pith. sign in

arxiv: 2606.00102 · v1 · pith:S5OFWRAWnew · submitted 2026-05-26 · 💻 cs.AI · math.PR

On the evolution of the concept of probability as a mirror of the evolution of reason

Jean-Louis Le Mou\"el , Vincent Courtillot , Dominique Gibert , Vladimir Kossobokov , Jean-Baptiste Boul\'e , Pierpaolo Zuddas , Fernando Lopes , Pa\"ikan Marccagi
show 1 more author
Alexis Maineult
This is my paper

Pith reviewed 2026-06-29 17:58 UTC · model grok-4.3

classification 💻 cs.AI math.PR
keywords probability theoryfuzzy logicdeep learningBayesian inferencescientific rationalityuncertaintyvaguenesshistory of probability
0
0 comments X

The pith

Scientific rationality must articulate uncertainty via probability, vagueness via fuzzy logic, and inference explicitly rather than relying on data performance alone.

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

The paper traces probability's development from Pascal and Fermat's combinatorial games through Bayes, Laplace, Poisson, and Kolmogorov's axioms to its mature form as a logic of information in Bayesian methods. This trajectory shows how reason incorporated uncertainty, time, and coherence into judgment. The analysis identifies a boundary: probability handles uncertainty about defined propositions but does not formalize vagueness in the concepts themselves. Fuzzy logic is introduced as the rigorous extension for graded meaning and qualitative judgment, while deep learning is positioned as a separate mode of geometric optimization and prediction. The resulting claim is that contemporary rationality requires the joint, explicit use of uncertainty, vagueness, and inference.

Core claim

Probability theory evolved from a calculus of chance into an axiomatic framework and then into Tarantola's logic of information that coherently merges prior knowledge with data. This evolution mirrors the growth of scientific reason yet stops short when propositions involve vague concepts that probability cannot grade. Fuzzy logic supplies a distinct formal language for such graded meaning, while deep learning supplies powerful interpolation through optimization without explicit inference steps. Placing the three in historical perspective clarifies their complementary roles and shows that rationality cannot be reduced to data-driven performance.

What carries the argument

The historical trajectory of probability from combinatorial symmetry to information logic, limited by its inability to grade concept vagueness and therefore extended by fuzzy logic for graded meaning.

If this is right

  • Bayesian inference supplies a coherent method for updating beliefs with new data while preserving prior knowledge.
  • Fuzzy logic formalizes qualitative and graded judgments that probability leaves outside its scope.
  • Deep learning achieves high predictive performance through geometric optimization rather than through explicit uncertainty or inference rules.
  • Full scientific rationality therefore requires the explicit articulation of uncertainty, vagueness, and inference together.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Hybrid models that embed fuzzy components inside probabilistic frameworks could be tested on tasks where concept boundaries are inherently graded.
  • The same historical lens applied to other formalisms might reveal similar limits when applied to real-world scientific description.
  • If the separation of vagueness from uncertainty holds, then purely data-driven systems will continue to require post-hoc human interpretation for concept-level claims.

Load-bearing premise

Probability theory inherently cannot formalize the vagueness of the concepts used to describe propositions, and fuzzy logic supplies the missing rigorous language for graded meaning.

What would settle it

A demonstration that every case of conceptual vagueness encountered in scientific propositions can be restated as precise, non-vague statements inside standard probability theory without loss of expressive power or predictive accuracy.

read the original abstract

Over the centuries, probability theory has grown from the calculus of games of chance into a central framework for reasoning under uncertainty. This article interprets that evolution not merely as a mathematical history, but as a transformation of rationality itself. From Pascal and Fermat's combinatorial symmetry to the inductive logic of Bayes and Laplace, from Poisson's statistics of events to Kolmogorov's axiomatic formalization, probability progressively incorporated uncertainty, time, and coherence into scientific judgment. This trajectory reaches a mature epistemological form in modern Bayesian inference, especially in Tarantola's view of probability as a logic of information, where prior knowledge and data are combined coherently. Yet this framework also exposes a limit: probability quantifies uncertainty about well-defined propositions, but does not by itself formalize the vagueness of the concepts used to describe them. The article therefore examines how rationality extends beyond probability. Fuzzy logic is presented as a rigorous language for graded meaning and qualitative judgment, while deep learning is analyzed as a distinct, powerful mode of prediction based on geometric interpolation and optimization rather than explicit inference. By situating probability, fuzzy logic, and deep learning in a common historical and epistemological perspective, the article clarifies their roles and limits. It argues that contemporary scientific rationality cannot be reduced to data-driven performance alone, but requires the explicit articulation of uncertainty, vagueness, and inference.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript traces the historical evolution of probability theory from Pascal and Fermat's combinatorial approach through Bayes-Laplace induction, Poisson statistics, Kolmogorov's axioms, and Tarantola's information logic, interpreting this as a mirror of developing scientific rationality. It argues that probability formalizes uncertainty over crisp propositions but cannot address conceptual vagueness, positioning fuzzy logic as the rigorous language for graded meaning and deep learning as a non-inferential geometric optimization mode. The central claim is that contemporary rationality requires explicit articulation of uncertainty, vagueness, and inference beyond data-driven performance alone.

Significance. If the interpretive distinctions hold, the paper offers a useful epistemological synthesis situating probability, fuzzy logic, and deep learning within a shared historical trajectory, clarifying their complementary limits in AI reasoning. It merits credit for the coherent narrative linking Tarantola's view to broader rationality and for highlighting that probability alone does not exhaust vagueness, a point with precedent in philosophy of vagueness but usefully reframed for AI contexts. As a purely interpretive essay without derivations, predictions, or empirical tests, its technical impact on cs.AI remains modest.

major comments (1)
  1. [Abstract] Abstract (paragraph on limits of probability): The assertion that probability 'quantifies uncertainty about well-defined propositions, but does not by itself formalize the vagueness of the concepts used to describe them' is load-bearing for the claim that fuzzy logic supplies a necessary separate framework, yet the text provides no formal argument, counterexample, or engagement with extensions (e.g., imprecise probabilities) showing why such extensions are insufficient; the distinction therefore rests on assertion rather than demonstrated necessity.
minor comments (2)
  1. The historical trajectory would be clearer with explicit section headings or numbered subsections separating the evolution of probability from the discussion of its limits and the roles of fuzzy logic and deep learning.
  2. Primary-source citations (e.g., to specific works by Tarantola or Kolmogorov) are referenced only in passing; adding a short reference list or inline pointers would improve traceability without altering the interpretive character.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive report and the opportunity to respond. We address the single major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract (paragraph on limits of probability): The assertion that probability 'quantifies uncertainty about well-defined propositions, but does not by itself formalize the vagueness of the concepts used to describe them' is load-bearing for the claim that fuzzy logic supplies a necessary separate framework, yet the text provides no formal argument, counterexample, or engagement with extensions (e.g., imprecise probabilities) showing why such extensions are insufficient; the distinction therefore rests on assertion rather than demonstrated necessity.

    Authors: The manuscript is an interpretive historical and epistemological essay, not a formal technical derivation. The distinction is presented as a conceptual observation drawn from the historical trajectory of probability (Pascal-Fermat through Kolmogorov and Tarantola), which consistently presupposes propositions with determinate truth values, versus the philosophical problem of vagueness in concept application (e.g., sorites paradoxes). We do not offer a formal proof of necessity because that lies outside the paper's scope. Imprecise probabilities extend the representation of uncertainty over probability measures but continue to operate on crisp propositions and do not model graded conceptual membership. We will add a short clarifying sentence in the abstract and introduction noting the interpretive basis and briefly contrasting with imprecise-probability approaches. revision: partial

Circularity Check

0 steps flagged

No significant circularity

full rationale

The manuscript is a philosophical-historical essay tracing the development of probability from Pascal-Fermat through Bayes, Laplace, Poisson, Kolmogorov, and Tarantola, then contrasting it with fuzzy logic and deep learning. It contains no equations, derivations, fitted parameters, or quantitative predictions. The central claim is an interpretive distinction between uncertainty over crisp propositions and vagueness of concepts, supported by standard historical references rather than any self-referential construction or self-citation chain. No load-bearing step reduces to its own inputs by definition or by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No mathematical derivations or empirical claims are present; the text relies on standard historical facts and philosophical distinctions without introducing fitted parameters, new axioms, or postulated entities.

pith-pipeline@v0.9.1-grok · 5807 in / 1006 out tokens · 20312 ms · 2026-06-29T17:58:51.371605+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

51 extracted references · 1 canonical work pages

  1. [1]

    Archive for history of exact sciences , volume=

    A history of the axiomatic formulation of probability from Borel to Kolmogorov: Part I , author=. Archive for history of exact sciences , volume=. 1978 , publisher=

    1978

  2. [2]

    By the late Rev

    An essay towards solving a problem in the doctrine of chances. By the late Rev. Mr. Bayes, FRS communicated by Mr. Price, in a letter to John Canton, AMFR S , author=. Philosophical transactions of the Royal Society of London , number=. 1763 , publisher=

  3. [3]

    Ars conjectandi, opus posthumum: accedit tractatus de seriebus infinitis, et epistola Gallice scripta de ludo pil

    Bernoulli, Jakob , year=. Ars conjectandi, opus posthumum: accedit tractatus de seriebus infinitis, et epistola Gallice scripta de ludo pil

  4. [4]

    Les probabilit

    Borel,. Les probabilit. Rendiconti del Circolo Matematico di Palermo (1884-1940) , volume=. 1909 , publisher=

    1940

  5. [5]

    Statistics , volume=

    Bayesian filtering: From Kalman filters to particle filters, and beyond , author=. Statistics , volume=

  6. [6]

    Essai sur l'application de l'analyse

    Condorcet, Nicolas , year=. Essai sur l'application de l'analyse

  7. [7]

    1961 , publisher=

    The algebra of probable inference , author=. 1961 , publisher=

    1961

  8. [8]

    Mathematics of control, signals and systems , volume=

    Approximation by superpositions of a sigmoidal function , author=. Mathematics of control, signals and systems , volume=. 1989 , publisher=

    1989

  9. [9]

    The Invention of Physical Science: Intersections of Mathematics, Theology and Natural Philosophy Since the Seventeenth Century Essays in Honor of Erwin N

    The doctrine of chances without chance: determinism, mathematical probability, and quantification in the seventeenth century , author=. The Invention of Physical Science: Intersections of Mathematics, Theology and Natural Philosophy Since the Seventeenth Century Essays in Honor of Erwin N. Hiebert , pages=. 1992 , publisher=

    1992

  10. [10]

    , author=

    Bayesian statistical inference for psychological research. , author=. Psychological review , volume=. 1963 , publisher=

    1963

  11. [11]

    De Finetti, Bruno , booktitle=. La pr

  12. [12]

    Mathematical Demography: Selected Papers , pages=

    On the probability of the extinction of families , author=. Mathematical Demography: Selected Papers , pages=. 1814 , publisher=

  13. [13]

    Theoria motus corporum coelestium in sectionibus conicis solem ambientium auctore Carolo Friderico Gauss , author=

  14. [14]

    arXiv preprint arXiv:2408.16361 , year=

    Information theory, Signal Analysis and Inverse Problem , author=. arXiv preprint arXiv:2408.16361 , year=

    work page arXiv

  15. [15]

    2005 , publisher=

    A history of probability and statistics and their applications before 1750 , author=. 2005 , publisher=

    2005

  16. [16]

    Neural computation , volume=

    Long short-term memory , author=. Neural computation , volume=. 1997 , publisher=

    1997

  17. [17]

    Neural networks , volume=

    Approximation capabilities of multilayer feedforward networks , author=. Neural networks , volume=. 1991 , publisher=

    1991

  18. [18]

    Definitions and preliminary lemmas , author=

    A theory of steady-state activity in nerve-fiber networks: I. Definitions and preliminary lemmas , author=. The bulletin of mathematical biophysics , volume=. 1941 , publisher=

    1941

  19. [19]

    1657 , pages=

    Exercitationum Mathematicarum Liber V, De Ratiociniis in Ludo Aleae , author=. 1657 , pages=

  20. [20]

    2003 , publisher=

    Probability theory: The logic of science , author=. 2003 , publisher=

    2003

  21. [21]

    1939 , publisher=

    Theory of probability , author=. 1939 , publisher=

    1939

  22. [22]

    A new approach to linear filtering and prediction problems , author=

  23. [23]

    science , volume=

    Optimization by simulated annealing , author=. science , volume=. 1983 , publisher=

    1983

  24. [24]

    1933 , publisher=

    Grundbegriffe der wahrscheinlichkeitsrechnung , author=. 1933 , publisher=

    1933

  25. [25]

    Laplace, Pierre Simon , journal=. M

  26. [26]

    Laplace, Pierre Simon , volume=. Th. 1812 , publisher=

  27. [27]

    1814 , publisher=

    Essai philosophique sur les Prababilités , author=. 1814 , publisher=

  28. [28]

    science , volume=

    The parable of Google Flu: traps in big data analysis , author=. science , volume=. 2014 , publisher=

    2014

  29. [29]

    Nouvelles m

    Legendre, Adrien Marie , year=. Nouvelles m

  30. [30]

    International journal of control , volume=

    Input-output parametric models for non-linear systems part I: deterministic non-linear systems , author=. International journal of control , volume=. 1985 , publisher=

    1985

  31. [31]

    International journal of control , volume=

    Input-output parametric models for non-linear systems part II: stochastic non-linear systems , author=. International journal of control , volume=. 1985 , publisher=

    1985

  32. [32]

    1972 , publisher=

    Bayesian statistics: A review , author=. 1972 , publisher=

    1972

  33. [33]

    L'Oeuvre de Blaise Pascal , year =

  34. [34]

    The journal of chemical physics , volume=

    Equation of state calculations by fast computing machines , author=. The journal of chemical physics , volume=. 1953 , publisher=

    1953

  35. [35]

    1718 , publisher =

    de Moivre, Abraham , title =. 1718 , publisher =

  36. [36]

    Advances in neural information processing systems , volume=

    On the number of linear regions of deep neural networks , author=. Advances in neural information processing systems , volume=

  37. [37]

    Geophysics , volume=

    Quest for consistency, symmetry, and simplicity—The legacy of Albert Tarantola , author=. Geophysics , volume=. 2011 , publisher=

    2011

  38. [38]

    London: Philosophical Transactions of the Royal Society of London , year=

    On the problem of the most efficient tests of statistical , author=. London: Philosophical Transactions of the Royal Society of London , year=

  39. [39]

    Synthese , volume=

    Frequentist probability and frequentist statistics , author=. Synthese , volume=. 1977 , publisher=

    1977

  40. [40]

    The American Mathematical Monthly , volume=

    Pascal and the invention of probability theory , author=. The American Mathematical Monthly , volume=. 1960 , publisher=

    1960

  41. [41]

    Pascal, Blaise , year=. Trait

  42. [42]

    Recherches sur la probabilit

    Poisson, Sim. Recherches sur la probabilit. 1837 , publisher=

  43. [43]

    1931 , publisher=

    Foundations of mathematics and other logical essays , author=. 1931 , publisher=

    1931

  44. [44]

    The secularization of chance: Toward understanding the impact of the probability revolution on Christian belief in divine providence , author=. Zygon. 2015 , publisher=

    2015

  45. [45]

    2005 , publisher=

    Inverse problem theory and methods for model parameter estimation , author=. 2005 , publisher=

    2005

  46. [46]

    1954 , publisher=

    The foundations of statistics , author=. 1954 , publisher=

    1954

  47. [47]

    Journal of geophysics , volume=

    Inverse problems= quest for information , author=. Journal of geophysics , volume=

  48. [48]

    , author=

    The Logic of Chance. , author=. New York: Chelsea , year=

  49. [49]

    JC Willis, FR S , author=

    II—A mathematical theory of evolution, based on the conclusions of Dr. JC Willis, FR S , author=. Philosophical transactions of the Royal Society of London. Series B, containing papers of a biological character , volume=. 1925 , publisher=

    1925

  50. [50]

    Information and control , volume=

    Fuzzy sets , author=. Information and control , volume=. 1965 , publisher=

    1965

  51. [51]

    Fuzzy sets and systems , volume=

    Fuzzy sets as a basis for a theory of possibility , author=. Fuzzy sets and systems , volume=. 1978 , publisher=

    1978