Chapter I: On the Nature of Mathematical Reasoning — Summary¶
Analytical Digest of the Prefaces, Introduction & Part I (Chapter I)
1. Architecture of Scientific Hypotheses (Introduction)¶
Poincaré opens La Science et l’Hypothèse by rejecting both naive dogmatism (that science deduces reality infallibly from self-evident axioms) and superficial nominalism (as radicalized by Édouard Le Roy, who treated scientific laws as arbitrary caprices). Instead, Poincaré distinguishes three fundamentally different kinds of hypotheses across the hierarchy of the sciences:
Verifiable Empirical Generalizations — fruitful physical hypotheses tested continually against experiment;
Indifferent Representational Hypotheses — auxiliary images (such as continuous fluid vs. atomic models in elasticity or optics) useful for fixing thought without altering observable differential equations;
Apparent Hypotheses or Disguised Conventions (conventions ou définitions déguisées) — decrees of the mind guided, yet never logically or empirically forced, by experience, which give geometry and classical mechanics their exactitude.
Even as conventions frame our language, science reaches genuine objective reality because it grasps invariant relations between things (les rapports entre les choses).
2. Chapter I: Synthetic A Priori Reasoning by Recurrence¶
How can pure mathematics be simultaneously rigorous (free of empirical approximation) and creative (not a gigantic tautology repeating $A = A$)?
Verification vs. Proof (§§ I–II): Examining Leibniz’s analytical proof that $2 + 2 = 4$ from $x + 2 = (x + 1) + 1$, Poincaré observes that purely syllogistic substitution only performs a particular verification; it cannot establish a universal law over infinitely many integers.
Recursive Definitions & Cascade of Syllogisms (§§ III–V): Defining addition $x + a = [x + (a - 1)] + 1$ and multiplication recursively, Poincaré proves associativity, commutativity, and distributivity via reasoning by recurrence: establishing a theorem for $n = 1$, proving that if it holds for $n - 1$ it holds for $n$, and concluding it holds for all positive integers.
The Irreducible Synthetic *A Priori* Intuition (§§ VI–VIII): Complete mathematical induction condenses an infinite cascade of hypothetical syllogisms into a single formula. Because neither formal logic (restricted to finite syllogisms) nor empirical induction (restricted to finite observations) can license this leap to the infinite, Poincaré affirms—against Kant on geometry, yet with Kant on arithmetic—that mathematical induction is the true synthetic a priori intuition of the mind’s power to conceive the indefinite repetition of a once-possible act.