Pith. sign in

REVIEW 4 major objections 5 minor 46 references

Cognitive maps are generative programs

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read The paper claims that human planning in structured spaces works by representing maps as generative programs and reusing per-fragment plans, which makes search locally optimal and globally suboptimal, resource-efficient, and not…

desk verdict A real behavioral effect with a plausible model, but the fragment inference step is under-specified and the model comparison lacks statistical detail. read the letter →

arxiv 2504.20628 v1 pith:WBTWX34F submitted 2025-04-29 cs.AI cs.ET

Marta Kryven , Cole Wyeth , Aidan Curtis , Kevin Ellis This is my paper · ORCID
classification cs.AIcs.ET
keywords navigationplanningcognitivemapscomputationalmodelinglargelanguagemodelsprograminductionmodularitypartialobservability
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 is trying to establish that human navigation in structured, partially visible spaces is not driven by a full internal map or by shallow planning alone: people effectively store the environment as a generative program that describes repeated map fragments, plan a search policy once per fragment, and reuse that policy whenever the fragment reappears. This would explain why people are efficient in structured real-world environments yet look suboptimal under classical planning models: their plans are locally optimal inside each repeated unit and only globally suboptimal across fragments. To test this, the authors ran a maze-search experiment in which nearly all participants' paths showed a modular signature—visiting repeated fragments in order—across all analyzed mazes, and they built a computational model that recovers fragment programs using large-language-model program synthesis and plans fragment by fragment. The model predicted people's choices better than an optimal planner and a limited-horizon planner, while requiring far less computation and memory. The authors also report that a pilot correlation between cognitive-reflection scores and modularity did not replicate in the final sample.

What carries the argument

The load-bearing object is the generative map program: a compact program, synthesized by prompting a large language model, that reconstructs the observed grid from a small set of fragment units placed by translations, rotations, and reflections. Candidate programs are ranked by Equation 2, a weighted score combining grid-level reconstruction error with a minimum-description-length penalty on fragment complexity and transform counts. The Fragment-based Planning module then solves the partially observable Markov decision process (POMDP) belief-tree search once per fragment and reuses that decision subtree whenever the fragment recurs, using value iteration to move between fragments and falling back to non-modular planning when observations conflict with the reconstructed map. This machinery converts repeated structure into reusable computations, producing the predicted local-optimal/global-suboptimal search signature.

What would settle it

Have participants directly segment the same mazes—mark the repeated chunks they perceive as units—and compare those chunks with the fragments GPT-4 recovers under Equation 2; if people segment the mazes differently on the very maps where the model predicts their search paths, the central claim that human plans track program-like fragments loses its support.

Watch

Extended reading notes

Core claim

The central claim, stated on the paper's own terms, is that human planning in structured environments is predicted by a generative-map model and cannot be explained by unstructured planning with cognitive constraints. A person's cognitive map is not a pixel-by-pixel layout but a small program that regenerates the environment from repeated fragments; the planner computes a policy once for each fragment and reuses it, and interpolates between fragments by solving a local search. Behavioral results show that people choose modular paths that visit all fragments in order in every analyzed maze, matching this signature of locally optimal, globally suboptimal planning, while the optimal expected-utility planner and the discounted-utility planning-horizon model predict non-modular or indifferent paths. The paper presents this as evidence that human deviations from optimality in structured environments arise, at least in part, from reasoning about environmental structure rather than only from limited planning depth.

Load-bearing premise

The load-bearing premise is that the fragment decomposition recovered by GPT-4 program synthesis under the paper's scoring function matches the fragments people actually form in their heads; the authors state this as an assumption that LLM training data gives implicit access to human prior knowledge, and if that link fails, the behavioral fit could be an artifact of scoring weights and prompt design rather than evidence about human cognition.

Editorial extensions

If this is right

  • In structured environments, human planning cost should scale with the number and complexity of distinct fragments rather than with map size, so large regular spaces should be much easier for people than their classical planning complexity suggests.
  • Planning models that ignore environmental structure, such as optimal belief-space search or a fixed discount factor, will systematically mispredict behavior in structured mazes even when they fit behavior in unstructured laboratory tasks.
  • LLM program synthesis can serve as a practical source of human-like priors about spatial chunking, since the recovered fragments are what let the model predict human search better than unstructured planners.
  • The remaining variability in behavior, with people sometimes skipping fragments, points to non-uniform beliefs about which fragments are likely to contain the reward; the paper identifies this as the next extension of the model.

Reading between the lines

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

  • Beyond the paper: if fragment policies are genuinely reusable, a person who has searched one maze containing a given fragment should navigate a second maze containing the same fragment faster, without relearning; this transfer effect is not tested in the paper and would be a direct behavioral check of the reuse mechanism.
  • Beyond the paper: the weighting parameters $w_1, w_2$ and the fragment priors could be fitted per participant, turning the paper's acknowledged variability in modularity into a measurement of individual differences in chunking rather than residual noise.
  • Beyond the paper: an independent probe of human chunking, in which participants freely segment maps into repeated units, could be compared against the fragments the language model recovers; agreement on the same mazes would separate the cognitive-claim contribution from the particular inductive biases of the language model.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper argues that human planning in structured environments is supported by cognitive maps represented as generative programs, in which repeated spatial fragments are discovered and planned over separately, yielding modular and resource-efficient but globally suboptimal behavior. The authors introduce the Generative Modular Planning (GMP) model, which uses GPT4 program synthesis to recover fragment decompositions from maze layouts and plans within each fragment, and they test it against two existing maze-search models using a 30-participant behavioral experiment. They report that people's search paths are predominantly modular according to their definition and that GMP predicts behavior better than Expected Utility and Discounted Utility planning. The central claim, in the paper's own words, is that human planning in structured environments is predicted by the generative map model and cannot be explained by unstructured planning with cognitive constraints.

Significance. If the central claim were fully supported, the paper would make a substantial contribution by connecting program induction, LLM-based priors, hierarchical planning, and behavioral data in a single framework. The experimental design is thoughtful: the mazes were constructed so that the alternative models predict non-modular policies or are indifferent, and the conservative modularity definition works against the authors' hypothesis. The use of GPT4 program synthesis as a bridge to human spatial priors is a novel and potentially valuable methodological step. The paper also provides a concrete efficiency argument for why program-structured maps could be adaptive. However, the strong comparative claim is currently under-supported: no inferential statistics are reported for the model comparison, the free weights in the scoring function are not specified or fitted, the LLM-derived fragment decomposition is not validated against human chunking, and the comparison set lacks simple heuristic baselines. These issues are load-bearing because the behavioral metric is defined relative to the GMM's fragments.

major comments (4)
  1. [Generative Map Module, Eq. (2), and Algorithm 1] The scoring function in Eq. (2) contains two free weights, w1 and w2, but the manuscript never states their values, how they were chosen, or whether the reported results are sensitive to them. Algorithm 1 also leaves the threshold t and the number of completions C unspecified. Because the score determines which fragments GMM treats as programmatic structure, and because modular paths and discriminating decisions are computed relative to those fragments, the downstream model comparison in Fig. 4 can be shaped by these choices. Please report the parameter values, the fitting procedure (if any), and a sensitivity analysis over w1, w2, t, and C.
  2. [Results, Fig. 4] The claim that GMP predicts people 'significantly better' than the optimal planner or the limited-horizon planner is made without reporting any test statistic, p-value, effect size, or confidence interval for the model comparison. Fig. 4c shows per-participant fractions with 95% CIs, but no aggregate statistical test is reported. The assertion that the design makes alternative models 'indifferent between modular and non-modular search' with probability 'at most 0.25' also needs a derivation or supporting analysis. Without these quantitative details, the paper's central comparative claim is not established.
  3. [Background and Results] The paper explicitly relies on the assumption that GPT4 program synthesis 'can implicitly access human prior knowledge of the world embedded in language and code used in LLM training.' This assumption is load-bearing because the behavioral modularity metric is defined relative to the GMM's fragment decomposition: if GPT4's fragments do not correspond to the chunks humans actually use, the behavioral fit could come from the scoring function and prompt design rather than from generative-program cognition. The Discussion does not address this grounding problem. Please validate the fragment decomposition directly (for example, by eliciting human segmentations of the same mazes and measuring agreement with GMM fragments), and include a non-LLM baseline segmentation (such as wall-connected components, uniform tile grids, or an information-gain-based exploration policy) to test whether the predictive advantage comes from generative-program structure specifically.
  4. [Alternative models] The model comparison in the Alternative models section includes only the Expected Utility and Discounted Utility planners. The paper's strong claim that modular planning 'can not be explained by alternative models based on unstructured planning with cognitive constraints' is not supported by this comparison alone, because a simple greedy or nearest-unvisited-cell policy may also produce fragment-ordered search behavior in these mazes. Please add one or more simple heuristic baselines or explicitly weaken the claim to the two models actually tested.
minor comments (5)
  1. [Methods, Procedure] The exclusion of one maze because the exit was 'randomly placed in plain sight' is a post hoc exclusion and should be reported with a robustness check; if the maze is included in the analysis, the modularity results should be shown to be unchanged.
  2. [Fig. 3] The figure caption says the examples have 3, 2, and 5 structural fragments, while the text says 4, 2, and 5; this inconsistency should be corrected.
  3. [Abstract and Introduction] There are several grammatical errors that should be fixed, including 'Our models leverages' in the abstract, 'We makes three contributions' in the Discussion, 'In the the remainder' near Eq. (1), 'perviously' in the Results, and 'Massachussetts' in the author affiliations.
  4. [Discussion] The CRT-modularity regression is reported only as non-significant; please provide the correlation coefficient, sample size, and p-value so the reader can assess the null result, and consider whether the pilot-based sample size is sufficient for this secondary analysis.
  5. [General] The manuscript does not include a data or code availability statement. For reproducibility, please provide the code for GMM/FP, the exact prompts used, and the parameter values for Algorithm 1 and Eq. (2).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the model's fragment decomposition is produced independently by GPT4 program synthesis and tested against external behavioral data.

full rationale

The paper's central chain—LLM-based generative map construction, fragment-based planning, and behavioral comparison—does not reduce to its own inputs. The GMM's fragments are synthesized by GPT4 using an MDL reconstruction score (Eq. 2); no participant data enter the fragment-finding procedure, and the free weights w1,w2 are not reported as fit to the behavioral responses. The 'modular path' metric is explicitly 'consistent with our implementation of GMP,' but this is a faithful operationalization of the hypothesis, not a tautology: participants could have failed to visit GMM's fragments in order, and the paper reports that they did not. The alternative models (Expected Utility and Discounted Utility, gamma = 0.7) are imported from prior work as comparators; the comparison is a test against an external baseline, not a premise that entails GMP's success. The paper openly identifies the load-bearing assumption that LLM program synthesis can implicitly access human prior knowledge through language and code; an unverified assumption is a validity risk, not a circular derivation. There are self-citations (Kryven et al. 2021, 2024), but they provide the task and baseline models and do not themselves assert the generative-program conclusion. No uniqueness theorem, ansatz smuggled through citation, or fitted-parameter-renamed-as-prediction step appears in the derivation.

Assumptions & free parameters 4 free parameters · 7 assumptions · 1 invented entities

The central model depends on a small set of free scoring weights and algorithm thresholds that are not specified, plus explicit assumptions about uniform priors and the LLM's ability to stand in for human priors. The only invented entity is the generative-program cognitive map, which currently lacks independent falsifiable evidence outside the reported experiment.

free parameters (4)
  • w1
    Weight on grid-level similarity in the GMM score (Eq. 2). No value or fitting procedure is reported; changes fragment selection and model predictions.
  • w2
    Weight on program complexity (MDL) penalty in Eq. (2). Same status as w1; central to the model's trade-off.
  • t
    Threshold in Algorithm 1 that stops the search when reconstructed area S' reaches t. Not specified; affects which programs are accepted.
  • C
    Number of LLM completions sampled per iteration in Algorithm 1. Not specified; affects stochastic search and reproducibility.
assumptions (7)
  • standard math POMDP belief-state update and Bellman equation (Eq. 1) describe optimal planning under partial observability.
    Used to define the optimal planner and the within-fragment planner; standard result from Kaelbling et al.
  • domain assumption Exit location is uniformly distributed over valid unseen cells in the initial belief b0.
    Methods, Problem Formulation: 'the exit is uniformly distributed over all valid, unseen cells.' This prior is assumed, not derived from participant behavior.
  • domain assumption The prior probability of finding the exit is uniform across fragments when planning between fragments.
    Fragment-based Planning module: FP 'moves toward the closest fragment ... on the assumption that the prior probability of finding the exit is uniform across fragments.'
  • ad hoc to paper LLM training on human text and code embeds human prior knowledge about spatial structure.
    Background: 'on the assumption that by doing so, we can implicitly access human prior knowledge of the world embedded in language and code used in LLM training.' This is load-bearing for the GMM's fragment proposals.
  • ad hoc to paper Symmetry-based transformations (rotation, reflection, translation) are the relevant structural primitives for human planning.
    Methods, GMM: the generative programs are required to use these transformations because they 'encode symmetries which are relevant to planning.'
  • domain assumption Participants search to minimize time or path length to the exit.
    Methods, Alternative models: path minimization is 'an implicit goal spontaneously reported by participants in the post-experiment questionnaire.'
  • standard math Minimum Description Length is a valid parsimony criterion for choosing among map programs.
    Methods, GMM scoring; based on Rissanen 1978.
invented entities (1)
  • Generative program cognitive map
    purpose: Postulated mental representation that compresses a spatial layout into a small program with repeated fragments and reuses plans across fragments.
    The paper's own experiment is the only evidence offered; there is no independent task, neural measure, or out-of-sample replication that would falsify the construct separately from the GMP model.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Cognitive maps are generative programs." pith.science (2026). https://pith.science/paper/WBTWX34F

@misc{pith2026250420628,
  author       = {Pith},
  title        = {Pith review of: Cognitive maps are generative programs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WBTWX34F}},
  note         = {Machine review of arXiv:2504.20628}
}
read the original abstract

Making sense of the world and acting in it relies on building simplified mental representations that abstract away aspects of reality. This principle of cognitive mapping is universal to agents with limited resources. Living organisms, people, and algorithms all face the problem of forming functional representations of their world under various computing constraints. In this work, we explore the hypothesis that human resource-efficient planning may arise from representing the world as predictably structured. Building on the metaphor of concepts as programs, we propose that cognitive maps can take the form of generative programs that exploit predictability and redundancy, in contrast to directly encoding spatial layouts. We use a behavioral experiment to show that people who navigate in structured spaces rely on modular planning strategies that align with programmatic map representations. We describe a computational model that predicts human behavior in a variety of structured scenarios. This model infers a small distribution over possible programmatic cognitive maps conditioned on human prior knowledge of the world, and uses this distribution to generate resource-efficient plans. Our models leverages a Large Language Model as an embedding of human priors, implicitly learned through training on a vast corpus of human data. Our model demonstrates improved computational efficiency, requires drastically less memory, and outperforms unstructured planning algorithms with cognitive constraints at predicting human behavior, suggesting that human planning strategies rely on programmatic cognitive maps.

Figures

Figures reproduced from arXiv: 2504.20628 by the authors.

Figure 1
Figure 1. Real-world planning tends to occur in predictably [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. (a) The GMP framework. Generative map representations recovered from the input map constitute a list of fragments, [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Examples of maps used in the experiment with 3, [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: People are highly modular, according to the con [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 34 canonical work pages

  1. [1]

    Acquaviva, S., Pu, Y., Kryven, M., Sechopoulos, T., Wong, C., Ecanow, G., Nye, M., Tessler, M., and Tenenbaum, J. (2022). Communicating natural programs to humans and machines. Advances in Neural Information Processing Systems , 35:3731--3743

    work page 2022

  2. [2]

    K., Éltető, N., Griffiths, T

    Binz, M., Akata, E., Bethge, M., Brändle, F., Callaway, F., Coda-Forno, J., Dayan, P., Demircan, C., Eckstein, M. K., Éltető, N., Griffiths, T. L., Haridi, S., Jagadish, A. K., Ji-An, L., Kipnis, A., Kumar, S., Ludwig, T., Mathony, M., Mattar, M., Modirshanechi, A., Nath, S. S., Peterson, J. C., Rmus, M., Russek, E. M., Saanum, T., Scharfenberg, N., Schub...

    work page 2024

  3. [3]

    Bongiorno, C., Zhou, Y., Kryven, M., Theurel, D., Rizzo, A., Santi, P., Tenenbaum, J., and Ratti, C. (2021). Vector-based pedestrian navigation in cities

    work page 2021

  4. [4]

    M., Griffiths, T

    Callaway, F., van Opheusden, B., Gul, S., Das, P., Krueger, P. M., Griffiths, T. L., and Lieder, F. (2022). Rational use of cognitive resources in human planning. Nature Human Behaviour , 6(8):1112--1125

    2022

  5. [5]

    Chandler, J., Mueller, P., and Paolacci, G. (2014). Nonnaivete among amazon mechanical turk workers: Consequences and solutions for behavioral researchers. Behavior research methods , 46(1):112--130

    work page 2014

  6. [6]

    Chollet, F., Knoop, M., Kamradt, G., and Landers, B. (2025). Arc prize 2024: Technical report

    work page 2025

  7. [7]

    G., Sanborn, S., Ho, M

    Correa, C. G., Sanborn, S., Ho, M. K., Callaway, F., Daw, N. D., and Griffiths, T. L. (2023). Exploring the hierarchical structure of human plans via program generation. arXiv preprint arXiv:2311.18644

    arXiv 2023

  8. [8]

    B., Lozano-P \'e rez, T., and Kaelbling, L

    Curtis, A., Silver, T., Tenenbaum, J. B., Lozano-P \'e rez, T., and Kaelbling, L. (2022). Discovering state and action abstractions for generalized task and motion planning. In Proceedings of the AAAI conference on artificial intelligence , volume 36, pages 5377--5384

    work page 2022

Show all 46 references
  1. [9]

    Dehaene, S., Izard, V., Pica, P., and Spelke, E. (2006). Core knowledge of geometry in an amazonian indigene group. Science , 311(5759):381--384

    2006

  2. [10]

    Feldman, J. (2013). Tuning your priors to the world. Topics in cognitive science , 5(1):13--34

    2013

  3. [11]

    Ferreira, D. R. (2013). The impact of the search depth on chess playing strength. ICGA journal , 36(2):67--80

    2013

  4. [12]

    Frederick, S. (2005). Cognitive reflection and decision making. The Journal of Economic Perspectives , 19(4):25--42

    2005

  5. [13]

    Hirtle, S. C. and Jonides, J. (1985). Evidence of hierarchies in cognitive maps. Memory & cognition , 13(3):208--217

    1985

  6. [14]

    K., Abel, D., Correa, C

    Ho, M. K., Abel, D., Correa, C. G., Littman, M. L., Cohen, J. D., and Griffiths, T. L. (2022). People construct simplified mental representations to plan. Nature , 606(7912):129--136

    2022

  7. [15]

    J., Lally, N., Faulkner, P., Eshel, N., Seifritz, E., Gershman, S

    Huys, Q. J., Lally, N., Faulkner, P., Eshel, N., Seifritz, E., Gershman, S. J., Dayan, P., and Roiser, J. P. (2015). Interplay of approximate planning strategies. Proceedings of the National Academy of Sciences , 112(10):3098--3103

    2015

  8. [16]

    G., Dingle, K., Greenbury, S

    Johnston, I. G., Dingle, K., Greenbury, S. F., Camargo, C. Q., Doye, J. P., Ahnert, S. E., and Louis, A. A. (2022). Symmetry and simplicity spontaneously emerge from the algorithmic nature of evolution. Proceedings of the National Academy of Sciences , 119(11):e2113883119

    2022

  9. [17]

    P., Littman, M

    Kaelbling, L. P., Littman, M. L., and Cassandra, A. R. (1998). Planning and acting in partially observable stochastic domains. Artificial intelligence , 101(1-2):99--134

    1998

  10. [18]

    J., and Dayan, P

    Keramati, M., Smittenaar, P., Dolan, R. J., and Dayan, P. (2016). Adaptive integration of habits into depth-limited planning defines a habitual-goal--directed spectrum. Proceedings of the National Academy of Sciences , 113(45):12868--12873

    2016

  11. [19]

    M., Pick Jr, H

    Kosslyn, S. M., Pick Jr, H. L., and Fariello, G. R. (1974). Cognitive maps in children and men. Child development , pages 707--716

    1974

  12. [20]

    D., Cowan, W., and Tenenbaum, J

    Kryven, M., Ullman, T. D., Cowan, W., and Tenenbaum, J. B. (2021). Plans or outcomes: How do we attribute intelligence to others? Cognitive Science , 45(9):13--41

    2021

  13. [21]

    Kryven, M., Yu, S., Kleiman-Weiner, M., Ullman, T., and Tenenbaum, J. (2024). Approximate planning in spatial search. PLOS Computational Biology , 20(11):e1012582

    2024

  14. [22]

    G., Dasgupta, I., Marjieh, R., Hu, M., Hawkins, R

    Kumar, S., Correa, C. G., Dasgupta, I., Marjieh, R., Hu, M., Hawkins, R. D., Cohen, J., Daw, N., Narasimhan, K. R., and Griffiths, T. L. (2022). Using natural language and program abstractions to instill human inductive biases in machines. In Oh, A. H., Agarwal, A., Belgrave, ...

    2022

  15. [23]

    Lake, B. M. and Piantadosi, S. T. (2020). People infer recursive visual concepts from just a few examples. Computational Brain & Behavior , 3(1):54--65

    2020

  16. [24]

    M., Tang, H., Naim, M., Nguyen, D., et al

    Li, W.-D., Hu, K., Larsen, C., Wu, Y., Alford, S., Woo, C., Dunn, S. M., Tang, H., Naim, M., Nguyen, D., et al. (2024). Combining induction and transduction for abstract reasoning. arXiv preprint arXiv:2411.02272

    2024 arXiv

  17. [25]

    G., Behrens, T

    Liu, Y., Mattar, M. G., Behrens, T. E., Daw, N. D., and Dolan, R. J. (2020). Experience replay supports non-local learning. BioRxiv , pages 2020--10

    2020

  18. [26]

    Madani, O., Hanks, S., and Condon, A. (2003). On the undecidability of probabilistic planning and related stochastic optimization problems. Artificial Intelligence , 147(1-2):5--34

    2003

  19. [27]

    F., Casasanto, D., Gibson, E., and Piantadosi, S

    Pitt, B., Ferrigno, S., Cantlon, J. F., Casasanto, D., Gibson, E., and Piantadosi, S. T. (2021). Spatial concepts of number, size, and time in an indigenous culture. Science Advances , 7(33):eabg4141

    2021

  20. [28]

    Rissanen, J. (1978). Modeling by shortest data description. Automatica , 14(5):465--471

    1978

  21. [29]

    Russell, S. J. and Norvig, P. (2016). Artificial intelligence: a modern approach . Pearson

    2016

  22. [30]

    T., and Gershman, S

    Schulz, E., Franklin, N. T., and Gershman, S. J. (2018). Finding structure in multi-armed bandits. bioRxiv

    2018

  23. [31]

    B., Duvenaud, D., Speekenbrink, M., and Gershman, S

    Schulz, E., Tenenbaum, J. B., Duvenaud, D., Speekenbrink, M., and Gershman, S. J. (2017). Compositional inductive biases in function learning. Cognitive psychology , 99:44--79

    2017

  24. [32]

    Sharma, S., Curtis, A., Kryven, M., Tenenbaum, J., and Fiete, I. (2022). Map induction: Compositional spatial submap learning for efficient exploration in novel environments. International Conference of Learning Representations

    2022

  25. [33]

    Silver, D., Schrittwieser, J., Simonyan, K., Antonoglou, I., Huang, A., Guez, A., Hubert, T., Baker, L., Lai, M., Bolton, A., et al. (2017). Mastering the game of go without human knowledge. nature , 550(7676):354--359

    2017

  26. [34]

    Singh, S., James, M., and Rudary, M. (2012). Predictive state representations: A new theory for modeling dynamical systems. arXiv preprint arXiv:1207.4167

    2012 arXiv

  27. [35]

    Spelke, E. S. and Kinzler, K. D. (2007). Core knowledge. Developmental Science , 10(1):89--96

    2007

  28. [36]

    and Coupe, P

    Stevens, A. and Coupe, P. (1978). Distortions in judged spatial relations. Cognitive psychology , 10(4):422--437

    1978

  29. [37]

    S., Precup, D., and Singh, S

    Sutton, R. S., Precup, D., and Singh, S. (1999). Between mdps and semi-mdps: A framework for temporal abstraction in reinforcement learning. Artificial intelligence , 112(1-2):181--211

    1999

  30. [38]

    Tang, H., Key, D., and Ellis, K. (2024). Worldcoder, a model-based llm agent: Building world models by writing code and interacting with the environment. arXiv preprint arXiv:2402.12275

    2024 arXiv

  31. [39]

    Tian, L., Ellis, K., Kryven, M., and Tenenbaum, J. (2020). Learning abstract structure for drawing by efficient motor program induction. Advances in Neural Information Processing Systems , 33:2686--2697

    2020

  32. [40]

    U., De Cola, G., Deleu, T., Goul \ a o, M., Kallinteris, A., Krimmel, M., KG, A., et al

    Towers, M., Kwiatkowski, A., Terry, J., Balis, J. U., De Cola, G., Deleu, T., Goul \ a o, M., Kallinteris, A., Krimmel, M., KG, A., et al. (2024). Gymnasium: A standard interface for reinforcement learning environments. arXiv preprint arXiv:2407.17032

    2024 arXiv

  33. [41]

    M., Rahm, B., Kaller, C

    Unterrainer, J. M., Rahm, B., Kaller, C. P., Leonhart, R., Quiske, K., Hoppe-Seyler, K., Meier, C., M \"u ller, C., and Halsband, U. (2004). Planning abilities and the tower of london: is this task measuring a discrete cognitive function? Journal of clinical and experimental n...

    2004

  34. [42]

    van Opheusden, B., Kuperwajs, I., Galbiati, G., Bnaya, Z., Li, Y., and Ma, W. J. (2023). Expertise increases planning depth in human gameplay. Nature , pages 1000--1005

    2023

  35. [43]

    S., Hutter, M., Uther, W., and Silver, D

    Veness, J., Ng, K. S., Hutter, M., Uther, W., and Silver, D. (2011). A monte-carlo aixi approximation. Journal of Artificial Intelligence Research , 40:95--142

    2011

  36. [44]

    Verhoef, T., Kirby, S., and De Boer, B. (2014). Emergence of combinatorial structure and economy through iterated learning with continuous acoustic signals. Journal of Phonetics , 43:57--68

    2014

  37. [45]

    Wilson, A., Fern, A., and Tadepalli, P. (2012). Transfer learning in sequential decision problems: A hierarchical bayesian approach. In Proceedings of ICML Workshop on Unsupervised and Transfer Learning , pages 217--227. JMLR Workshop and Conference Proceedings

    2012

  38. [46]

    Xie, Y., Yu, C., Zhu, T., Bai, J., Gong, Z., and Soh, H. (2023). Translating natural language to planning goals with large-language models. arXiv preprint arXiv:2302.05128

    2023 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.