Pith. sign in

REVIEW 4 major objections 4 minor 75 references

Standard Neural Computation Alone Is Insufficient for Logical Intelligence

T0 review · 4 major / 4 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read The paper argues that ordinary inner-product neural computation cannot by itself deliver reliable deductive inference, and that neural architectures must embed differentiable logical operations such as AND, OR, and NOT to achieve logical…

desk verdict Honest position paper whose central insufficiency thesis is undercut by the fact that thresholded linear units already compute Boolean functions exactly; the toy experiment can't carry it. read the letter →

arxiv 2502.02135 v1 pith:QVHQYNJW submitted 2025-02-04 cs.AI cs.LG

Youngsung Kim This is my paper · ORCID
classification cs.AIcs.LG
keywords logicalintelligenceneuralnetworksdeductivereasoningUnitsneuro-symbolicAIuniversalapproximationtheoremfuzzylogicdifferentiable
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 position paper claims that standard neural computation—inner-product layers followed by nonlinear activations—cannot, by itself, produce robust deductive reasoning or logical consistency. Scaling up models or adding prompt-based heuristics does not close the gap, because approximation is not the same as the exactness and determinism that logical rules require. The proposed fix is to embed Logical Neural Units (LNUs), trainable modules that implement differentiable approximations of AND, OR, and NOT directly inside neural layers. A small toy experiment reports that an LNU-based model generalizes better than inner-product perceptrons on a logical function with few training examples. The paper should be read as a call to change architecture, not merely to add data or parameters.

What carries the argument

The central object is the Logical Neural Unit (LNU), a module that replaces a standard dense layer's inner product and nonlinearity with differentiable logical connectives. Each input feature $x_i$ is weighted to form $z_i = x_i w_i$; a sharpness parameter $\beta$ controls how closely softmax and softmin approximate max and min, and the layer outputs soft-OR($z$) $= \sum_i \text{softmax}(\beta z)_i z_i$ and soft-AND($z$) $= \sum_i \text{softmin}(\beta z)_i z_i$, optionally with soft-NOT($x$) $= 1-x$. Because $\beta$ can be raised, the same unit interpolates from graded fuzzy logic to near-Boolean gates; because the weights and gating are learned, the unit decides which features matter. Stacked LNU layers with logical residual connections of the form soft-IMPLY($A,B$) $=$ soft-OR($1-A, B$) are proposed as a deep architecture that keeps decision boundaries interpretable.

What would settle it

Find a standard inner-product network—no logic modules, no post-hoc thresholding, no external prover—that, after ordinary gradient training, outputs exactly the correct {0,1} truth value on every instance of a nontrivial deductive problem, such as propositional entailment with arbitrarily many variables or first-order theorem proving, with outputs strictly equal to the logical values rather than merely close. One such counterexample would refute the paper's insufficiency claim.

Watch

Extended reading notes

Core claim

The central claim is stated outright in Section 7.1: standard inner-product-based neural networks are good at sub-symbolic pattern recognition but inherently lack the capacity for structured logical reasoning. The paper grounds this in the universal approximation theorem, which guarantees approximation of continuous functions on compact sets but does not guarantee exact discrete outputs, deterministic convergence, global consistency over unbounded domains, or interpretable inference chains. Because logical rules demand such properties, the insufficiency is treated as architectural rather than a matter of scale or training data. The remedy is LNU layers that replace dense inner-product computations with learnable t-norm and t-conorm operators, so that logical inference happens inside the network's own differentiable computation.

Load-bearing premise

The thesis rests on the premise that logical reasoning requires exact, discrete, symbolic outputs, so that approximate or graded truth values are not enough; if that premise fails, standard neural networks can no longer be ruled out.

Editorial extensions

If this is right

  • If the thesis holds, no amount of scaling or prompting in standard transformer or MLP models can guarantee deductive correctness; auxiliary symbolic or logic-embedded machinery will always be needed for strict inference.
  • Large-scale models should not rely on brute-force scaling as the sole path to reasoning; selectively replacing dense layers with LNUs could yield logical consistency at lower parameter and data cost.
  • LNU-based decisions are transparent by construction, because AND/OR semantics define the decision boundary, so individual unit outputs can be read as logical statements without post-hoc explanation tools.
  • Neurosymbolic systems that treat logic as an external module or a hard-coded rule set are a stopgap; logic must live inside the differentiable computation for scalable, stable reasoning.
  • The toy experiment indicates that in low-data logical tasks, LNU modules can generalize from fewer examples than inner-product perceptrons with comparable parameter counts.

Reading between the lines

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

  • An implication the paper leaves implicit is that LNUs are essentially trainable fuzzy-logic operators, so the line between 'embedded logic' and 'learned approximation of logic' is not sharp; architectural guarantees would need to be tested on worst-case behavior rather than average accuracy.
  • The paper's own survey admits fuzzy, many-valued, and probabilistic logic as legitimate continuous-valued logics, which suggests a weaker version of the thesis—standard networks suffice for graded inference, while exact deduction needs logic-embedded units—is more defensible than the blanket insufficiency claim.
  • A testable extension is to benchmark LNU-augmented transformers against standard transformers on propositional entailment instances whose truth values are not near 0 or 1; the sharpness parameter $\beta$ should make LNU performance degrade gracefully, whereas inner-product layers should fail exactly when graded partial truths matter.
  • If the argument is right, chain-of-thought prompting should be understood as a heuristic that surfaces patterns rather than a mechanism of reasoning, which would shift interpretability audits toward counting hidden logical steps instead of tokens.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. This position paper argues that standard inner-product-based neural networks, even with nonlinear activations, are insufficient for logical intelligence, and it proposes Logical Neural Units (LNUs) as differentiable modules that embed approximate AND/OR/NOT operations into deep architectures. The paper reviews universal approximation theorems and several neurosymbolic frameworks, presents a toy experiment comparing an LNU-based model against perceptrons on the logical function f=(x1∨x2)∧¬x3, and closes with a discussion of limitations and future work.

Significance. If the central impossibility thesis were correct, the paper would have substantial implications for neural architecture design, motivating a fundamental shift away from standard dense and attention layers. The paper has some strengths as a position statement: it surveys a broad literature, proposes a concrete modular design, and acknowledges unresolved issues such as soundness, completeness, and first-order extensions. However, the paper does not prove the claimed insufficiency, and the main theoretical argument is contradicted by classical results on threshold circuits. The toy experiment is too narrow and too circular to support the 'demonstrated' claim in Section 7.1. As a proof of impossibility the paper fails; its plausible residue is a modest proposal for logic-biased inductive priors, which is a different and weaker claim.

major comments (4)
  1. [Sections 4.1-4.2, Eq. (1)] The universal-approximation argument does not establish insufficiency. UAT concerns approximation of continuous functions on compact sets, but Boolean logic does not require approximation: a single thresholded linear unit computes AND and OR exactly with appropriate weights and bias, and two-layer threshold networks compute every Boolean function. Since the paper's own definition of standard neural computation includes inner-product transformations and nonlinear activations, thresholded units are inside the class the paper declares insufficient. The baseline in Figure 1 is therefore a selection artifact: ReLU with weights 0.5 and biases 0.0 or -0.5 is not a representative member of the class.
  2. [Section 4.2 and Section 7.2] The premise that logical reasoning requires exact, discrete, and symbolic outputs is asserted without proof. This premise is in tension with Section 3, which presents fuzzy, many-valued, and probabilistic logic as legitimate continuous-valued foundations for reasoning; if graded truth is acceptable, approximate continuous computation is not automatically disqualified. Moreover, Section 7.2 concedes that soundness and completeness guarantees are unsolved for LNUs, so the exactness requirement is not met by the proposed alternative either.
  3. [Section 6.5 and Figure 2] The toy experiment is self-confirming by construction: Equations (8)-(9) define LNUs as soft-AND and soft-OR aggregators, and the target function f=(x1∨x2)∧¬x3 is literally a composition of AND, OR, and NOT. A single perceptron as baseline is a known weak model for such functions, and the reported test accuracies (about 80% for perceptrons versus 84.7% for Logicron) do not demonstrate that standard networks 'inherently lack the capacity' for the task. The wording 'We have demonstrated' in Section 7.1 is therefore not supported by the evidence presented.
  4. [Abstract, Section 7.1] The central modal claim—that no amount of standard inner-product computation can deliver logical intelligence—is never given a precise formal statement. The paper supplies no theorem, no complexity lower bound, and no definition of the relevant architecture class that would exclude threshold or hardtanh activations. Without such a statement the negative thesis cannot be evaluated, and the Section 7.3 alternative-view paragraph concedes that sufficiently large MLPs can approximate logical functions, which further weakens the claim as stated.
minor comments (4)
  1. [Section 6.4 and Figure 1] The notation is inconsistent: the paper introduces LNUs but repeatedly uses 'NLUs' in Section 6.4 and Figure 1 (e.g., 'NLUs can filter them out' and 'NLUs exhibit decision boundaries'). Please standardize the terminology.
  2. [Appendix B] The matrix formulation defines z_ijk = x_ij · w_jk, but Equations (8)-(9) take a vector z as input; the relationship between the broadcasted elementwise product and the softmin/softmax operations in Equations (8)-(9) is not specified.
  3. [Section 6.5] The experimental setup omits optimization details such as learning rate, batch size, number of random seeds, and parameter initialization, which makes the reported accuracy gaps difficult to interpret.
  4. [Section 6.5] The term 'Logicron' is used without a definition; the reader must infer that it denotes the LNU-based model.

Circularity Check

2 steps flagged · score 6.0 of 10

The empirical demonstrations of LNU superiority are self-confirming: LNUs are defined as soft AND/OR operators and then tested on an AND/OR formula, while the inner-product baseline is pinned to parameters that preclude logical behavior; the theoretical insufficiency claim rests on a stipulated exactness requirement rather than on a proof that standard networks cannot implement Boolean logic.

  1. self definitional [Section 6.3 (Eqs. 8-9) and Section 6.5 (Toy Example)]
    "soft-OR(z) = \sum_i (softmax(βz))_i · z_i, (8) soft-AND(z) = \sum_i (softmin(βz))_i · z_i. (9) ... Task: Given the logical function: f (x1, x2, x3) = (x1 ∨ x2) ∧ ¬(x3) ... Logicron uses an LNU layer and Perceptrons use an inner-product-based linear unit with non-linear activation functions."

    The LNU layer is constructed out of soft-AND and soft-OR aggregators, so its native computational primitives are exactly the connectives appearing in the test formula f = (x1 ∨ x2) ∧ ¬x3. The reported generalization advantage of Logicron over perceptrons is therefore a measure of how much logical inductive bias was hard-wired into the architecture, not evidence that standard inner-product networks inherently lack logical capacity. The experiment's outcome is entailed by the definition of the LNU: a model whose units are logic gates is evaluated on a logic-gate composition, so the 'prediction' reduces to the construction of the model.

  2. other [Section 6.4, Figure 1]
    "The third row presents an inner-product unit with ReLU activation, using the same fixed weights (wi = 0.5, ∀i) and different biases (0.0 or −0.5) over the input domain (x1, x2) ∈ [0, 1]. ... In contrast, the inner-product unit, which relies on summation-based arithmetic, does not exhibit logical function behavior."

    The displayed failure of the inner-product unit is manufactured by the chosen biases. On binarized inputs, ReLU(0.5x1 + 0.5x2) with a decision threshold at 0.5 computes AND exactly, and with a threshold at 0 it computes OR exactly; this is the McCulloch-Pitts threshold unit that the paper itself cites as standard neural computation. Because the baseline is not thresholded and its biases are fixed to values that avoid the AND/OR thresholds, the observation that it 'does not exhibit logical function behavior' is a property of the selected parameters, not of inner-product-based layers. The conclusion is thus forced by the experimental setup rather than derived from the nature of inner-product computation.

full rationale

No load-bearing self-citation is present: the LNU proposal is not justified by a prior uniqueness theorem or by the author's own cited results, so the self-citation patterns do not apply. The circularity lies in the two empirical demonstrations offered for the insufficiency thesis. Section 6.3 defines LNUs as differentiable AND/OR aggregators, and Section 6.5 then benchmarks them on an AND/OR formula; the control is a single inner-product unit, and Figure 1 fixes baseline biases so that the ReLU output cannot match Boolean AND/OR at the displayed settings. These observations are consequences of the chosen definitions and parameters, not independent evidence that standard inner-product computation is insufficient. The theoretical UAT argument in Sections 4.1-4.2 is better classified as a correctness risk than a circular step: it stipulates that logical reasoning requires exact, discrete outputs without proving that thresholded or step-activation networks are excluded, and it overlooks that continuous extensions plus thresholds implement Boolean functions. Section 7.2 also concedes that LNUs themselves lack soundness and completeness guarantees, which weakens the exactness-based argument further. Because the central position is asserted rather than derived and the only demonstrations reduce to construction, the circularity score is 6.

Assumptions & free parameters 5 free parameters · 5 assumptions · 4 invented entities

The insufficiency thesis rests on contested premises about what logical intelligence requires (exactness and determinism) rather than on a derived theorem, and the only empirical support is a toy task whose central constant (beta) is unreported. The proposed units, LNU, Logicron, soft-IMPLY residual, and the Locally Gated Logical Consistency mechanism, are named entities introduced here with no evidence beyond the paper's own comparisons; the paper itself lists first-order logic support, soundness guarantees, and real-world validation as open problems (Sec 7.2, 7.4).

free parameters (5)
  • beta (sharpness of soft-AND/soft-OR) = unreported for the toy experiment; 1, 10, 100 in figures
    beta controls how closely Eqs. 8-9 approximate Boolean functions (Sec 6.3). The paper never states which beta produced the reported test accuracies, so the central empirical comparison is underdetermined.
  • Inner-product baseline biases = 0.0 and -0.5 (Figure 1)
    Hand-selected for the dense comparator; the conclusion that inner-product units cannot do logic changes if biases (say -0.75 with a threshold at 0) are chosen that realize AND exactly.
  • Fixed comparison weights = wi = 0.5 for all i
    Set by hand in Figures 1 and 3 to 'ensure an even distribution that aligns with hard AND/OR logic' (Sec 6.4), a choice that predetermines the LNU boundary shape and the comparison outcome.
  • sqrt(d) output normalization = 1/sqrt(d)
    Introduced in Sec 6.3 to 'mitigate activation scaling issues'; no experiment or analysis validates the constant.
  • Toy-run training constants = unreported (epochs = 30 only)
    Learning rate, optimizer, seeds, and architecture dimensions are unstated; parameter counts 90-110 are inconsistent with a single hidden unit on 3 inputs, so the run cannot be reconstructed.
assumptions (5)
  • standard math Universal Approximation Theorem for feedforward networks on compact domains (Hornik et al. 1989, Eq. 1)
    Invoked in Sec 4.1 as the backdrop; standard background mathematics. The insufficiency conclusions drawn from it are not part of the theorem.
  • domain assumption Logical reasoning requires exact, discrete, and symbolic representations and deterministic convergence (Sec 4.2)
    This premise is load-bearing for the insufficiency thesis. It is asserted, not proved, and sits in tension with the paper's own Sec 3, which presents fuzzy and many-valued logic as legitimate continuous-valued logical frameworks.
  • domain assumption Compact-domain approximation cannot ensure global consistency over unbounded first-order domains (Sec 4.2)
    Technically true for UAT, but the paper jumps from this to 'standard networks cannot do systematic reasoning' without a derivation.
  • standard math softmax(beta z) and softmin(beta z) approximate max and min as beta goes to infinity (Sec 6.3)
    Standard and correct as a limit, but the paper states neither the uniformity of the approximation nor the finite beta values needed for Boolean behavior.
  • domain assumption Scaling laws do not inherently enable structured rule-based reasoning (Sec 7.1)
    Asserted without evidence; Sec 7.3 concedes large MLPs 'can approximate logical functions' given sufficient data and compute, which partially undermines this premise.
invented entities (4)
  • Logical Neural Unit (LNU)
    purpose: Differentiable module embedding soft-AND, soft-OR, and optional soft-NOT inside neural layers (Eqs. 8-11, Sec 6.2-6.3).
    No external benchmark or falsifiable prediction is provided; evidence is limited to the authors' toy task (Sec 6.5) and fixed-weight visualizations (Figure 1). Sec 7.4 concedes real-world benchmarking is still needed.
  • Logicron
    purpose: LNU-based model benchmarked against perceptrons on the toy task (Sec 6.5, Figure 2).
    Named only in this paper; the reference list contains no prior 'Logicron' work, and its reported advantage rests on a single under-specified 20-sample run.
  • soft-IMPLY residual connection
    purpose: Logical residual across stacked LNU layers via soft-OR(1 - A, B) (Eq. 11).
    Proposed in Sec 6.3 with no experiment isolating its effect; behavior in deep stacks is untested.
  • Locally Gated Logical Consistency mechanism
    purpose: Broadcast elementwise gating of inputs by learned matrices WAND and WOR before logical aggregation (Appendix B).
    A new name for per-dimension gating similar to attention; the paper analogizes it to self-attention (Sec 6.4) but gives no analysis of whether gating preserves logical semantics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Standard Neural Computation Alone Is Insufficient for Logical Intelligence." pith.science (2026). https://pith.science/paper/QVHQYNJW

@misc{pith2026250202135,
  author       = {Pith},
  title        = {Pith review of: Standard Neural Computation Alone Is Insufficient for Logical Intelligence},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QVHQYNJW}},
  note         = {Machine review of arXiv:2502.02135}
}
read the original abstract

Neural networks, as currently designed, fall short of achieving true logical intelligence. Modern AI models rely on standard neural computation-inner-product-based transformations and nonlinear activations-to approximate patterns from data. While effective for inductive learning, this architecture lacks the structural guarantees necessary for deductive inference and logical consistency. As a result, deep networks struggle with rule-based reasoning, structured generalization, and interpretability without extensive post-hoc modifications. This position paper argues that standard neural layers must be fundamentally rethought to integrate logical reasoning. We advocate for Logical Neural Units (LNUs)-modular components that embed differentiable approximations of logical operations (e.g., AND, OR, NOT) directly within neural architectures. We critique existing neurosymbolic approaches, highlight the limitations of standard neural computation for logical inference, and present LNUs as a necessary paradigm shift in AI. Finally, we outline a roadmap for implementation, discussing theoretical foundations, architectural integration, and key challenges for future research.

Figures

Figures reproduced from arXiv: 2502.02135 by the authors.

Figure 1
Figure 1. Comparison of logical function approximations across different models. The first row represents Hard Logic, where AND and OR operations are applied to binarized inputs (xi > 0.5, ∀i). The second row shows LNUs with fixed weights (wi = 0.5, ∀i) and a sharp gating parameter β = 100. The third row presents an inner-product unit with ReLU activation, using the same fixed weights (wi = 0.5, ∀i) and different biases (0.0 … view at source ↗
Figure 2
Figure 2. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Effect of gating parameter β ∈ {1, 10, 100} on LNU decision boundaries with fixed weights (wi = 0.5, ∀i). Increasing β sharpens the boundary, approaching hard logic. A. Observations on Decision Boundary Expanding on [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

75 extracted references · 60 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    Organon, 4th century BCE

    Aristotle. Organon, 4th century BCE

    work page

  3. [3]

    and Hitzler, P

    Bader, S. and Hitzler, P. Dimensions of neural-symbolic integration a structured survey. In Artemov, S., Barringer, H., Garcez, A. S. d., Lamb, L. C., and Woods, J. (eds.), We Will Show Them: Essays in Honour of Dov Gabbay, volume 1, pp.\ 167--194. King's College Publications, 2005

    work page 2005

  4. [4]

    Explaining neural scaling laws

    Bahri, Y., Dyer, E., Kaplan, J., Lee, J., and Sharma, U. Explaining neural scaling laws. Proceedings of the National Academy of Sciences, 121 0 (27): 0 e2311878121, 2024. doi:10.1073/pnas.2311878121. URL https://www.pnas.org/doi/abs/10.1073/pnas.2311878121

    doi:10.1073/pnas.2311878121 2024

  5. [5]

    Barsalou, L. W. Perceptual symbol systems. Behavioral and Brain Sciences, 22 0 (4): 0 577--660, 1999

    work page 1999

  6. [6]

    R., d'Avila Garcez, A., Bader, S., Bowman, H., Domingos, P., Hitzler, P., Kühnberger, K.-U., Lamb, L

    Besold, T. R., d'Avila Garcez, A., Bader, S., Bowman, H., Domingos, P., Hitzler, P., Kühnberger, K.-U., Lamb, L. C., Lowd, D., Lima, P. M. V., de Penning, L., Pinkas, G., Poon, H., and Zaverucha, G. Neural-symbolic learning and reasoning: A survey and interpretation. arXiv preprint arXiv:1711.03902, 2017. URL https://arxiv.org/abs/1711.03902

    arXiv 2017

  7. [7]

    T., Li, Y., Lundberg, S., Nori, H., Palangi, H., Ribeiro, M

    Bubeck, S., Chandrasekaran, V., Eldan, R., Gehrke, J., Horvitz, E., Kamar, E., Lee, P., Lee, Y. T., Li, Y., Lundberg, S., Nori, H., Palangi, H., Ribeiro, M. T., and Zhang, Y. Sparks of artificial general intelligence: Early experiments with gpt-4. arXiv preprint arXiv:2303.12712, 2023

    arXiv 2023

  8. [8]

    Large language models can be easily distracted by irrelevant context

    Cai, T., Liu, H., Liu, Z., Abbeel, P., Song, D., and Mordatch, I. Large language models can be easily distracted by irrelevant context. In International Conference on Machine Learning (ICML), 2023

    work page 2023

Show all 75 references
  1. [9]

    Cattell, R. B. The measurement of adult intelligence. Psychological Bulletin, 40 0 (3): 0 153--193, 1943

    1943

  2. [10]

    Cattell, R. B. Theory of fluid and crystallized intelligence: A critical experiment. Journal of Educational Psychology, 54: 0 1--22, 1963

    1963

  3. [11]

    Algebraic Foundations of Many-Valued Reasoning

    Cignoli, R., Dubuc, E., and Mundici, D. Algebraic Foundations of Many-Valued Reasoning. Springer Science & Business Media, 2013

    2013

  4. [12]

    Selection-inference: Exploiting large language models for interpretable logical reasoning

    Creswell, A., Shanahan, M., and Higgins, I. Selection-inference: Exploiting large language models for interpretable logical reasoning. In International Conference on Learning Representations (ICLR), 2022. URL https://openreview.net/forum?id=3Pf3Wg6o-A4

    2022

  5. [13]

    Approximation by superpositions of a sigmoidal function

    Cybenko, G. Approximation by superpositions of a sigmoidal function. Mathematics of Control, Signals and Systems, 2 0 (4): 0 303--314, 1989

    1989

  6. [14]

    C., Serafini, L., Spranger, M., and Tran, S

    d'Avila Garcez, A., Gori, M., Lamb, L. C., Serafini, L., Spranger, M., and Tran, S. N. Neural-symbolic computing: An effective methodology for principled integration of machine learning and reasoning. arXiv preprint arXiv:1905.06088, 2019. URL https://arxiv.org/abs/1905.06088

    1905 arXiv

  7. [15]

    Dennett, D. C. The nature of images and the introspective trap. In Content and Consciousness, pp.\ 132--146. Routledge, London, 1969

    1969

  8. [16]

    Neural logic machines

    Dong, H., Mao, J., Lin, T., Wang, C., Li, L., and Zhou, D. Neural logic machines. arXiv preprint arXiv:1904.11694, 2019

    1904 arXiv

  9. [17]

    and Prade, H

    Dubois, D. and Prade, H. Fuzzy sets and systems: Theory and applications. Academic Press, 1980

    1980

  10. [18]

    and Lamb, L

    d’Avila Garcez, A. and Lamb, L. C. Neurosymbolic ai: The 3rd wave. Artificial Intelligence Review, 56 0 (11): 0 12387--12406, nov 2023

    2023

  11. [19]

    and Halpern, J

    Fagin, R. and Halpern, J. Y. Uncertainty, belief, and probability. Computational Intelligence, 7 0 (1): 0 160--173, 1990

    1990

  12. [20]

    Fodor, J. A. The Language of Thought. Harvard University Press, Cambridge, MA, 1975

    1975

  13. [21]

    and Taskar, B

    Getoor, L. and Taskar, B. (eds.). Introduction to Statistical Relational Learning. Adaptive Computation and Machine Learning. The MIT Press, 2007. ISBN 978-0-262-07286-1. URL https://doi.org/10.7551/mitpress/7432.001.0001

    2007 doi

  14. [22]

    Metamathematics of Fuzzy Logic

    Hajek, P. Metamathematics of Fuzzy Logic. Springer Science & Business Media, 1998

    1998

  15. [23]

    The symbol grounding problem

    Harnad, S. The symbol grounding problem. Physica D: Nonlinear Phenomena, 42 0 (1-3): 0 335--346, 1990

    1990

  16. [24]

    and Prentzas, J

    Hatzilygeroudis, I. and Prentzas, J. Neurules: Integrating symbolic rules and neurocomputing. In Fotiades, D. and Nikolopoulos, S. (eds.), Advances in Informatics, pp.\ 122--133. World Scientific, Singapore, 2000

    2000

  17. [25]

    Artificial Intelligence: The Very Idea

    Haugeland, J. Artificial Intelligence: The Very Idea. MIT Press, Cambridge, MA, 1985

    1985

  18. [26]

    An overview of strategies for neurosymbolic integration

    Hilario, M. An overview of strategies for neurosymbolic integration. In Proceedings of the Workshop on Connectionist-Symbolic Integration: From Unified to Hybrid Approaches, Montreal, Canada, 1995

    1995

  19. [27]

    Multilayer feedforward networks are universal approximators

    Hornik, K., Stinchcombe, M., and White, H. Multilayer feedforward networks are universal approximators. Neural Networks, 2 0 (5): 0 359--366, 1989

    1989

  20. [28]

    and Kalinke, Y

    Hölldobler, S. and Kalinke, Y. Towards a massively parallel computational model for logic programming. In Proceedings of the ECAI'94 Workshop on Combining Symbolic and Connectionist Processing, pp.\ 68--77, 1994

    1994

  21. [29]

    Thinking, Fast and Slow

    Kahneman, D. Thinking, Fast and Slow. Farrar, Straus and Giroux, 2011

    2011

  22. [30]

    P., and B Murthy, A

    Kambhampati, S., Valmeekam, K., Guan, L., Verma, M., Stechly, K., Bhambri, S., Saldyt, L. P., and B Murthy, A. Position: LLM s can’t plan, but can help planning in LLM -modulo frameworks. In Salakhutdinov, R., Kolter, Z., Heller, K., Weller, A., Oliver, N., Scarlett, J., and B...

    2024

  23. [31]

    B., Chess, B., Child, R., Gray, S., Radford, A., Wu, J., and Amodei, D

    Kaplan, J., McCandlish, S., Henighan, T., Brown, T. B., Chess, B., Child, R., Gray, S., Radford, A., Wu, J., and Amodei, D. Scaling laws for neural language models. arXiv preprint arXiv:2001.08361, 2020

    2001 arXiv

  24. [32]

    The third ai summer: Aaai robert s

    Kautz, H. The third ai summer: Aaai robert s. engelmore memorial lecture. AI Magazine, 43 0 (1): 0 105--125, 2022

    2022

  25. [33]

    Kleene, S. C. Introduction to Metamathematics. North-Holland Publishing Company, 1952

    1952

  26. [34]

    P., Mesiar, R., and Pap, E

    Klement, E. P., Mesiar, R., and Pap, E. Triangular Norms. Springer, 2000

    2000

  27. [35]

    Manhaeve, R., Dumančić, S., Kimmig, A., Demeester, T., and Raedt, L. D. Deepproblog: Neural probabilistic logic programming. In Advances in Neural Information Processing Systems 31 (NeurIPS 2018), pp.\ 3749--3759, 2018

    2018

  28. [36]

    From statistical relational to neurosymbolic artificial intelligence: A survey

    Marra, G., Dumančić, S., Manhaeve, R., and De Raedt , L. From statistical relational to neurosymbolic artificial intelligence: A survey. Artificial Intelligence, 328: 0 104062, 2024

    2024

  29. [37]

    and Hayes, P

    McCarthy, J. and Hayes, P. J. Some philosophical problems from the standpoint of artificial intelligence. Machine Intelligence, 4: 0 463--502, 1969

    1969

  30. [38]

    McClelland, J. L. and Rumelhart, D. E. Parallel Distributed Processing: Explorations in the Microstructure of Cognition, Volume 2: Psychological and Biological Models. MIT Press, Cambridge, MA, 1986

    1986

  31. [39]

    McCulloch, W. S. and Pitts, W. A logical calculus of the ideas immanent in nervous activity. Bulletin of Mathematical Biophysics, 5: 0 115--133, 1943

    1943

  32. [40]

    Statistical metrics

    Menger, K. Statistical metrics. Proceedings of the National Academy of Sciences of the United States of America, 28 0 (12): 0 535--537, 1942

    1942

  33. [41]

    Steps toward artificial intelligence

    Minsky, M. Steps toward artificial intelligence. Proceedings of the IRE, 49 0 (1): 0 8--30, 1961. doi:10.1109/JRPROC.1961.287775. URL https://ieeexplore.ieee.org/document/4066245

    1961

  34. [42]

    Learning from positive data

    Muggleton, S. Learning from positive data. In Proceedings of the 6th International Workshop on Inductive Logic Programming (ILP), pp.\ 358--376, 1996

    1996

  35. [43]

    Limitations of the current stock of ideas about problem solving

    Newell, A. Limitations of the current stock of ideas about problem solving. In Kent, A. and Taulbee, O. (eds.), Proceedings of the Conference on Electronic Information Handling, pp.\ 195--208, New York, 1965. Spartan Books

    1965

  36. [44]

    and Simon, H

    Newell, A. and Simon, H. The logic theory machine--a complex information processing system. IRE Transactions on Information Theory, 2 0 (3): 0 61--79, 1956

    1956

  37. [45]

    and Simon, H

    Newell, A. and Simon, H. A. Human Problem Solving. Prentice-Hall, Englewood Cliffs, NJ, 1972

    1972

  38. [46]

    Nilsson, N. J. Principles of Artificial Intelligence. Tioga Publishing, 1980

    1980

  39. [47]

    Nilsson, N. J. Probabilistic logic. Artificial Intelligence, 28 0 (1): 0 71--87, 1986

    1986

  40. [48]

    Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference

    Pearl, J. Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference. Morgan Kaufmann, 1988

    1988

  41. [49]

    Rae, J. W., Borgeaud, S., Cai, T., Millican, K., Hoffmann, J., Song, F., Aslanides, J., Henderson, S., Ring, R., Young, S., Rutherford, E., Hennigan, T., Menick, J., Cassirer, A., Powell, R., van den Driessche, G., Hendricks, L. A., Rauh, M., Huang, P.-S., Glaese, A., Welbl, J...

    2021 arXiv

  42. [50]

    D., Dumančić, S., Manhaeve, R., and Marra, G

    Raedt, L. D., Dumančić, S., Manhaeve, R., and Marra, G. From statistical relational to neuro-symbolic artificial intelligence. In Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence (IJCAI-20), pp.\ 4943--4950, 2020

    2020

  43. [51]

    Many-Valued Logic

    Rescher, N. Many-Valued Logic. McGraw-Hill, New York, 1969

    1969

  44. [52]

    Riegel, R., Gray, A., and Laird, J. E. Logical neural networks. arXiv preprint arXiv:2006.13155, 2020

    2006 arXiv

  45. [53]

    and Riedel, S

    Rockt\" a schel, T. and Riedel, S. End-to-end differentiable proving. In Guyon, I., Luxburg, U. V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., and Garnett, R. (eds.), Advances in Neural Information Processing Systems, volume 30. Curran Associates, Inc., 2017

    2017

  46. [54]

    Rumelhart, D. E. and McClelland, J. L. Parallel Distributed Processing: Explorations in the Microstructure of Cognition, Volume 1: Foundations. MIT Press, Cambridge, MA, 1986

    1986

  47. [55]

    and Norvig, P

    Russell, S. and Norvig, P. Artificial Intelligence: A Modern Approach. Pearson, 4th edition, 2020

    2020

  48. [56]

    K., Zhou, L., Eberhart, A., and Hitzler, P

    Sarker, M. K., Zhou, L., Eberhart, A., and Hitzler, P. Neuro-symbolic artificial intelligence. AI Commun., 34 0 (3): 0 197–209, January 2021. ISSN 0921-7126

    2021

  49. [57]

    and Sklar, A

    Schweizer, B. and Sklar, A. Associative functions and statistical triangle inequalities. Publications of the Mathematical Society of Japan, 8: 0 169--186, 1961

    1961

  50. [58]

    and Sklar, A

    Schweizer, B. and Sklar, A. Probabilistic Metric Spaces. Elsevier, 1983

    1983

  51. [59]

    Seda, A. K. On the integration of connectionist and logic-based systems. Electronic Notes in Theoretical Computer Science, 161: 0 109--130, 2006. ISSN 1571-0661. doi:https://doi.org/10.1016/j.entcs.2006.04.028. URL https://www.sciencedirect.com/science/article/pii/S15710661060...

    2006 doi

  52. [60]

    Sen, P., de Carvalho, B. W. S. R., Riegel, R., and Gray, A. Neuro-symbolic inductive logic programming with logical neural networks. In Proceedings of the AAAI Conference on Artificial Intelligence, pp.\ 8212--8219, 2022

    2022

  53. [61]

    and Garcez, A

    Serafini, L. and Garcez, A. d. Logic tensor networks: Deep learning and logical reasoning from data and knowledge. In Conference of the Italian Association for Artificial Intelligence, pp.\ 334--348. Springer, 2016

    2016

  54. [62]

    Smith, K. C. Multiple-valued logic: A tutorial and appreciation. Computer, 21 0 (4): 0 17--27, 1988

    1988

  55. [63]

    D eterm LR : Augmenting LLM -based logical reasoning from indeterminacy to determinacy

    Sun, H., Xu, W., Liu, W., Luan, J., Wang, B., Shang, S., Wen, J.-R., and Yan, R. D eterm LR : Augmenting LLM -based logical reasoning from indeterminacy to determinacy. In Ku, L.-W., Martins, A., and Srikumar, V. (eds.), Proceedings of the 62nd Annual Meeting of the Associatio...

    2024 doi

  56. [64]

    The semantic conception of truth

    Tarski, A. The semantic conception of truth. Philosophy and Phenomenological Research, 4: 0 341--376, 1944

    1944

  57. [65]

    Diagnosing the first-order logical reasoning ability through L ogic NLI

    Tian, J., Li, Y., Chen, W., Xiao, L., He, H., and Jin, Y. Diagnosing the first-order logical reasoning ability through L ogic NLI . In Moens, M.-F., Huang, X., Specia, L., and Yih, S. W.-t. (eds.), Proceedings of the 2021 Conference on Empirical Methods in Natural Language Pro...

    2021 doi

  58. [66]

    Analyzing differentiable fuzzy logic operators

    van Krieken , E., Acar, E., and van Harmelen , F. Analyzing differentiable fuzzy logic operators. Artificial Intelligence, 302: 0 103602, 2022. ISSN 0004-3702. doi:https://doi.org/10.1016/j.artint.2021.103602. URL https://www.sciencedirect.com/science/article/pii/S0004370221001533

    2022

  59. [67]

    Satnet: Bridging deep learning and logical reasoning using a differentiable satisfiability solver

    Wang, P.-W., Donti, P., Wilder, B., and Kolter, Z. Satnet: Bridging deep learning and logical reasoning using a differentiable satisfiability solver. In Proceedings of the International Conference on Machine Learning (ICML), pp.\ 6545--6554. PMLR, 2019

    2019

  60. [68]

    Towards data-and knowledge-driven ai: A survey on neuro-symbolic computing

    Wang, W., Yang, Y., and Wu, F. Towards data-and knowledge-driven ai: A survey on neuro-symbolic computing. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47 0 (2): 0 878--899, 2025

    2025

  61. [69]

    Chain of thought prompting elicits reasoning in large language models

    Wei, J., Wang, X., Schuurmans, D., Bosma, M., Chi, E., Le, Q., and Zhou, D. Chain of thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems (NeurIPS), 2022. URL https://arxiv.org/abs/2201.11903

    2022 arXiv

  62. [70]

    Whitehead, A. N. and Russell, B. Principia Mathematica. Cambridge University Press, 1910

    1910

  63. [71]

    Learning algorithms via neural logic networks

    Yang, F., Song, L., and Zhang, M. Learning algorithms via neural logic networks. arXiv preprint arXiv:1904.01554, 2019

    1904 arXiv

  64. [72]

    Neurasp: Embracing neural networks into answer set programming

    Yang, Z., Ishay, A., and Lee, J. Neurasp: Embracing neural networks into answer set programming. In Bessiere, C. (ed.), Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence, IJCAI-20 , pp.\ 1755--1762. International Joint Conferences on Art...

    2020

  65. [73]

    Tree of thoughts: Deliberate problem solving with large language models

    Yao, S., Yu, D., Zhao, J., Shafran, I., Griffiths, T., Cao, Y., and Narasimhan, K. Tree of thoughts: Deliberate problem solving with large language models. In Advances in Neural Information Processing Systems (NeurIPS), 2023

    2023

  66. [74]

    Zadeh, L. A. Fuzzy sets. Information and Control, 8 0 (3): 0 338--353, 1965

    1965

  67. [75]

    O logice trójwartościowej

    Łukasiewicz, J. O logice trójwartościowej. Ruch Filozoficzny, 5: 0 169--171, 1920

    1920

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.