Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

Probing for Consciousness in Machines

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims that a reinforcement-learning agent trained to navigate a virtual dungeon spontaneously forms rudimentary world and self models, evidenced by probes that decode the agent's coordinates from its LSTM activations with…

desk verdict Position decoding is real, but the leap from probe accuracy to 'rudimentary world and self models' is unsupported; this is a modest interpretability result dressed as a consciousness finding. read the letter →

arxiv 2411.16262 v1 pith:5XFVYQOO submitted 2024-11-25 cs.AI q-bio.NC

Mathis Immertreu , Achim Schilling , Andreas Maier , Patrick Krauss This is my paper · ORCID
classification cs.AIq-bio.NC
keywords machineconsciousnesscoreworldmodelselfprobingclassifiersreinforcementlearningLSTMNetHack
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that an artificial agent trained only by reinforcement learning to play a game in a virtual environment develops preliminary forms of a world model and a self model as a byproduct of its task. The evidence comes from probes, small classifiers trained on the agent's neural activations, which can predict the agent's current grid position with accuracy well above chance. A sympathetic reading is that the agent's recurrent memory stores its location, a necessary ingredient for the kind of core consciousness described in the paper's theoretical framework. The authors are careful to say this is preliminary and not a claim that the agent is conscious.

What carries the argument

The central object is the probe, a small feedforward classifier trained on the activation vector of a single layer of a trained network to predict a property of interest. Here, each probe takes the LSTM's hidden or cell state, a 512-dimensional vector, and outputs a score for each of the 225 possible grid coordinates; the readout is accurate far above chance. The LSTM is the load-bearing component, because with only a small crop as input, reconstructing the agent's location requires memory of past observations, which is precisely the kind of internal representation a world model would need.

What would settle it

Train the same probe pipeline on an agent whose LSTM is replaced by a feedforward network with the same crop input, or shuffle the order of observations so that no temporal integration is possible; if probe accuracy remains at the same level, the position signal comes from single observations, not from a memory-based world model. Alternatively, test whether a probe trained on one map can decode positions on a novel map at the same accuracy, which would indicate a general internal model rather than a map-specific code.

Watch

Extended reading notes

Core claim

The paper's central claim is that an RL agent playing MiniHack, a lightweight version of NetHack, ends up representing its own spatial position in the hidden and cell states of its LSTM even though it is never told its coordinates. In three experiments, probes trained on those memory states achieved accuracies of roughly 25% to 67% at predicting the agent's x and y position, against chance levels of 6.7% to 9.1%. The highest accuracies appeared when the agent's visual input was reduced to a 3x3 crop, forcing it to integrate observations over time. The authors interpret this above-chance decodability as evidence of a rudimentary world model, which they take as a step toward the core consciousness described by the theory they use.

Load-bearing premise

The argument rests on the premise that a classifier's ability to decode the agent's position from its memory is evidence of an internal world model, rather than just a convenient encoding that emerged without representing anything about the world.

Editorial extensions

If this is right

  • If the claim holds, supposedly model-free reinforcement learning can still produce implicit internal models as a side effect of maximizing reward.
  • The probing pipeline offers a quantitative, task-agnostic way to look for primitive world models in artificial agents, using only their activations and ground-truth position.
  • The differences between hidden and cell states, and between linear and nonlinear probes, suggest the cell state carries a slightly richer positional code that may merit further study.
  • The paper's distinction between a stable self (the agent is always at the crop's center) and variable external observations gives a concrete direction for separating self models from world models in future experiments.
  • The authors note that the current environments are too simple to test for richer forms of consciousness, so follow-up work should use more complex environments and architectures.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Above-chance probe accuracy alone does not prove a world model; a positional code in the LSTM could yield the same decoding without representing environmental structure. A control that tests whether the probe generalizes to unseen maps, or whether the representation supports planning, would separate a code from a model.
  • The paper's discussion of successor representations suggests a connection: the same discount factor that makes the RL objective convergent also shapes expected future state occupancy, so the LSTM's hidden state may literally encode a predictive map rather than just the current location.
  • The same probing approach could be extended to transformers or to agents with explicit internal-state inputs such as hit points or resources, which would test the self-model half of the theory.
  • A sharper falsification would edit the supposed position code in the hidden state and show that the agent's behavior changes as if it believed it was elsewhere, following the activation-editing style used in the Othello-lineage work the paper cites.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. This manuscript applies linear and nonlinear probe classifiers to the activations of PPO agents trained in MiniHack rooms, with the goal of predicting the agent's current coordinates. Across three experiments—full-map input, 5x5 centered crop, and 3x3 centered crop—the probes achieve above-chance accuracies, such as 59.7% versus a 7.7% chance level in Table III. The authors interpret these results as evidence that the agent has formed rudimentary world and self models, and they link this to Damasio's theory of core consciousness and a possible pathway to machine consciousness.

Significance. The probe methodology is a standard and potentially useful tool, and the paper measures a real decodability signal: the agent's position is indeed linearly and nonlinearly decodable from LSTM activations. The strength is that the probes are evaluated on held-out data and the study spans multiple environment variants. However, the significance is limited by a large gap between the operational result and the interpretive claim. Above-chance coordinate decoding is expected for a task where position is reward-relevant, and it does not establish the existence of a world model in the sense the paper itself defines, namely a representation containing essential structures and dynamics for prediction and planning. The self-model claim is even weaker because no internal or homeostatic variable is probed. The paper could be a starting point for mechanistic studies of spatial encoding in RL agents, but as presented the central conclusion is not supported by the evidence.

major comments (4)
  1. [Figure 2 and Table III] The inference illustrated in Figure 2—that above-chance probe accuracy for the agent's current position implies 'the agent developed a world model'—is not valid. A supervised probe can read out a single spatial feature from recurrent activations without the network storing the essential structures and dynamics required by the paper's own definition of a world model. The random-map accuracy of 59.7% versus a 7.7% chance level is equally consistent with the LSTM maintaining a task-relevant positional code of its recent trajectory and current location. No predictive or structural content is tested, so the central claim goes beyond what the data support.
  2. [Abstract and Discussion] The abstract states that the agent 'can form rudimentary world and self models,' but the experiments probe only external coordinates; no internal state variable such as hitpoints, resources, or reward-derived feelings is measured, so at most a form of self-location is demonstrated, not a self model. The Discussion itself concedes that 'an agent's ability to discern its position may suggest basic core consciousness, but this is not conclusive' and that future research must distinguish self-models from world models. These concessions directly contradict the definitive language of the abstract and the statement in the Discussion that the findings 'robustly confirm' a world model.
  3. [Results, dataset construction] The manuscript does not describe the train/test split as episode-stratified or temporally blocked. LSTM hidden and cell states are highly autocorrelated across consecutive time steps; if samples from the same episode appear in both the training and test sets, a probe can exploit time-neighbor similarity rather than learn a generalizable position code. The paper should state whether entire episodes were held out, and ideally report probe accuracy on held-out episodes or with temporal blocking. This is a load-bearing issue for the decodability result itself, not merely for the world-model interpretation.
  4. [Tables I-III] The results are reported as single accuracies without confidence intervals, significance tests, or information about the number of training seeds. Because chance levels differ across tables due to excluded edge rows and because the paper interprets differences among maps and between hidden and cell states, the absence of variability measures makes the comparisons unreliable. At minimum, the authors should provide multiple seeds with error bars or permutation-based chance intervals, and a statistical test for the 'significantly higher than chance' claim in Figure 2.
minor comments (4)
  1. [Abstract and Discussion] The language should be hedged to match the evidence; 'demonstrate' and 'robustly confirm' are too strong for single accuracies without statistical support.
  2. [Methods, Equations (1)-(3)] There are minor typesetting issues, such as the use of 'T' both as the episode length and in the summation, and the rendering of umlauts in author names appears corrupted in places.
  3. [Methods, Probes] The probe training details are incomplete: the manuscript should specify input normalization, optimization details for each experiment, the number of probe parameters, and the exact procedure for splitting the 230,000 samples into 200,000 training and 30,000 test instances.
  4. [Discussion] The paragraph referring to citation [25] is poorly integrated; it is unclear how that citation supports or contrasts with the preceding distinction between self-models and world models.

Circularity Check

1 steps flagged · score 2.0 of 10

Probe measurements are not circular, but the paper's world-model conclusion rests on an interpretive leap and a mild self-citation for the theoretical framework.

  1. self citation load bearing [Introduction, paragraph 4]
    "In [7] it is argued that this theory is uniquely well-suited for application to artificial intelligence (AI) and machine learning (ML) systems."

    Reference [7] is Krauss and Maier (2020), authored by two of the present authors. The paper's interpretive premise—that Damasio's theory and its self/world-model constructs are the appropriate lens for interpreting probe results—is imported from the authors' own prior argument rather than from an independent, externally verified source. The probe accuracies themselves are genuinely measured on held-out data, so this self-citation does not force the numerical results; it is a load-bearing appeal to the authors' own framework for the central conceptual claim.

full rationale

The empirical chain is not circular: a probe is trained on 200,000 samples and tested on 30,000 held-out instances, and the reported accuracies (e.g., 59.7% vs 7.7% chance in Table III) are real decodability signals. The central weakness is construct validity, not circularity: the paper's own definition of a world model—'an internal representation containing essential structures and dynamics for the agent'—is never directly tested; a position decoder measures only that spatial location is readable from LSTM states. The strong conclusion in the abstract and Figure 2 ('Thus the agent developed a world model') is an interpretive step that goes beyond the measured quantity. The Discussion itself concedes the first experiment is ambiguous, states that 'An agent's ability to discern its position may suggest basic core consciousness, but this is not conclusive,' and admits that 'Future research will have to distinguish between self-models and world models.' These limitations are acknowledged, so the paper does not hide the gap; it simply labels an over-reaching conclusion. The only circularity-adjacent issue is the self-citation [7] that supplies the Damasio framework, and this does not compromise the independent probe measurements.

Assumptions & free parameters 3 free parameters · 4 assumptions · 2 invented entities

The paper's central result depends on hand-chosen experimental design choices (crop size, edge exclusion, LSTM size) and on a chain of interpretive axioms. The only measured quantity is probe accuracy on position. The world model and self model labels are constructs added on top of that measurement, and the paper provides no independent evidence for them. This is why the ledger lists them as axioms and invented entities rather than as derived results.

free parameters (3)
  • Observation crop size = 9x9 in first experiment, 5x5 in second, 3x3 in third
    Chosen by hand; the conclusions about memory and indirect position inference depend on this choice.
  • Edge-row exclusion width = 2 rows/cols in second experiment, 1 row/col in third
    Post hoc data selection intended to block direct visual localization; it changes the chance level and the measured accuracies.
  • LSTM hidden and cell state size = 512
    Architecture hyperparameter chosen by hand; the paper attributes the position information to the LSTM memory.
assumptions (4)
  • domain assumption Damasio's theory of consciousness, specifically that core consciousness requires a self model and a world model, is accepted as the interpretive framework.
    The entire research question and the meaning of the probe results are framed in terms of Damasio's levels of consciousness as cited in [6] and [7].
  • domain assumption Above-chance probe accuracy implies that the feature is present in the network's activations.
    This is the standard probing assumption from [13] and [14], used without additional validation in this setting.
  • ad hoc to paper Decodability of current position from activations implies the agent has a world model.
    The Discussion states that position encoding 'indicate[s] that the agent has developed a world model'; no test of prediction, structure, or dynamics is provided.
  • domain assumption The small crop and edge exclusions prevent the agent from inferring its position directly from observations.
    Plausible but not verified; local glyphs such as staircases, traps, and monsters can still carry position-correlated information.
invented entities (2)
  • Implicit world model attributed to the RL agent
    purpose: Explains above-chance probe accuracy as a sign that the agent internally represents the environment's structure.
    No falsifiable handle outside the paper's probe setup; no prediction of future observations, no causal intervention, and no definition of the predicted 'structured dynamics.'
  • Rudimentary self model attributed to the agent
    purpose: Supports the abstract's claim that the agent forms both world and self models.
    No internal-state variables, no body-state input, and no measure of a self representation are present in the experiments. The term is used without any direct evidence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Probing for Consciousness in Machines." pith.science (2026). https://pith.science/paper/5XFVYQOO

@misc{pith2026241116262,
  author       = {Pith},
  title        = {Pith review of: Probing for Consciousness in Machines},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5XFVYQOO}},
  note         = {Machine review of arXiv:2411.16262}
}
read the original abstract

This study explores the potential for artificial agents to develop core consciousness, as proposed by Antonio Damasio's theory of consciousness. According to Damasio, the emergence of core consciousness relies on the integration of a self model, informed by representations of emotions and feelings, and a world model. We hypothesize that an artificial agent, trained via reinforcement learning (RL) in a virtual environment, can develop preliminary forms of these models as a byproduct of its primary task. The agent's main objective is to learn to play a video game and explore the environment. To evaluate the emergence of world and self models, we employ probes-feedforward classifiers that use the activations of the trained agent's neural networks to predict the spatial positions of the agent itself. Our results demonstrate that the agent can form rudimentary world and self models, suggesting a pathway toward developing machine consciousness. This research provides foundational insights into the capabilities of artificial agents in mirroring aspects of human consciousness, with implications for future advancements in artificial intelligence.

Figures

Figures reproduced from arXiv: 2411.16262 by the authors.

Figure 1
Figure 1. Simplified representation of Damasio’s model of consciousness (taken from [7]): The protoself operates at an unconscious level, processing emotions and sensory input. Core consciousness emerges from the protoself, creating the initial self and world models, allowing the self to relate to its environment. Projections of emotions evolve into higher-order feelings. With access to memory and the integration of complex f… view at source ↗
Figure 3
Figure 3. The basic agent-environment interaction cycle: The agent observes the current state/observation st/ot, decides on an action at based on its policy and the environment reacts to this action by returning a reward rt and the next state/observation st+1/ot+1 beginning the next cycle. for trajectories τ = (s0, a0, r0, s1, a1, r1..., sT +1) ∈ T , the space of trajectories, a discount factor γ ∈ [0, 1] and a reward functio… view at source ↗
Figure 2
Figure 2. Schematized overview of the approach: 1.) An agent is trained with RL 2.) A dataset of the trained agent’s position and neural network’s activations is sampled 3.) Using this dataset on each layer’s activations a probe is trained to predict the true position. 4.) If one of the probes can predict the true agent position (with an accuracy significantly higher than chance), it shows that the necessary information is co… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: An example of the ultimate type map: The small figure is the agent, the staircases up and down are the start and goal, respectively, the eyes show uncovered teleportation traps and the bones are remains of a defeated monster. The dark grey areas have already been visit…
Figure 5
Figure 5. Figure 5: A simplified illustration of the agent architecture.: The map (dashed circle) was only included as input in the first experiment and the LSTM cell (dotted rounded rectangle) became part of the architecture from the second experiment onward. Training RLlib [24] was used…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Probing Internal Representations of Multi-Word Verbs in Large Language Models

    cs.CL 2025-02 reject novelty 4.0 of 10

    A study claims BERT stores phrasal and prepositional verbs in a non-linearly separable way, but its own linear classifiers separate the two categories almost perfectly.

Reference graph

Works this paper leans on

35 extracted references · 23 canonical work pages · cited by 1 Pith paper

  1. [1]

    I.—COMPUTING MACHINERY AND INTELLIGENCE,

    A. M. TURING, “I.—COMPUTING MACHINERY AND INTELLIGENCE,” Mind, vol. LIX, no. 236, pp. 433–460, 10

    work page

  2. [2]

    People cannot distinguish gpt- 4 from a human in a turing test,

    C. R. Jones and B. K. Bergen, “People cannot distinguish gpt- 4 from a human in a turing test,” 2024. [Online]. Available: https://arxiv.org/abs/2405.08007

    arXiv 2024

  3. [3]

    Minds, brains, and programs,

    J. R. Searle, “Minds, brains, and programs,” Behavioral and Brain Sciences, vol. 3, no. 3, p. 417–424, 1980

    work page 1980

  4. [4]

    Integrated information theory: from consciousness to its physical substrate,

    G. Tononi, M. Boly, M. Massimini, and C. Koch, “Integrated information theory: from consciousness to its physical substrate,” Nature Reviews Neuroscience , vol. 17, no. 7, pp. 450–461, Jul 2016. [Online]. Available: https://doi.org/10.1038/nrn.2016.44

    doi:10.1038/nrn.2016.44 2016

  5. [5]

    A global workspace theory of conscious experience,

    B. J. Baars, “A global workspace theory of conscious experience,” in Consciousness in philosophy and cognitive neuroscience . Psychology Press, 2013, pp. 161–184

    work page 2013

  6. [6]

    Consciousness: An overview of the phe- nomenon and of its possible neural basis,

    A. Damasio and K. Meyer, “Consciousness: An overview of the phe- nomenon and of its possible neural basis,” The neurology of conscious- ness: Cognitive neuroscience and neuropathology , pp. 3–14, 2009

    work page 2009

  7. [7]

    Will we ever have conscious machines?

    P. Krauss and A. Maier, “Will we ever have conscious machines?” Fron- tiers in Computational Neuroscience, vol. 14, 2020. [Online]. Available: https://www.frontiersin.org/articles/10.3389/fncom.2020.556544

    work page arXiv 2020

  8. [8]

    Deep reinforcement learning for robotic manipulation with asynchronous off-policy updates,

    S. Gu, E. Holly, T. Lillicrap, and S. Levine, “Deep reinforcement learning for robotic manipulation with asynchronous off-policy updates,” in 2017 IEEE International Conference on Robotics and Automation (ICRA), 2017, pp. 3389–3396

    work page 2017

Show all 35 references
  1. [9]

    Magnetic control of tokamak plasmas through deep reinforcement learning,

    J. Degrave, F. Felici, J. Buchli, M. Neunert, B. Tracey, F. Carpanese, T. Ewalds, R. Hafner, A. Abdolmaleki, D. de las Casas, C. Donner, L. Fritz, C. Galperti, A. Huber, J. Keeling, M. Tsimpoukelli, J. Kay, A. Merle, J.-M. Moret, S. Noury, F. Pesamosca, D. Pfau, O. Sauter, C. ...

    2022

  2. [10]

    Playing atari with deep reinforcement learning,

    V . Mnih, K. Kavukcuoglu, D. Silver, A. Graves, I. Antonoglou, D. Wierstra, and M. A. Riedmiller, “Playing atari with deep reinforcement learning,” CoRR, vol. abs/1312.5602, 2013. [Online]. Available: http://arxiv.org/abs/1312.5602

    2013 arXiv

  3. [11]

    Emergent world representations: Exploring a sequence model trained on a synthetic task,

    K. Li, A. K. Hopkins, D. Bau, F. Vi ´egas, H. Pfister, and M. Wattenberg, “Emergent world representations: Exploring a sequence model trained on a synthetic task,” arXiv preprint arXiv:2210.13382 , 2022

    2022 arXiv

  4. [12]

    R. S. Sutton and A. G. Barto, Reinforcement Learning: An Introduction, 2nd ed. The MIT Press, 2018. [Online]. Available: http://incompleteideas.net/book/the-book-2nd.html

    2018

  5. [13]

    Understanding intermediate layers using linear classifier probes,

    G. Alain and Y . Bengio, “Understanding intermediate layers using linear classifier probes,” 2018

    2018

  6. [14]

    Probing classifiers: Promises, shortcomings, and advances,

    Y . Belinkov, “Probing classifiers: Promises, shortcomings, and advances,” Computational Linguistics , vol. 48, no. 1, pp. 207–219, Mar. 2022. [Online]. Available: https://aclanthology.org/2022.cl-1.7

    2022

  7. [15]

    The NetHack Learning Envi- ronment,

    H. K ¨uttler, N. Nardelli, A. H. Miller, R. Raileanu, M. Selvatici, E. Grefenstette, and T. Rockt ¨aschel, “The NetHack Learning Envi- ronment,” in Proceedings of the Conference on Neural Information Processing Systems (NeurIPS) , 2020

    2020

  8. [16]

    Minihack the planet: A sandbox for open-ended reinforcement learning research,

    M. Samvelyan, R. Kirk, V . Kurin, J. Parker-Holder, M. Jiang, E. Hambro, F. Petroni, H. Kuttler, E. Grefenstette, and T. Rockt ¨aschel, “Minihack the planet: A sandbox for open-ended reinforcement learning research,” in Thirty-fifth Conference on Neural Information Processing ...

    2021

  9. [17]

    Insights from the neurips 2021 nethack challenge,

    E. Hambro, S. Mohanty, D. Babaev, M. Byeon, D. Chakraborty, E. Grefenstette, M. Jiang, J. Daejin, A. Kanervisto, J. Kim et al. , “Insights from the neurips 2021 nethack challenge,” in NeurIPS 2021 Competitions and Demonstrations Track . PMLR, 2022, pp. 41–52

    2021

  10. [18]

    Motif: Intrinsic motivation from artificial intelligence feedback,

    M. Klissarov, P. D’Oro, S. Sodhani, R. Raileanu, P.-L. Bacon, P. Vincent, A. Zhang, and M. Henaff, “Motif: Intrinsic motivation from artificial intelligence feedback,” arXiv preprint arXiv:2310.00166 , 2023

    2023 arXiv

  11. [19]

    Evolving curricula with regret- based environment design,

    J. Parker-Holder, M. Jiang, M. Dennis, M. Samvelyan, J. Foerster, E. Grefenstette, and T. Rockt ¨aschel, “Evolving curricula with regret- based environment design,” in International Conference on Machine Learning. PMLR, 2022, pp. 17 473–17 498

    2022

  12. [20]

    Learning to query internet text for informing reinforcement learning agents,

    K. Nottingham, A. Pyla, S. Singh, and R. Fox, “Learning to query internet text for informing reinforcement learning agents,” arXiv preprint arXiv:2205.13079, 2022

    2022 arXiv

  13. [21]

    Skillhack: A benchmark for skill transfer in open- ended reinforcement learning,

    M. Matthews, M. Samvelyan, J. Parker-Holder, E. Grefenstette, and T. Rockt ¨aschel, “Skillhack: A benchmark for skill transfer in open- ended reinforcement learning,” in ICLR Workshop on Agent Learning in Open-Endedness, 2022

    2022

  14. [22]

    Oracle-sage: Planning ahead in graph-based deep reinforcement learning,

    A. Chester, M. Dann, F. Zambetta, and J. Thangarajah, “Oracle-sage: Planning ahead in graph-based deep reinforcement learning,” in Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Springer, 2022, pp. 52–67

    2022

  15. [23]

    Cora: Bench- marks, baselines, and metrics as a platform for continual reinforcement learning agents,

    S. Powers, E. Xing, E. Kolve, R. Mottaghi, and A. Gupta, “Cora: Bench- marks, baselines, and metrics as a platform for continual reinforcement learning agents,” in Conference on Lifelong Learning Agents . PMLR, 2022, pp. 705–743

    2022

  16. [24]

    RLlib: Abstractions for distributed reinforcement learning,

    E. Liang, R. Liaw, R. Nishihara, P. Moritz, R. Fox, K. Goldberg, J. Gonzalez, M. Jordan, and I. Stoica, “RLlib: Abstractions for distributed reinforcement learning,” in Proceedings of the 35th International Conference on Machine Learning , ser. Proceedings of Machine Learning ...

    2018

  17. [25]

    Homeostatic feelings and the biology of consciousness,

    A. Damasio and H. Damasio, “Homeostatic feelings and the biology of consciousness,” Brain, vol. 145, no. 7, pp. 2231–2235, 05 2022. [Online]. Available: https://doi.org/10.1093/brain/awac194

    2022 doi

  18. [26]

    The hip- pocampus as a predictive map,

    K. L. Stachenfeld, M. M. Botvinick, and S. J. Gershman, “The hip- pocampus as a predictive map,” Nature neuroscience, vol. 20, no. 11, pp. 1643–1653, 2017

    2017

  19. [27]

    The successor representation: its computational logic and neural substrates,

    S. J. Gershman, “The successor representation: its computational logic and neural substrates,” Journal of Neuroscience , vol. 38, no. 33, pp. 7193–7200, 2018

    2018

  20. [28]

    Neural network based successor representations to form cognitive maps of space and language,

    P. Stoewer, C. Schlieker, A. Schilling, C. Metzner, A. Maier, and P. Krauss, “Neural network based successor representations to form cognitive maps of space and language,” Scientific Reports, vol. 12, no. 1, p. 11233, 2022

    2022

  21. [29]

    Neural network based formation of cognitive maps of semantic spaces and the putative emergence of abstract concepts,

    P. Stoewer, A. Schilling, A. Maier, and P. Krauss, “Neural network based formation of cognitive maps of semantic spaces and the putative emergence of abstract concepts,” Scientific Reports , vol. 13, no. 1, p. 3644, 2023

    2023

  22. [30]

    Conceptual cognitive maps formation with neural successor net- works and word embeddings,

    ——, “Conceptual cognitive maps formation with neural successor net- works and word embeddings,” in 2023 IEEE International Conference on Development and Learning (ICDL) . IEEE, 2023, pp. 391–395

    2023

  23. [31]

    Multi-modal cognitive maps based on neural networks trained on successor representations,

    ——, “Multi-modal cognitive maps based on neural networks trained on successor representations,” arXiv preprint arXiv:2401.01364 , 2023

    2023 arXiv

  24. [32]

    Word class representations spontaneously emerge in a deep neural network trained on next word prediction,

    K. Surendra, A. Schilling, P. Stoewer, A. Maier, and P. Krauss, “Word class representations spontaneously emerge in a deep neural network trained on next word prediction,” in 2023 International Conference on Machine Learning and Applications (ICMLA) . IEEE, 2023, pp. 1481– 1486

    2023

  25. [33]

    The successor representation in human reinforcement learning,

    I. Momennejad, E. M. Russek, J. H. Cheong, M. M. Botvinick, N. D. Daw, and S. J. Gershman, “The successor representation in human reinforcement learning,” Nature human behaviour , vol. 1, no. 9, pp. 680–692, 2017

    2017

  26. [34]

    Reinforcement learning, fast and slow,

    M. Botvinick, S. Ritter, J. X. Wang, Z. Kurth-Nelson, C. Blundell, and D. Hassabis, “Reinforcement learning, fast and slow,”Trends in cognitive sciences, vol. 23, no. 5, pp. 408–422, 2019

    2019

  27. [1950]

    Available: https://doi.org/10.1093/mind/LIX.236.433

    [Online]. Available: https://doi.org/10.1093/mind/LIX.236.433

    doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.