- Atlas
- Philosophy
- Mathematical Explanation
Mathematical Explanation
Mathematical explanation is the study of what it is for mathematics to explain something—why a proof can do more than establish that a theorem is true, and why a mathematical fact can sometimes explain a physical or mathematical phenomenon rather than merely describe it.
9 frameworks, from 350 BCE to today · Philosophy Of Mathematics · Philosophy
- Aristotelian Demonstrative Explanation 350 BCE-1600
- Deductive-Nomological Model 1948-1975
- Causal Theory of Explanation 1971-1995
- Unificationist Account 1974-2010
- Explanatory Proof in Pure Mathematics 1980-Present
- Mathematical Explanation in Science 1980-Present
- Understanding-Based Account 2004-Present
- Distinctively Mathematical Explanation 2005-Present
- Pluralism about Explanation 2010-Present
Overview
Mathematical explanation is the study of what it is for mathematics to explain something—why a proof can do more than establish that a theorem is true, and why a mathematical fact can sometimes explain a physical or mathematical phenomenon rather than merely describe it. The stakes are real for both philosophy and practice. Mathematicians often say that one proof "explains" while another merely "proves," and scientists sometimes invoke mathematics not as a calculational tool but as the reason a phenomenon must be the way it is. If those judgments are meaningful, philosophy needs an account of what makes them true. If they are not, a great deal of mathematical and scientific rhetoric needs deflating. The history of the subfield is largely the history of attempts to make sense of that distinction, first within mathematics and then across the mathematics–science boundary.
Aristotle and the Origins of Demonstrative Explanation
The earliest systematic treatment comes from Aristotle's Posterior Analytics, where he distinguishes between knowing a fact (hoti) and knowing the reasoned fact (dioti). To know that the planets are near the Earth is one thing; to know why—because they do not twinkle, and non-twinkling belongs to nearness—is another. For Aristotle, a proper demonstration (apodeixis) explains by displaying the cause: the middle term of a syllogism must be the reason the conclusion holds, not merely a co-occurring feature. This gave the first clear criterion for explanatory proof: an argument explains when its premises identify the cause of what is concluded, and a demonstration that runs from effect to cause (or from a coincidental middle term) fails to explain even if it is valid.
Aristotle's framework set the terms for centuries of debate. It also embedded a tension that never disappeared: explanation was tied to causation and to a hierarchical structure of sciences, so mathematical demonstrations—which for Aristotle often lacked efficient causes—sat awkwardly within it. The question of whether mathematical proofs can be explanatory in the Aristotelian sense, or whether mathematics explains only by revealing formal structure, runs through the whole later history.
The Early Modern Split: Proof and Explanation in Mathematics
Through the medieval and early modern periods, the Aristotelian distinction survived mainly in logic and in the organization of the sciences, while mathematics developed its own informal culture of explanatory proof. Descartes, Leibniz, and Newton all distinguished between showing that a result holds and showing why, though not always in those words. Leibniz's insistence on sufficient reason, and his search for proofs that display the "why," is an early modern echo of the dioti ideal. Newton's Principia is instructive: its mathematical derivations from laws of motion were presented as demonstrations, but Newton and his readers debated which derivations genuinely explained the phenomena and which were merely predictive bookkeeping—a debate that anticipated twentieth-century discussions of explanation in science.
By the nineteenth century, the rise of rigor in analysis—Cauchy, Bolzano, Weierstrass—created a new context. Proofs were being rewritten to meet standards of logical precision, and some of the resulting proofs were widely felt to be less explanatory than the informal arguments they replaced. The epsilon-delta definition of a limit made theorems provable but, many mathematicians complained, obscured why they were true. This tension between rigor and insight became a recurring theme: the more a proof is formalized, the less it may seem to explain.
The Deductive-Nomological Model and Its Discontents
The modern philosophical discussion begins with the logical empiricists. Carl Hempel and Paul Oppenheim's deductive-nomological (DN) model, developed in the 1940s and 1950s, held that to explain an event is to deduce a statement describing it from general laws plus initial conditions. Explanation and prediction became structurally identical: the same derivation that lets you predict also explains1. Applied to mathematics, the model suggested that a mathematical explanation of a physical fact is a derivation of that fact from mathematical laws or principles together with boundary conditions.
The DN model had the virtue of precision and the defect of over-permissiveness. A derivation from any true law explains, on this account, even when the law is irrelevant to the phenomenon—the classic counterexamples involve barometers and storms, or flags and shadows, where a valid derivation exists but the "explanation" is clearly not explanatory1. Within mathematics, the model struggled to distinguish proofs that explain from proofs that merely derive. Hempel himself was cautious about extending the model to mathematics, and the difficulties pushed later philosophers toward causal and then non-causal accounts.
The Causal Theory and the Problem of Mathematical Explanation
In the 1960s and 1970s, Wesley Salmon and others developed the causal theory of explanation: to explain an event is to exhibit its causal history. This worked well for many scientific cases and sharpened the contrast with mathematics, where causes in the ordinary sense seem absent. The causal theory thus raised a pointed question: if explanation is causal, can mathematics explain anything at all? Either mathematical explanations must be reduced to causal ones (mathematics describes causal structure), or the causal theory must be supplemented.
The debate crystallized around examples in which mathematics appears to do genuine explanatory work in science. A frequently discussed case is the explanation of the life cycle of the cicada: periodical cicadas have prime-numbered life cycles (13 or 17 years), and the standard explanation is that prime periods minimize overlap with predators that have shorter, periodic cycles. The mathematics of prime numbers—specifically, the fact that primes have no divisors other than 1 and themselves—is offered as the reason the life cycle is what it is. Whether this is a genuine mathematical explanation or a biological one dressed in mathematical clothing became a test case for the field.
Mathematical Explanation in Science
From the 1980s onward, philosophers began systematically asking whether mathematics can explain physical phenomena. Mark Steiner's The Applicability of Mathematics as a Philosophical Problem (1998) argued that mathematics sometimes plays an ineliminable explanatory role, and that this raises deep questions about why abstract structures fit the world. The cicada case, the hexagonal honeycomb, the optimality of certain physical configurations, and the use of topological or group-theoretic facts to explain physical possibilities all entered the literature2.
Two broad positions emerged. One holds that mathematical explanations in science are real and irreducible: the mathematics does explanatory work that no purely physical description could do. The other holds that such explanations are either causal explanations in disguise or pragmatic redescriptions, and that the appearance of mathematical explanation is an artifact of how we represent the world. The debate turned on criteria for explanatory relevance: what makes a mathematical fact the reason a phenomenon is as it is, rather than merely a feature that happens to correlate with it.
Unification and the Search for a General Account
Parallel to the science-focused debate, philosophers developed general accounts of explanation and tested them on mathematics. The unificationist account, associated with Michael Friedman and later Philip Kitcher, holds that an explanation is an argument that reduces the number of independent assumptions needed to derive a body of results—it unifies. On this view, a mathematical proof explains when it shows how a theorem follows from a small set of principles that also generate other theorems, thereby displaying the theorem as part of a coherent system. Kitcher's The Advancement of Science (1993) developed this for science generally, and the idea was applied to mathematical cases where a proof is explanatory because it reveals a unifying pattern.
Unificationism captured something real about mathematical practice—mathematicians do value proofs that reveal connections—but it faced counterexamples: some unifying derivations seem not to explain, and some explanatory proofs are not especially unifying. The account also struggled to say precisely what counts as unification without circularity.
Distinctively Mathematical Explanation
A major turning point came with the articulation of "distinctively mathematical explanation," especially in work by Alan Baker and later Marc Lange. Baker's 2005 paper on the cicada case argued that the prime-number fact explains the life cycle in a way that is not reducible to causal explanation, and he proposed criteria for when a mathematical fact is explanatory. Lange's Because Without Cause (2017) developed a systematic account: a distinctively mathematical explanation is one in which the explanans is a mathematical fact and the explanation works by showing that the phenomenon is mathematically necessary—it could not have failed to occur given the mathematics, regardless of the physical details. On Lange's view, such explanations are non-causal and involve mathematical necessity rather than natural necessity.
This work sharpened the field's central question: what kind of modality does mathematical explanation invoke? If the cicada's prime cycle is explained by the mathematics of divisors, the explanation seems to say that no non-prime cycle could have avoided predator overlap—a claim of mathematical, not biological, necessity. Critics pressed whether such necessity is genuine or whether the biological facts are doing the real explanatory work and the mathematics is a convenient representation.
Pluralism and the Understanding-Based Turn
By the 2010s, the field had accumulated enough competing accounts that pluralism became a serious option. Pluralism about explanation holds that there is no single correct account of what explanation is; different explanations succeed in different ways—causal, unifying, mathematical, mechanistic—and the philosopher's job is to describe the variety rather than reduce it. Applied to mathematical explanation, pluralism suggests that explanatory proofs may explain in different ways depending on context: some by revealing cause-like structure, some by unifying, some by displaying necessity, some by enabling understanding.
The understanding-based account, developed by philosophers including Henk de Regt in the science context and applied to mathematics by others, shifts the criterion from objective structure to cognitive achievement: an explanation is what produces understanding in a suitably prepared agent. On this view, a proof explains when it makes the theorem intelligible—when it lets a mathematician see why it must be true, not just that it is. This account has the advantage of matching mathematicians' actual judgments, which are often about insight and grasp rather than about formal properties. Its difficulty is that understanding is agent-relative and hard to characterize precisely, risking a slide into "explanation is whatever convinces someone."
The Present State of the Field
Today the subfield is active on several fronts. The debate over mathematical explanation in science continues, with cases from biology, physics, and computer science used to test whether mathematics can be explanatory without causation. The project of characterizing explanatory proof within pure mathematics—what makes one proof of the Pythagorean theorem explanatory and another not—remains open, with candidates including symmetry, unification, and the display of essential structure. The relationship between explanation and understanding is a live area, as is the question of whether explanation is a single phenomenon or a family of related achievements. Pluralism has gained ground as a description of practice, though not as a settled theory. What unites the field is the conviction, inherited from Aristotle and sharpened by twentieth-century philosophy of science, that the difference between proving and explaining is real and worth understanding—and that mathematics, which seems to prove so much and explain so little in causal terms, is the hardest and most revealing place to study it.
References
- Deductive-nomological model, Wikipedia
- Explanatory indispensability argument, Wikipedia
The frameworks of Mathematical Explanation
Each framework is a way of working the field has used: when it held, and the question it answers.
- 350 BCE-1600
Aristotelian Demonstrative Explanation
What makes a demonstration explain rather than merely prove?
- 1948-1975
Deductive-Nomological Model
When does a derivation from laws explain an event?
- 1971-1995
- 1974-2010
Unificationist Account
Does explanation consist in unifying many facts under few principles?
- 1980-Present
Explanatory Proof in Pure Mathematics
What makes a mathematical proof explanatory?
- 1980-Present
Mathematical Explanation in Science
Can mathematics explain physical phenomena?
- 2004-Present
- 2005-Present
Distinctively Mathematical Explanation
Are some explanations distinctively mathematical?
- 2010-Present
Pluralism about Explanation
Are there multiple legitimate forms of explanation?
How the ideas connect
Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.
Causal Theory of Explanation reacted against Deductive-Nomological Model
The causal theory arose directly as a rejection of the DN model's over-permissiveness, arguing that derivation from laws is not sufficient and that explanation must trace causal history.
Carried by Wesley Salmon.
Distinctively Mathematical Explanation grew out of Mathematical Explanation in Science
Distinctively mathematical explanation grew out of the debate on mathematical explanation in science, with Baker's cicada paper arguing that the prime-number fact provides an irreducible mathematical explanation.
Carried by Alan Baker.
Deductive-Nomological Model influenced by Aristotelian Demonstrative Explanation
Hempel and Oppenheim's deductive-nomological model inherited Aristotle's idea that explanation is a deductive argument displaying why the conclusion must hold, recast in terms of general laws and initial conditions rather than causes.
Carried by Carl Hempel; Paul Oppenheim.
Unificationist Account reacted against Deductive-Nomological Model
Friedman's unificationist account was proposed as an alternative to the DN model, arguing that explanation consists in reducing the number of independent assumptions rather than in deriving from laws.
Carried by Michael Friedman.
Mathematical Explanation in Science influenced by Causal Theory of Explanation
The causal theory's challenge that explanation must be causal framed the question of whether mathematics can explain physical phenomena, prompting the search for non-causal mathematical explanations in science.
Carried by Wesley Salmon.
Distinctively Mathematical Explanation reacted against Causal Theory of Explanation
The distinctively mathematical explanation account was framed as a challenge to the causal theory, arguing that some explanations are mathematical rather than causal.
Carried by Alan Baker; Marc Lange.
Explanatory Proof in Pure Mathematics influenced by Unificationist Account
Unificationist ideas about explanatory proofs that reveal connections influenced the development of accounts of explanatory proof in pure mathematics, which often appeal to unification as a mark of explanation.
Carried by Philip Kitcher.
Understanding-Based Account influenced by Unificationist Account
The understanding-based account drew on unificationist themes, treating explanation as what produces understanding, though it shifted focus from unification to cognitive grasp.
Understanding-Based Account influenced by Explanatory Proof in Pure Mathematics
Work on explanatory proof in pure mathematics, which often ties explanation to understanding why a theorem holds, contributed to the development of understanding-based accounts of explanation.
Pluralism about Explanation influenced by Understanding-Based Account
The understanding-based account, by emphasizing the diverse epistemic functions of explanation, helped motivate pluralist views that allow multiple legitimate forms of explanation.
Pluralism about Explanation influenced by Distinctively Mathematical Explanation
The distinctively mathematical explanation debate, by showing that mathematical explanations do not fit causal models, supported pluralist conclusions that different kinds of explanation coexist.
Beyond the field
Where Mathematical Explanation's ideas came from in other fields, and what they led to there: the strongest 12 of 32 links.
Causal Theory of Explanation reacted against Deductive-nomological model in Scientific Explanation
The causal theory of explanation arose as a rejection of the deductive-nomological model, arguing that subsumption under laws is neither necessary nor sufficient and that citing causes is what explains.
Carried by Wesley Salmon.
Causal Theory of Explanation led to Causal-process account in Scientific Explanation
Wesley Salmon's causal theory of explanation, developed from the 1970s, supplied the causal-connexion requirement that the causal-process account of explanation built on and refined. The later account treats causal processes and interactions as the explanatory core, extending Salmon's programme.
Carried by Wesley C. Salmon.
Understanding-Based Account grew out of Mathematical explanation in Mathematical Practice
The understanding-based account of mathematical explanation was developed as a refinement of the earlier debate over whether mathematics can genuinely explain, taking Steiner's 1978 question of explanatory proofs as its starting point and recasting explanation in terms of mathematical understanding.
Carried by Mark Steiner; Philip Kitcher.
Distinctively Mathematical Explanation grew out of Mathematical explanation in Mathematical Practice
The distinctively mathematical explanation framework grew directly out of the broader literature on mathematical explanation, narrowing the question to cases where mathematics itself does the explanatory work in empirical science.
Carried by Mark Steiner.
Pluralism about Explanation grew out of Explanatory pluralism in Scientific Explanation
Pluralism about explanation in mathematics grew out of the broader explanatory pluralism debate, extending the claim that different sciences need different explanatory forms to the case of mathematical explanation.
Aristotelian Demonstrative Explanation influenced by Aristotelian modal syllogistic in Modal Logic
Aristotle's modal syllogistic in the Prior Analytics supplies the necessity and modality apparatus on which demonstrative explanation in the Posterior Analytics depends, since demonstrations must proceed from necessary premises.
Carried by Aristotle.
Deductive-Nomological Model influenced by Hypothetico-deductive model in Confirmation And Evidence
The hypothetico-deductive account of confirmation, developed by Hempel and earlier by Popper, supplied the logical apparatus of derivation from laws and antecedent conditions that Hempel and Oppenheim turned into the covering-law model of explanation.
Carried by Carl Hempel; Karl Popper.
Deductive-Nomological Model led to Inductive-statistical model in Scientific Explanation
Hempel's inductive-statistical model was framed as an extension of the deductive-nomological model to probabilistic laws, defining itself by modifying the earlier account.
Carried by Carl Hempel.
Deductive-Nomological Model provoked Statistical-relevance model in Scientific Explanation
The statistical-relevance model was put forward by Salmon as a replacement for the deductive-nomological account of statistical explanation, which he argued failed on cases of irrelevant premises.
Carried by Wesley Salmon.
Causal Theory of Explanation rival of Explanatory unification in Scientific Explanation
Explanatory unification was developed by Friedman and Kitcher as a rival to causal theories of explanation, competing over what makes an explanation explanatory.
Carried by Michael Friedman; Philip Kitcher.
Causal Theory of Explanation influenced by Probabilistic Causation in Philosophy Of Causation
Probabilistic theories of causation, especially Suppes's probabilistic causality, supplied the notion of causal relevance that Salmon and others used to replace mere statistical relevance in explaining events.
Carried by Patrick Suppes.
Causal Theory of Explanation influenced by Counterfactual Theory in Philosophy Of Causation
The causal theory of explanation held that to explain an event is to give information about its causal history. Lewis's counterfactual analysis of causation supplied the account of that causal history which the explanation theory then invoked.
Carried by David Lewis.

Mathematical Explanation is a planet in Nousfarer
Its city has a tower for each of these 9 frameworks. Land, walk in, and climb one concept at a time.
Land on Mathematical ExplanationRelated fields
Fields its ideas are linked to
- Scientific Explanation 19 links
- Mathematical Practice 5 links
- Philosophy Of Mathematics 2 links
- Philosophy Of Causation 2 links
- Epistemology 1 link
- Modal Logic 1 link
- Confirmation And Evidence 1 link
- Philosophy Of Science 1 link
More in Philosophy Of Mathematics
See where Mathematical Explanation sits on the Map of Human Knowledge