Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Experience-driven discovery of planning strategies

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read People discover new planning strategies by learning which mental operations pay off, not just by choosing among familiar ones.

desk verdict A solid empirical demonstration of experience-driven strategy discovery, but the central mechanistic claim rests on a tuned simulation and a hand-crafted feature space, so it is not yet established. read the letter →

arxiv 2412.03111 v1 pith:VKZV46EK submitted 2024-12-04 cs.AI

Ruiqi He , Falk Lieder This is my paper · ORCID
classification cs.AI
keywords metacognitivelearningplanningstrategydiscoverycognitivemodelingreinforcementMouselab-MDPresourcerationality
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

This paper asks where planning strategies come from, not just how people choose among strategies they already have. It proposes that new strategies are discovered through metacognitive reinforcement learning: people treat mental operations, such as clicking to reveal a node in a planning tree, as actions and update preferences for those operations based on the rewards they bring. In a custom Mouselab-MDP planning task where the resource-rational strategy was deliberately unfamiliar, the proportion of participants using the adaptive strategy rose from 0.79% in the first trial to 28.57% by trial 120, and scores rose accordingly. When fitted to individual click sequences, metacognitive reinforcement learning models explained more participants than strategy-selection learning, value-free mental habit formation, or no learning, with 61% best explained by the reinforcement mechanism and the hybrid variant preferred within it. The models nevertheless discovered the strategy more slowly than people did, which the authors attribute to missing "Eureka"-style insight learning and possibly to a feature representation that does not cover every intermediate strategy.

What carries the argument

The central object is the metacognitive reinforcement learning model, specifically the Reinforce algorithm applied to a meta-level Markov decision process whose states are mental belief states and whose actions are mental computations such as revealing a node. Planning strategies are represented by 63 features of belief states and cognitive operations, with weights approximating meta-level Q-values; the next planning operation is chosen via a softmax over those Q-values, and after each trial the weights are updated by the Reinforce gradient-ascent rule using the return of that trial's planning episode. The load-bearing mechanism is this gradient ascent in strategy space: starting from an arbitrary initial weighting, repeated updates should move the model toward the resource-rational strategy, and the model comparison then tests whether human click sequences follow the same update rule.

What would settle it

Simulate click sequences from a known mental-habit model with an abrupt switch to the adaptive strategy, then run the paper's Bayesian model selection on those simulated data; if the procedure attributes a majority of these simulations to hybrid Reinforce, the reported preference for metacognitive RL over habit learning is not diagnostic.

Watch

Extended reading notes

Core claim

The paper claims that experience-driven strategy discovery in planning is best explained by metacognitive reinforcement learning (MCRL), in which the brain learns to approximate an optimal meta-level policy by updating feature weights attached to cognitive operations. To test this, the authors designed a Mouselab-MDP environment whose resource-rational strategy—check immediate nodes until the branch with a positive start is identified, then inspect exactly one outer node on that branch—was unlikely to be in participants' pre-existing mental toolboxes. Across 120 trials, the proportion of participants using the adaptive strategy increased significantly, and Bayesian model selection over click sequences favored the hybrid Reinforce model for about 62% of the participants best described by a Reinforce variant, with rational strategy selection learning explaining under 1% overall. The authors conclude that metacognitive reinforcement learning is a key mechanism for discovering new planning strategies, while acknowledging that their models have a slower discovery rate than humans and that individual differences exist, with about 39% of participants better described by a value-free mental habit model.

Load-bearing premise

The load-bearing assumption is that the 63-feature strategy space (with seven features added for this task) can represent every intermediate strategy participants actually tried; if a real intermediate strategy falls outside this representation, the model can only discover strategies within its own vocabulary and the comparison loses force.

Editorial extensions

If this is right

  • Strategy discovery is experience-driven: the adaptive strategy was almost never present on trial 1 (0.79%) but was used by 28.57% of participants by trial 120.
  • Metacognitive reinforcement learning is a better explanation of strategy discovery than choosing among pre-existing strategies: rational strategy selection learning explained under 1% of participants, while the Reinforce mechanism accounted for about 61%.
  • Strategy acquisition is not a single mechanism for everyone: about 39% of participants were better explained by value-free mental habit formation than by learning from reward feedback.
  • Current MCRL models under-predict discovery speed, so a full account of human strategy discovery must include something like insight-driven "Eureka" jumps or active learning.
  • Because the models learn from feedback on mental computations, interventions aimed at improving planning should target the value of thinking itself, not only the value of external actions.

Reading between the lines

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

  • Editorial inference: if metacognitive reinforcement learning generalizes beyond this task, training environments could be designed so that the cost structure of mental operations is discoverable, potentially accelerating planning-skill acquisition in education and AI.
  • Editorial inference: the slower-model-faster-human gap suggests a low-probability "insight" sampler that reinitializes feature weights when cumulative reward is poor would improve the model; this is a testable extension the paper does not implement.
  • Editorial inference: the 21 habitual participants who examined all nodes early and then snapped to the optimal strategy could be reinterpreted as performing deliberate exploration first; modeling this as an explore-then-exploit meta-policy is a natural extension that current gradual-learning models cannot capture.
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

3 major / 5 minor

Summary. The paper proposes that new planning strategies are discovered via metacognitive reinforcement learning (MCRL). It introduces a Mouselab-MDP experiment in which participants must learn a task-specific resource-rational (RR) strategy, shows that the proportion of adaptive strategies increases significantly over 120 trials, and then fits two MCRL variants (hybrid and model-free Reinforce) against alternative accounts (RSSL, mental habit formation, and a non-learning model) using family-level Bayesian model selection. The BMS results favor learning models over non-learning, and Reinforce over RSSL and mental habit, with the hybrid variant best among the Reinforce models. The paper also demonstrates, via simulations, that the Reinforce models can in principle discover the adaptive strategy, but acknowledges that when fitted to human data the models discover at a slower rate than participants.

Significance. If the central claim holds, the paper would provide a mechanistic account of how new planning heuristics enter the mental toolbox, extending prior work on strategy selection to strategy discovery. The experimental design is a clear contribution: it creates a novel, task-specific RR strategy that participants are unlikely to know in advance, and the empirical trend (rising proportion of adaptive strategies) is supported by both a logistic regression and a non-parametric Mann-Kendall test. The paper also reports model comparison using random-effects BMS and gives credit for the honest discussion of the models' slower discovery rate and the possible role of 'Eureka' moments. However, the significance is tempered by two load-bearing concerns: the capability simulations use hyperparameters optimized to produce the optimal click sequence, and the seven task-specific features added to the strategy representation are not described in the manuscript, so the representational envelope may be engineered to contain the target strategy. These issues limit the strength of the evidence for MCRL as the actual mechanism of human strategy discovery.

major comments (3)
  1. [Capability to represent and to discover the novel planning strategy (Figure 3)] The claim that the Reinforce models 'are capable of discovering the adaptive strategy from scratch' is not supported by the simulations as reported, because the hyperparameters were 'found by optimizing for the optimal click sequence.' This is a circular step: the simulation is tuned to produce the target sequence, so convergence to that sequence demonstrates representational capacity and favorable tuning, not discovery from experience. To support the discovery claim, the authors should show that the model discovers the strategy for a broad, principled range of hyperparameters (e.g., random search, default values, or cross-validated fits) and that the optimized hyperparameters are not special.
  2. [Strategy representation (paragraph on the 63-feature set)] The manuscript does not describe the seven new features added for this task; it merely states that they are 'essential to capture preferences and avoidance related to specific node levels' and provides an OSF link. Because the target RR strategy specifically involves examining immediate nodes and one outer node on the positive branch, these features could, in the worst case, directly encode the target strategy. This concern is load-bearing for both the capability demonstration and the model comparison: if the feature space is engineered to contain the target, then the BMS advantage of Reinforce over the mental habit model (which shares the same features) may reflect the flexibility of weight updating in a hand-crafted space rather than metacognitive value learning. The authors should list the seven features in the paper, explain what they capture, and ideally show that the target strategy is not representable by a small subset of them.
  3. [Model fitting and model selection (Tables 3 and 4)] The mental habit model is implemented as a Reinforce model that uses the same features as the model-free variant but does not update the feature weights. Thus the Level-2 BMS contrast (Reinforce vs. mental habit) reduces to updating versus not updating the same feature weights. This comparison does not isolate metacognitive value learning as the mechanism; any time-varying weight model could fit better because it can track changes in behavior. To strengthen the mechanistic interpretation, the authors should add a control model that updates weights without value-based reinforcement (e.g., random drift of weights) or otherwise demonstrate that the value-based update rule is essential for the empirical advantage.
minor comments (5)
  1. [Equation (2)] Equation (2) sums over 56 features, whereas the text states that the feature set has 63 features; please reconcile this discrepancy.
  2. [Empirical results (first-trial proportion)] The text reports that the proportion of adaptive strategies increased from 0.79% in the first trial to 28.57% in the last trial, but the reported 95% confidence interval for the first trial is [0; 2.22%], which includes zero; the authors should acknowledge this when interpreting the starting point.
  3. [Materials (adaptive strategy definition)] The classification of a click sequence as 'adaptive' is described only in prose; please provide a precise, algorithmic definition (e.g., exact conditions on click orders and counts) so that the classification is reproducible.
  4. [References] Reference 'He, Jain, & Lieder, 2021b' has a typo in the title: 'planning strategies the to structure of the environment' should be 'planning strategies to the structure of the environment'.
  5. [Figures 2 and 3] The captions of Figure 2 and Figure 3 are brief; please state the meaning of the shaded areas, the number of simulation runs, and the specific hyperparameter values for each curve in the captions.

Circularity Check

2 steps flagged · score 6.0 of 10

The 'from scratch' capability demonstration is a tuned reproduction: hyperparameters were optimized for the optimal click sequence and the feature set was extended to encode the target strategy, so the claimed discovery mechanism is not established independently of its inputs.

  1. self definitional [Modeling strategy discovery — Strategy representation]
    "To represent the unique planning strategy of our planning task, we extended the feature set outlined by Jain et al. (2022), who developed these features based on the Mouselab MDP paradigm, by adding seven new features that are essential to capture preferences and avoidance related to specific node levels, as well as one additional stopping criterion."

    The model's strategy space is deliberately extended with features defined to capture node-level preferences, which is exactly the dimension along which the to-be-discovered resource-rational strategy operates (examine immediate nodes, then one outer node). Representing the target strategy is therefore guaranteed by construction, so the later claim that the model can 'represent' the adaptive strategy is a restatement of this design choice rather than an empirical finding.

  2. fitted input called prediction [Modeling strategy discovery — Capability to represent and to discover the novel planning strategy]
    "To evaluate our model’s capacity to represent and discover the adaptive strategy, we conducted model simulations using hyperparameters that were found by optimizing for the optimal click sequence. Figure 3 shows that both the hybrid and model-free Reinforce models are capable of discovering the adaptive strategy from scratch"

    The demonstration that Reinforce 'discovers' the adaptive strategy depends on hyperparameters that were selected by optimizing for the very click sequence that defines the target strategy, and the initial feature weights are also free parameters. The simulation is therefore a fit to the target, not an out-of-sample prediction; showing convergence under this tuning is evidence of expressiveness and favorable tuning, not evidence that a new strategy emerges from unguided metacognitive reinforcement learning.

full rationale

The paper contains substantial independent content: the experiment measures genuine strategy discovery (proportion of adaptive strategies rises from 0.79% to 28.57% with significant trends), the RR strategy is confirmed by an external dynamic-programming computation, and the BMS comparison against RSSL and non-learning models is a self-contained model-selection exercise. These parts do not reduce to their inputs. However, the load-bearing support for the central mechanistic claim — that MCRL models 'are capable of discovering the adaptive strategy from scratch' — is partially circular. The feature set was explicitly extended with seven task-specific node-level preference features to make the target strategy representable, and the capability simulations used hyperparameters optimized for the optimal click sequence. Both choices encode the target before learning, so the 'discovery from scratch' result is by construction a reproduction of the target rather than an emergent prediction. The slower fitted discovery rate reported by the authors is an honest limitation but does not repair this: within an engineered feature space, a slower discovery rate may reflect the shape of the engineered space rather than the learning mechanism. Because the model-comparison results retain independent empirical content, the overall circularity is partial, not total: score 6.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new physical or mental entities; the MCRL model and feature set are computational constructs built from prior framework. The seven new features are extensions of existing feature dimensions, not independent entities.

free parameters (5)
  • Learning rate alpha (Reinforce weight update) = per participant, via Bayesian optimization
    Controls the step size of the weight update in Eq. 1; fitted to each participant's click sequence.
  • Discount factor gamma = per participant
    Discounts past reward contributions in the Reinforce update; fitted per participant.
  • Inverse temperature tau = per participant
    Softmax temperature in Eq. 3 controlling exploration; fitted.
  • Initial feature weights (63) = per participant
    Starting point of the strategy weight vector; fitted. This is a large number of free parameters.
  • Simulation hyperparameters for capability demo = optimized for optimal click sequence
    The demonstration that the model can discover the adaptive strategy uses hyperparameters 'found by optimizing for the optimal click sequence' (Capability section), so the model was set up to produce the target.
assumptions (5)
  • domain assumption Planning can be modeled as a meta-level MDP where mental operations are actions.
    Basis of MCRL; from Hay et al. (2014) and Callaway et al. (2018). Used throughout the modeling section.
  • ad hoc to paper The 63-feature representation is sufficient to represent the optimal strategy and plausible intermediate strategies.
    The paper adds seven features to make the RR strategy representable; if this assumption fails, 'discovery' is only within the feature envelope. Stated in 'Strategy representation' section and acknowledged in Discussion.
  • standard math Softmax action selection with temperature tau (Eq. 3).
    Standard choice rule mapping Q-values to probabilities.
  • domain assumption The resource-rational strategy computed by dynamic programming is indeed optimal for the environment.
    Footnote 1: confirmed by computing optimal metalevel policy with DP; assumes the Mouselab-MDP cost model is the right normative model.
  • domain assumption Click cost externalizes cognitive cost, making click patterns diagnostic of planning strategy.
    Core assumption of Mouselab-MDP paradigm (Callaway et al., 2017).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Experience-driven discovery of planning strategies." pith.science (2026). https://pith.science/paper/VKZV46EK

@misc{pith2026241203111,
  author       = {Pith},
  title        = {Pith review of: Experience-driven discovery of planning strategies},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VKZV46EK}},
  note         = {Machine review of arXiv:2412.03111}
}
read the original abstract

One explanation for how people can plan efficiently despite limited cognitive resources is that we possess a set of adaptive planning strategies and know when and how to use them. But how are these strategies acquired? While previous research has studied how individuals learn to choose among existing strategies, little is known about the process of forming new planning strategies. In this work, we propose that new planning strategies are discovered through metacognitive reinforcement learning. To test this, we designed a novel experiment to investigate the discovery of new planning strategies. We then present metacognitive reinforcement learning models and demonstrate their capability for strategy discovery as well as show that they provide a better explanation of human strategy discovery than alternative learning mechanisms. However, when fitted to human data, these models exhibit a slower discovery rate than humans, leaving room for improvement.

Figures

Figures reproduced from arXiv: 2412.03111 by the authors.

Figure 1
Figure 1. Each trial is randomly sampled from one of those six configurations. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The average score and the proportion of the adap [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Simulations of the hybrid and model-free Reinforce [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Model simulations using fitted hyperparameters of [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 4
Figure 4. Figure 4: Average scores of the participants grouped by the [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

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. Individual differences in the cognitive mechanisms of planning strategy discovery

    cs.AI 2025-05 conditional novelty 4.0 of 10

    Adding pseudo-rewards, subjective effort valuation, and termination deliberation to metacognitive reinforcement learning models captures individual differences in planning strategy discovery but does not close the mod...

Reference graph

Works this paper leans on

25 extracted references · 22 canonical work pages · cited by 1 Pith paper

  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]

    , Yamins, D

    bergstra2013making APACrefauthors Bergstra, J. , Yamins, D. \ Cox, D. APACrefauthors \ 2013 . Making a science of model search: Hyperparameter optimization in hundreds of dimensions for vision architectures Making a science of model search: Hyperparameter optimization in hundreds of dimensions for vision architectures . International conference on machine...

    work page 2013

  3. [3]

    , Sokol-Hessner, P

    Boureau2015 APACrefauthors Boureau, Y L. , Sokol-Hessner, P. \ Daw, N D. APACrefauthors \ 2015 . Deciding how to decide: Self-control and meta-decision making Deciding how to decide: Self-control and meta-decision making . Trends in cognitive sciences 19 11 700--710

    work page 2015

  4. [4]

    , Gul, S

    callaway2018learning APACrefauthors Callaway, F. , Gul, S. , Krueger, P. , Griffiths, T L. \ Lieder, F. APACrefauthors \ 2018 . Learning to select computations Learning to select computations . Uncertainty in Artificial Intelligence: Proceedings of the Thirty-Fourth Conference. Uncertainty in artificial intelligence: Proceedings of the thirty-fourth conference

    work page 2018

  5. [5]

    , Lieder, F

    Callaway2017 APACrefauthors Callaway, F. , Lieder, F. , Krueger, P M. \ Griffiths, T L. APACrefauthors \ 2017 . Mouselab-MDP: A new paradigm for tracing how people plan Mouselab-mdp: A new paradigm for tracing how people plan . The 3rd multidisciplinary conference on reinforcement learning and decision making. The 3rd multidisciplinary conference on reinf...

    work page 2017

  6. [6]

    , van Opheusden, B

    callaway2022rational APACrefauthors Callaway, F. , van Opheusden, B. , Gul, S. , Das, P. , Krueger, P M. , Lieder, F. \ Griffiths, T L. APACrefauthors \ 2022 . Rational use of cognitive resources in human planning Rational use of cognitive resources in human planning . Nature Human Behaviour 1--14

    work page 2022

  7. [7]

    APACrefauthors \ 2018

    Daw2018 APACrefauthors Daw, N D. APACrefauthors \ 2018 . Are we of two minds? Are we of two minds? Nature Neuroscience 21 11 1497

    work page 2018

  8. [8]

    , Russell, S

    hay2014selecting APACrefauthors Hay, N. , Russell, S. , Tolpin, D. \ Shimony, S E. APACrefauthors \ 2014 . Selecting computations: Theory and applications Selecting computations: Theory and applications . arXiv preprint arXiv:1408.2048

    arXiv 2014

Show all 25 references
  1. [9]

    , Jain, Y R

    HeJainLieder2021NIPS-Planning APACrefauthors He, R. , Jain, Y R. \ Lieder, F. APACrefauthors \ 2021 1 12 . Have I done enough planning or should I plan more? Have i done enough planning or should i plan more? Workshop on Metacognition in the Age of AI. Thirty-fifth Conference ...

    2021

  2. [10]

    , Jain, Y R

    HeJainLieder2021 APACrefauthors He, R. , Jain, Y R. \ Lieder, F. APACrefauthors \ 2021 2 . Measuring and modelling how people learn how to plan and how people adapt their planning strategies the to structure of the environment Measuring and modelling how people learn how to pl...

    2021

  3. [11]

    , Callaway, F

    Jain2021Computational APACrefauthors Jain, Y R. , Callaway, F. , Griffiths, T L. , Dayan, P. , He, R. , Krueger, P M. \ Lieder, F. APACrefauthors \ 2022 . A Computational Process-Tracing Method for Measuring People's Planning Strategies and How They Change Over Time A computat...

    2022 doi

  4. [12]

    , Callaway, F

    jain2019measuring APACrefauthors Jain, Y R. , Callaway, F. \ Lieder, F. APACrefauthors \ 2019 . Measuring how people learn how to plan. Measuring how people learn how to plan. CogSci Cogsci \ ( \ 1956--1962)

    2019

  5. [13]

    , Smittenaar, P

    keramati2016adaptive APACrefauthors Keramati, M. , Smittenaar, P. , Dolan, R J. \ Dayan, P. APACrefauthors \ 2016 . Adaptive integration of habits into depth-limited planning defines a habitual-goal--directed spectrum Adaptive integration of habits into depth-limited planning ...

    2016

  6. [14]

    , Lieder, F

    krueger2017enhancing APACrefauthors Krueger, P M. , Lieder, F. \ Griffiths, T. APACrefauthors \ 2017 . Enhancing metacognitive reinforcement learning using reward structures and feedback. Enhancing metacognitive reinforcement learning using reward structures and feedback. CogS...

    2017

  7. [15]

    \ Griffiths, T L

    LiederGriffiths2017 APACrefauthors Lieder, F. \ Griffiths, T L. APACrefauthors \ 2017 . Strategy selection as rational metareasoning Strategy selection as rational metareasoning . Psychological Review 124 6 762--794

    2017

  8. [16]

    \ Griffiths, T L

    lieder2020resource APACrefauthors Lieder, F. \ Griffiths, T L. APACrefauthors \ 2020 . Resource-rational analysis: Understanding human cognition as the optimal use of limited computational resources Resource-rational analysis: Understanding human cognition as the optimal use o...

    2020

  9. [17]

    , Shenhav, A

    miller2019habits APACrefauthors Miller, K J. , Shenhav, A. \ Ludvig, E A. APACrefauthors \ 2019 . Habits without values. Habits without values. Psychological review 126 2 292

    2019

  10. [18]

    APACrefauthors \ 2009

    niv2009reinforcement APACrefauthors Niv, Y. APACrefauthors \ 2009 . Reinforcement learning in the brain Reinforcement learning in the brain . Journal of Mathematical Psychology 53 3 139--154

    2009

  11. [19]

    APACrefauthors \ 1976

    payne1976task APACrefauthors Payne, J W. APACrefauthors \ 1976 . Task complexity and contingent processing in decision making: An information search and protocol analysis Task complexity and contingent processing in decision making: An information search and protocol analysis ...

    1976

  12. [20]

    \ Otto, P E

    rieskamp2006ssl APACrefauthors Rieskamp, J. \ Otto, P E. APACrefauthors \ 2006 . SSL: a theory of how people learn to select strategies. Ssl: a theory of how people learn to select strategies. Journal of Experimental Psychology: General 135 2 207

    2006

  13. [21]

    , Stephan, K E

    rigoux2014bayesian APACrefauthors Rigoux, L. , Stephan, K E. , Friston, K J. \ Daunizeau, J. APACrefauthors \ 2014 . Bayesian model selection for group studies—revisited Bayesian model selection for group studies—revisited . Neuroimage 84 971--985

    2014

  14. [22]

    \ Wefald, E

    Russell1991 APACrefauthors Russell, S. \ Wefald, E. APACrefauthors \ 1991 . Principles of metareasoning Principles of metareasoning . Artificial intelligence 49 1-3 361--395

    1991

  15. [23]

    APACrefauthors \ 1978

    schwarz1978estimating APACrefauthors Schwarz, G. APACrefauthors \ 1978 . Estimating the dimension of a model Estimating the dimension of a model . The annals of statistics 461--464

    1978

  16. [24]

    \ Crowley, K

    siegler1991microgenetic APACrefauthors Siegler, R S. \ Crowley, K. APACrefauthors \ 1991 . The microgenetic method: A direct means for studying cognitive development. The microgenetic method: A direct means for studying cognitive development. American psychologist 46 6 606

    1991

  17. [25]

    APACrefauthors \ 1992

    Williams1992 APACrefauthors Williams, R J. APACrefauthors \ 1992 . Simple statistical gradient-following algorithms for connectionist reinforcement learning Simple statistical gradient-following algorithms for connectionist reinforcement learning . Machine Learning 8 3 229--256

    1992

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.