Pith. sign in

REVIEW 3 major objections 5 minor 66 references

Towards a Formal Theory of the Need for Competence via Computational Intrinsic Motivation

T0 review · 3 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

Pith's one-line read The paper argues that the four facets of the SDT need for competence—effectance, skill use, task performance, and capacity growth—can each be aligned with existing intrinsic-reward formalisms from reinforcement learning, turning verbal…

desk verdict A useful, honest mapping of SDT competence facets to RL formalisms, but the abstract's 'demonstrate' oversells what is really a plausible, intuitive alignment with no explicit matching criteria. read the letter →

arxiv 2502.07423 v2 pith:SBLFMZXO submitted 2025-02-11 cs.AI

classification cs.AI
keywords needforcompetenceself-determinationtheoryintrinsicmotivationreinforcementlearninggoal-conditionedformalisationprogressempowerment
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 argues that the need for competence in self-determination theory (SDT) can be given a formal footing by pairing each of its four verbal facets with an existing reward formalism from intrinsically motivated reinforcement learning. The four facets are effectance (seeing that one's actions cause change), skill use (finding and recognizing opportunities to use one's capacities), task performance (performing well at intended tasks), and capacity growth (gaining strength or range of skills). If the pairings hold, psychologists can use implemented RL systems as concrete models that simulate competence-driven behaviour and yield empirically testable predictions. The mapping process also exposes preconditions that SDT leaves implicit, such as the agent needing to attribute change to its own actions and to distinguish among its own skills.

What carries the argument

The load-bearing device is the mapping from each verbal competence facet to an intrinsic reward function in goal-conditioned reinforcement learning. A reward function is the scalar signal an agent is built to maximise; intrinsic rewards are task-agnostic signals generated from agent-internal variables such as novelty, impact, empowerment, or learning progress. Each matched formalism rewrites a facet of competence as an optimisation objective: RIDE's impact reward formalises effectance; the variational empowerment objective of VIC and the skill-discrimination objective of DIAYN formalise skill use; the imagined-goal distance of RIG and the learning-progress module selection of CURIOUS formalise task performance; and the multi-time surprise bonus of IMRL and VIC's entropy term formalise capacity growth. These equations turn verbal definitions into executable specifications and expose hidden assumptions.

What would settle it

If human participants in a task environment were asked to choose goals and their choices were compared with the goal-selection probabilities of a model implementing the matched formalism for that facet (for example, CURIOUS for task performance), a robust absence of the predicted preference for intermediate-difficulty goals would falsify the claimed alignment for that facet.

Watch

Extended reading notes

Core claim

The central claim is that the four facets of competence identified in SDT texts—effectance, skill use, task performance, and capacity growth—can each be mapped to distinct existing reward formalisms in intrinsically motivated reinforcement learning. For effectance, RIDE rewards the agent for state changes it can control, with a novelty discount that matches the original effectance requirement of novel observations. For skill use, VIC and DIAYN use variational empowerment and skill discriminability to reward recognising opportunities to deploy skills and recognising that a skill is in use. Task performance is formalised by RIG's negative distance to a self-generated goal and by automatic curriculum learning methods like CURIOUS that select goals of appropriate difficulty. Capacity growth is captured by the intrinsic reward in the IMRL system and by the repertoire-expanding term of VIC. The paper claims this matching demonstrates that computational IM formalisms can plausibly match verbal SDT definitions, and that the process reveals implicit preconditions in SDT such as needing causal attribution and skill discrimination mechanisms.

Load-bearing premise

The alignment rests on the reward hypothesis, adopted from the RL literature without testing for human motivation: that every goal and purpose humans have can be represented as maximising the expected cumulative sum of a scalar reward.

Editorial extensions

If this is right

  • SDT researchers can reuse the existing implementations, simulators, and evaluation metrics from the cited RL systems to run simulation studies of competence-driven behaviour.
  • The paper gives SDT a formal criterion for distinguishing its facets: facets captured by different reward functions are likely distinct constructs, prompting conceptual review.
  • The matched formalisms make hidden preconditions explicit, such as the need for a competence-motivated agent to recognise its own causal impact and to discriminate between its own skills.
  • Learning-progress-based models that cover multiple facets (skill use, task performance, capacity growth) support the coherence of those facets within SDT.
  • The work initiates a theory-development cycle in which computational models of competence are built and then tested empirically to refine SDT.

Reading between the lines

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

  • A direct extension of the matching is to run the matched formalisms in a shared environment and compare their behaviour; the differences the paper notes between impact-driven and learning-progress-driven agents would become experimentally observable.
  • The authors' reliance on the reward hypothesis implies that a psychometric test of whether human competence ratings track the intrinsic reward magnitudes of the matched models would speak to whether these formalisms capture human motivation, a step the paper does not take.
  • The facet-to-formalism mapping could be used to design behavioural experiments that place human participants in the same task environments as the agents, allowing direct comparison of human and agent goal-selection probabilities.
  • If distinct facets map to formally distinct reward functions, the paper's approach gives SDT a principled way to test whether 'competence' is one construct or several.
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

3 major / 5 minor

Summary. The paper addresses the gap between verbal SDT theory and computational formalisms. It takes four competence facets--effectance, skill use, task performance, and capacity growth--from Deterding et al. (2024) and proposes candidate RL intrinsic-reward formalisms for each: RIDE for effectance, learning progress/VIC/DIAYN for skill use, RIG/CURIOUS for task performance, and IMRL/VIC for capacity growth. The authors argue that these alignments reveal hidden preconditions in SDT (e.g., that a competence-motivated agent must recognize its own causal efficacy and distinguish skills) and provide a resource for simulation-based theory testing. The paper is best read as an exploratory mapping rather than a formal derivation; its central claim exceeds the evidence currently presented.

Significance. If the alignments can be made rigorous, the paper would give SDT researchers concrete, implementable starting points and a practical bridge to the computational IM literature. Its concrete strengths include explicit equations for each reward function, direct links to source implementations and evaluation resources, an honest cataloguing of open questions, and a clear proposal for how simulation studies could generate empirically testable predictions. However, the paper does not yet supply a criterion for what counts as a match, and several of its matching choices are underdetermined. Its current value is therefore at the level of hypothesis generation and resource provision rather than formal theory.

major comments (3)
  1. [Matching Computational Models With SDT, Table 1] The paper never defines a criterion for a successful match between a verbal competence facet and a reward function. Each subsection asserts a fit (e.g., 'RIDE offers a simple formalisation' for C1, VIC 'fits the understanding' for C2.a, DIAYN 'shows that recognising which skill is in use ... can drive skill learning' for C2.b), but no property-based test distinguishes a faithful formalization from one with a suggestive name or a single matching term. The paper's own admission that LP 'can simultaneously exemplify' C2.a, C3.b, and C4.a shows that the mappings are not sharp. Absent a matching criterion, the central 'demonstrate' claim is unsupported; the authors should either provide explicit matching criteria or reframe the contribution as candidate alignments.
  2. [Advancing Competence Modelling in SDT] The 'underlying preconditions' are inferred from the same formalisms used for the matching, while the facet taxonomy comes from Deterding et al. (2024), which shares three authors with this paper. This makes the demonstration partially self-confirming: the categories and the features used to validate them originate in the same research program. The revision should address this circularity, for example by using independently derived facets, explicitly positioning the preconditions as interpretive hypotheses, or adding a cross-validation step in which formalisms are selected without knowledge of the facet labels.
  3. [Equations (1), (2), (3), and (6)] The reward functions are not decomposed into facet-specific components, so the same function can serve multiple facets by selecting different terms. For example, RIDE's Eq. (1) mixes an impact term with a novelty denominator; the text uses the impact term for C1 but appeals to novelty only via White (1959). VIC's Eq. (2) contains a discriminability term (alpha) and an entropy term (beta), and the paper attaches alpha to C2.a and beta to C4.b, while C2.b is matched to a reduced version of alpha (Eq. 3). A component-level mapping, such as a table listing each reward term and the facet it expresses, is necessary to make the alignments precise and to test whether the facets are formally distinct.
minor comments (5)
  1. [Equation (2)] Equation (2) contains LaTeX artifacts ('/bracehtipupleft' and similar markers); please typeset the equation cleanly.
  2. [Background: Computational IM & RL] The reward hypothesis is invoked without discussion of its psychological scope; even if it is the standard assumption in computational IM, the paper should state explicitly that the matched formalisms inherit this assumption and that its validity for human motivation is an open empirical question.
  3. [Advancing Competence Modelling in SDT] The observation that a single formalism (LP) simultaneously fits three facets is presented as supporting the theory, but it could equally indicate a lack of discriminant validity among facets; the authors should address this alternative interpretation.
  4. [Section heading for C3.b] The heading 'Specifically to an extent that requires a certain skill or skill level (C3.b)' is awkward; consider rephrasing to 'Performance at a level requiring a certain skill (C3.b)'.
  5. [References] Deterding et al. (2024) is cited as a PsyArXiv preprint; if a published version exists, it should be cited instead or in addition.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the facet taxonomy is an external (if overlapping-author) premise, and the RL formalisms are not fitted to it.

full rationale

The paper makes no empirical predictions and fits no parameters. Its central claim is explicitly conditional and hedged: given the Deterding et al. (2024) facet taxonomy of SDT competence, and given the reward hypothesis from the RL literature, the authors suggest that certain existing intrinsic-reward formalisms can plausibly be aligned with those facets. The facet taxonomy is cited to an overlapping-author prior work, but that prior work is a separate conceptual analysis of canonic SDT texts, externally checkable, and the present paper does not use the citation as a substitute for deriving its matching result. The RL equations (RIDE, VIC, DIAYN, RIG, CURIOUS, IMRL) are taken verbatim from external sources and are not fitted to or derived from the facets. The 'revealed preconditions' are conditional implications of adopting those formalisms, not predictions validated by the matching itself. The acknowledged fact that LP can exemplify multiple facets (C2.a, C3.b, C4.a) shows the matching is underdetermined, but underdetermination is a correctness or precision limitation, not circularity. No equation is shown to equal its input by construction, and no fitted parameter is renamed as a prediction.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No new entities or fitted parameters are introduced. The argument rests on the reward hypothesis, the prior facet taxonomy, and the transfer of algorithmic requirements to human psychology. These assumptions are either explicitly quoted or self-cited.

assumptions (4)
  • domain assumption The reward hypothesis: all goals and purposes can be thought of as maximization of expected cumulative scalar reward.
    Invoked in Background: Computational IM & RL, quoting Sutton and Barto 2018. Underlies every competence formalism matched in the paper.
  • domain assumption The four-facet decomposition of competence from Deterding et al. (2024) is a valid taxonomy.
    The paper uses C1-C4 as given, and several authors overlap with Deterding et al. The distinctness and completeness of the facets are not independently tested here.
  • domain assumption Requirements of the computational formalisms are necessary preconditions for human competence.
    In 'Advancing Competence Modelling in SDT', the paper infers that a competence-motivated agent must recognize action effects and distinguish skills, transferring algorithmic requirements to human psychology.
  • domain assumption Existing RL formalisms from the computational IM literature are reasonable models of human intrinsic motivation.
    The paper cites evidence for some components (e.g., Thompson sampling, empowerment) but does not validate the full mapping against human data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards a Formal Theory of the Need for Competence via Computational Intrinsic Motivation." pith.science (2026). https://pith.science/paper/SBLFMZXO

@misc{pith2026250207423,
  author       = {Pith},
  title        = {Pith review of: Towards a Formal Theory of the Need for Competence via Computational Intrinsic Motivation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SBLFMZXO}},
  note         = {Machine review of arXiv:2502.07423}
}
read the original abstract

Computational modelling offers a powerful tool for formalising psychological theories, making them more transparent, testable, and applicable in digital contexts. Yet, the question often remains: how should one computationally model a theory? We provide a demonstration of how formalisms taken from artificial intelligence can offer a fertile starting point. Specifically, we focus on the "need for competence", postulated as a key basic psychological need within Self-Determination Theory (SDT) -- arguably the most influential framework for intrinsic motivation (IM) in psychology. Recent research has identified multiple distinct facets of competence in key SDT texts: effectance, skill use, task performance, and capacity growth. We draw on the computational IM literature in reinforcement learning to suggest that different existing formalisms may be appropriate for modelling these different facets. Using these formalisms, we reveal underlying preconditions that SDT fails to make explicit, demonstrating how computational models can improve our understanding of IM. More generally, our work can support a cycle of theory development by inspiring new computational models, which can then be tested empirically to refine the theory. Thus, we provide a foundation for advancing competence-related theory in SDT and motivational psychology more broadly.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

66 extracted references · 57 canonical work pages

  1. [1]

    write newline

    " write newline " cite write " FUNCTION editor.postfix editor num.names #1 > "( )" "( )" if FUNCTION editor.trans.postfix editor num.names #1 > "( )" "( )" if FUNCTION trans.postfix translator num.names #1 > "( )" "( )" if FUNCTION authors.editors.reflist.apa5 'field := 'dot := field num.names 'numnames := numnames 'format.num.names := format.num.names na...

  2. [2]

    Aubret, A., Matignon, L., and Hassas, S. (2023). An information-theoretic perspective on intrinsic motivation in reinforcement learning: A survey. Entropy , 25(2)

  3. [3]

    Bagot, L., Mets, K., and Latr \'e , S. (2020). Learning intrinsically motivated options to stimulate policy exploration. In Lifelong Machine Learning Workshop at ICML 2020

  4. [4]

    Baldassarre, G. (2022). Intrinsic motivations for open-ended learning. In Cognitive Robotics , page 251–270. The MIT Press

  5. [5]

    and Agakov, F

    Barber, D. and Agakov, F. (2003). The IM algorithm: a variational approach to information maximization. In Advances in Neural Information Processing Systems , pages 201--208

  6. [6]

    G., Singh, S., and Chentanez, N

    Barto, A. G., Singh, S., and Chentanez, N. (2004). Intrinsically motivated learning of hierarchical collections of skills. In International Conference on Development and Learning , pages 112--120

  7. [7]

    Berlyne, D. (1965). Structure and Direction in Thinking . John Wiley & Sons

  8. [8]

    C., and Polani, D

    Biehl, M., Guckelsberger, C., Salge, C., Smith, S. C., and Polani, D. (2018). Expanding the active inference landscape: More intrinsic motivations in the perception-action loop. Frontiers in Neurorobotics , 12

Show all 66 references
  1. [9]

    J., Tenenbaum, J

    Br\" a ndle, F., Stocks, L. J., Tenenbaum, J. B., Gershman, S. J., and Schulz, E. (2023). Empowerment contributes to exploration behaviour in a creative video game. Nature Human Behaviour , 7(9):1481--1489

  2. [10]

    Burda, Y., Edwards, H., Pathak, D., Storkey, A., Darrell, T., and Efros, A. A. (2019). Large-scale study of curiosity-driven learning. In International Conference on Learning Representations

  3. [11]

    B., Zhao, B., and Niv, Y

    Byers, J. B., Zhao, B., and Niv, Y. (2024). Modeling goal selection with program synthesis. In Intrinsically Motivated Open-ended Learning Workshop at NeurIPS 2024

  4. [12]

    Choi, J., Sharma, A., Lee, H., Levine, S., and Gu, S. S. (2021). Variational empowerment as representation learning for goal-conditioned reinforcement learning. In International Conference on Machine Learning , pages 1953--1963

  5. [13]

    B., and Schulz, L

    Chu, J., Tenenbaum, J. B., and Schulz, L. E. (2024). In praise of folly: flexible goals and human cognition. Trends in Cognitive Sciences , 28(7):628--642

  6. [14]

    Colas, C., Chu, J., Molinaro, G., and Hawkins, R. (2024). What should i do now? goal-centric outlooks on learning, exploration, and communication. In Proceedings of the Annual Meeting of the Cognitive Science Society

  7. [15]

    Colas, C., Fournier, P., Chetouani, M., Sigaud, O., and Oudeyer, P.-Y. (2019). CURIOUS : Intrinsically motivated modular multi-goal reinforcement learning. In International Conference on Machine Learning , pages 1331--1340

  8. [16]

    Colas, C., Karch, T., Sigaud, O., and Oudeyer, P. (2022). Autotelic agents with intrinsically motivated goal-conditioned reinforcement learning: a short survey. Journal of Artificial Intelligence Research , 74:1159--1199

  9. [17]

    M., and Lake, B

    Davidson, G., Todd, G., Togelius, J., Gureckis, T. M., and Lake, B. M. (2024). Goals as reward-producing programs. arXiv preprint arXiv:2405.13242

  10. [18]

    and Balleine, B

    Dayan, P. and Balleine, B. W. (2002). Reward, motivation, and reinforcement learning. Neuron , 36(2):285--298

  11. [19]

    Deci, E. L. and Ryan, R. M. (1985). Intrinsic Motivation and Self-Determination in Human Behavior . Springer US

  12. [20]

    M., Guckelsberger, C., and Ady, N

    Deterding, S., Lintunen, E. M., Guckelsberger, C., and Ady, N. M. (2024). Why self-determination theory needs formal modelling: The case of competence and balanced challenge. PsyArXiv preprint 10.31234/osf.io/n6x8s\_v1

  13. [21]

    Eysenbach, B., Gupta, A., Ibarz, J., and Levine, S. (2019). Diversity is all you need: Learning skills without a reward function. In International Conference on Learning Representations

  14. [22]

    Faldor, M., Zhang, J., Cully, A., and Clune, J. (2024). Omni-epic: Open-endedness via models of human notions of interestingness with environments programmed in code. arXiv preprint arXiv:2405.15568

  15. [23]

    Florensa, C., Held, D., Geng, X., and Abbeel, P. (2018). Automatic goal generation for reinforcement learning agents. In International Conference on Machine Learning , pages 1515--1528

  16. [24]

    Gershman, S. J. (2018). Deconstructing the human algorithms for exploration. Cognition , 173:34--42

  17. [25]

    J., and Wierstra, D

    Gregor, K., Rezende, D. J., and Wierstra, D. (2016). Variational intrinsic control. arXiv preprint arXiv:1611.07507

  18. [26]

    Kaelbling, L. P. (1993). Learning to achieve goals. In International Joint Conference on Artificial Intelligence , pages 1094--8

  19. [27]

    and Dayan, P

    Kakade, S. and Dayan, P. (2002). Dopamine: generalization and bonuses. Neural Networks , 15(4):549--559

  20. [28]

    Lidayan, A., Dennis, M., and Russell, S. (2024). BAMDP shaping: a unified theoretical framework for intrinsic motivation and reward shaping. arXiv preprint arXiv:2409.05358

  21. [29]

    M., White, M., Degris, T., and White, A

    Linke, C., Ady, N. M., White, M., Degris, T., and White, A. (2020). Adapting behavior via intrinsic reward: A survey and empirical study. Journal of Artificial Intelligence Research , 69:1287--1332

  22. [30]

    M., Ady, N

    Lintunen, E. M., Ady, N. M., and Guckelsberger, C. (2024). Diversity progress for goal selection in discriminability-motivated RL . In Intrinsically Motivated Open-ended Learning Workshop at NeurIPS 2024

  23. [31]

    Liu, M., Zhu, M., and Zhang, W. (2022). Goal-conditioned reinforcement learning: Problems and solutions. In International Joint Conference on Artificial Intelligence , pages 5502--5511

  24. [32]

    and Silva, F

    Machado, A. and Silva, F. J. (2007). Toward a richer view of the scientific method: The role of conceptual analysis. American Psychologist , 62(7):671

  25. [33]

    Marsella, S., Gratch, J., and Petta, P. (2010). Computational models of emotion. In A Blueprint for Affective Computing: A sourcebook and manual . Oxford University Press

  26. [34]

    Retrieved from https://www.merriam-webster.com

    Merriam-Webster.com (2025). Retrieved from https://www.merriam-webster.com

  27. [35]

    Molinaro, G., Colas, C., Oudeyer, P.-Y., and Collins, A. (2024). Latent learning progress drives autonomous goal selection in human reinforcement learning. In Advances in Neural Information Processing Systems , pages 32251--32280

  28. [36]

    and Collins, A

    Molinaro, G. and Collins, A. G. (2023). A goal-centric outlook on learning. Trends in Cognitive Sciences , 27(12):1150--1164

  29. [37]

    and Jach, H

    Murayama, K. and Jach, H. K. (2025). A critique of motivation constructs to explain higher-order behavior: We should unpack the black box. Behavioral and Brain Sciences , 48:24

  30. [38]

    V., Pong, V., Dalal, M., Bahl, S., Lin, S., and Levine, S

    Nair, A. V., Pong, V., Dalal, M., Bahl, S., Lin, S., and Levine, S. (2018). Visual reinforcement learning with imagined goals. In Advances in Neural Information Processing Systems , pages 9191--9200

  31. [39]

    and Lewandowsky, S

    Oberauer, K. and Lewandowsky, S. (2019). Addressing the theory crisis in psychology. Psychonomic Bulletin & Review , 26(5):1596--1618

  32. [40]

    and Kaplan, F

    Oudeyer, P.-Y. and Kaplan, F. (2007). What is intrinsic motivation? A typology of computational approaches. Frontiers in Neurorobotics , 1:1--14. Article 6

  33. [41]

    and Kaplan, F

    Oudeyer, P.-Y. and Kaplan, F. (2008). How can we define intrinsic motivation? In International Conference on Epigenetic Robotics

  34. [42]

    Oudeyer, P.-Y., Kaplan, F., and Hafner, V. V. (2007). Intrinsic motivation systems for autonomous mental development. IEEE Transactions on Evolutionary Computation , 11(2):265--286

  35. [43]

    A., and Darrell, T

    Pathak, D., Agrawal, P., Efros, A. A., and Darrell, T. (2017). Curiosity-driven exploration by self-supervised prediction. In International Conference on Machine Learning , pages 2778--2787

  36. [44]

    Pekrun, R. (2024). Overcoming fragmentation in motivation science: Why, when, and how should we integrate theories? Educational Psychology Review , 36(1):27

  37. [45]

    Pong, V., Dalal, M., Lin, S., Nair, A., Bahl, S., and Levine, S. (2020). Skew-fit: State-covering self-supervised reinforcement learning. In International Conference on Machine Learning , pages 7783--7792

  38. [46]

    Portelas, R., Colas, C., Weng, L., Hofmann, K., and Oudeyer, P.-Y. (2021). Automatic curriculum learning for deep rl: a short survey. In International Joint Conference on Artificial Intelligence , pages 4819--4825

  39. [47]

    and Sutton, R

    Precup, D. and Sutton, R. S. (1997). Multi-time models for temporally abstract planning. In Advances in Neural Information Processing Systems , pages 1050--1056

  40. [48]

    and Rockt \" a schel, T

    Raileanu, R. and Rockt \" a schel, T. (2020). RIDE : Rewarding impact-driven exploration for procedurally-generated environments. In International Conference on Learning Representations

  41. [49]

    J., Haslbeck, J

    Robinaugh, D. J., Haslbeck, J. M. B., Ryan, O., Fried, E. I., and Waldorp, L. J. (2021). Invisible Hands and Fine Calipers : A Call to Use Formal Theory as a Toolkit for Theory Construction . Perspectives on Psychological Science , 16(4):725--743

  42. [50]

    Rutherford, A., Beukman, M., Willi, T., Lacerda, B., Hawes, N., and Foerster, J. N. (2024). No regrets: Investigating and improving regret approximations for curriculum discovery. In Advances in Neural Information Processing Systems , pages 16071--16101

  43. [51]

    Ryan, R. M. (2023). The Oxford Handbook of Self-Determination Theory . Oxford University Press

  44. [52]

    Ryan, R. M. and Deci, E. L. (2017). Self-determination theory: Basic psychological needs in motivation, development, and wellness . Guilford

  45. [53]

    Salge, C., Glackin, C., and Polani, D. (2014). Empowerment--an introduction. In Guided Self-Organization: Inception , pages 67--114. Springer Berlin Heidelberg

  46. [54]

    G., Baldassarre, G., and Mirolli, M

    Santucci, V. G., Baldassarre, G., and Mirolli, M. (2012). Intrinsic motivation mechanisms for competence acquisition. In International Conference on Development and Learning and Epigenetic Robotics , pages 1--6

  47. [55]

    Schaul, T., Horgan, D., Gregor, K., and Silver, D. (2015). Universal value function approximators. In International Conference on Machine Learning , pages 1312--1320

  48. [56]

    M., Tiokhin, L., Isager, P

    Scheel, A. M., Tiokhin, L., Isager, P. M., and Lakens, D. (2021). Why hypothesis testers should spend less time testing hypotheses. Perspectives on Psychological Science , 16(4):744--755

  49. [57]

    Schmidhuber, J. (1991). Curious model-building control systems. In IEEE International Joint Conference on Neural Networks , pages 1458--1463

  50. [58]

    Silver, D., Singh, S., Precup, D., and Sutton, R. S. (2021). Reward is enough. Artificial Intelligence , 299

  51. [59]

    Singh, S., Barto, A., and Chentanez, N. (2004). Intrinsically motivated reinforcement learning. In Advances in Neural Information Processing Systems , pages 1281--1288

  52. [60]

    C., Cueva, C

    Sucholutsky, I., Muttenthaler, L., Weller, A., Peng, A., Bobu, A., Kim, B., Love, B. C., Cueva, C. J., Grant, E., Groen, I., Achterberg, J., Tenenbaum, J. B., Collins, K. M., Hermann, K. L., Oktar, K., Greff, K., Hebart, M. N., Cloos, N., Kriegeskorte, N., Jacoby, N., Zhang, Q...

  53. [61]

    Sutton, R. S. and Barto, A. G. (2018). Reinforcement Learning: An Introduction . The MIT Press, second edition

  54. [62]

    Ten, A., Kaushik, P., Oudeyer, P.-Y., and Gottlieb, J. (2021). Humans monitor learning progress in curiosity-driven exploration. Nature communications , 12(1):5972

  55. [63]

    Thompson, W. R. (1933). On the likelihood that one unknown probability exceeds another in view of the evidence of two samples. Biometrika , 25(3–4):285--294

  56. [64]

    L., Robinaugh, D

    Van Dongen, N., van Bork, R., Finnemann, A., Haslbeck, J., van der Maas, H. L., Robinaugh, D. J., de Ron, J., Sprenger, J., and Borsboom, D. (2024). Productive explanation: A framework for evaluating explanations in psychological science. Psychological review

  57. [65]

    Van Rooij, I., Guest, O., Adolfi, F., de Haan, R., Kolokolova, A., and Rich, P. (2024). Reclaiming ai as a theoretical tool for cognitive science. Computational Brain & Behavior , 7(4):616--636

  58. [66]

    White, R. W. (1959). Motivation reconsidered: The concept of competence. Psychological Review , 66(5):297--333

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.