth of the self, 2012

    2012

  80. [80]

    The self as a center of narrative gravity

    Daniel C Dennett. The self as a center of narrative gravity. InSelf and consciousness, pages 103–115. Psychology Press, 2014

    2014

Showing first 80 references.