Loading field map
For much of the twentieth century, the philosophy of mathematics was dominated by a single question: which foundational system—logicism, formalism, or intuitionism—could finally secure all of mathematics on a firm footing? The failure of each program to deliver a complete, consistent, and intuitively satisfying foundation, sharpened by Gödel's incompleteness theorems, gradually eroded confidence that any single system could capture the whole of mathematical truth. By the 1970s, a growing number of philosophers began to suspect that the foundationalist project had been asking the wrong question. Instead of legislating what mathematics should be from an external philosophical standpoint, they argued, philosophers should attend to what mathematicians actually do: the practices of proving, conjecturing, revising, computing, and communicating that constitute the living discipline. This shift in attention—from foundations to practice—gave rise to a new subfield, the philosophy of mathematical practice, whose four main frameworks have reshaped philosophical engagement with mathematics over the past half-century.
The three great foundational programs of the early twentieth century—logicism (the reduction of mathematics to logic), formalism (the treatment of mathematics as a game of symbol manipulation), and intuitionism (the restriction of mathematics to mental constructions)—each aimed to provide a single, universal foundation for all mathematical knowledge. By mid-century, however, each had encountered serious obstacles. Logicism required axioms that seemed more mathematical than logical; formalism could not prove its own consistency in any interesting sense; and intuitionism rejected large swaths of classical mathematics that most mathematicians found indispensable. Gödel's incompleteness theorems (1931) dealt a further blow by showing that any sufficiently powerful formal system cannot prove its own consistency. The dream of a single, self-justifying foundation faded, and philosophers began to look for alternative ways of understanding mathematical knowledge that did not depend on a unique foundational system.
The first major framework to foreground mathematical practice was Imre Lakatos's heuristic philosophy, articulated in his 1976 book Proofs and Refutations. Lakatos rejected the formalist picture of mathematics as a static, deductive structure and instead portrayed it as a dynamic, fallible enterprise driven by conjecture, counterexample, and conceptual refinement. He studied historical episodes—most famously the development of the Euler characteristic for polyhedra—to show how mathematical concepts are not given in advance but are gradually generated through the interplay of proof and refutation. A proposed proof is met with a counterexample; the concept is refined to exclude the counterexample; a new proof is attempted; and the cycle continues. For Lakatos, this quasi-empirical process is not a regrettable departure from ideal rigor but the very engine of mathematical growth.
Lakatos's framework was a direct challenge to the foundationalist assumption that mathematical knowledge must be grounded in a fixed, axiomatic system. By treating mathematics as a historically situated, fallible activity, he opened the door for philosophers to study mathematical practice on its own terms. His work also introduced a distinctive methodological commitment: the philosopher of mathematics should be a historian of mathematics, tracing the actual development of concepts and proofs rather than reconstructing them in an idealized formal language. This historical orientation set Lakatos apart from later practice-oriented frameworks that would adopt a more synchronic, descriptive approach.
Penelope Maddy's mathematical naturalism, developed from the mid-1980s onward, offered a different way of taking mathematical practice seriously. Where Lakatos looked to history, Maddy looked to the current methods and standards of mathematics as the proper source of philosophical norms. In her 1997 book Naturalism in Mathematics, she argued that the philosopher should not impose external epistemological or metaphysical criteria on mathematics but should instead describe and justify mathematical practice from within. For example, when mathematicians accept the existence of large cardinal axioms because they unify and explain set-theoretic phenomena, the naturalist philosopher does not ask whether those axioms are epistemically justified by some independent standard; she asks whether they are justified by the standards of mathematics itself.
Maddy's naturalism differs from Lakatos's heuristic philosophy in a crucial respect. Lakatos treated mathematical practice as a historical process of conjecture and refutation, and he was willing to criticize actual mathematical developments (for example, the formalist turn in analysis) as a betrayal of the heuristic method. Maddy, by contrast, treats contemporary mathematical practice as largely authoritative: if mathematicians collectively accept a method or a result, the naturalist philosopher has no standing to reject it from an external philosophical vantage point. This makes Maddy's naturalism more conservative than Lakatos's framework—it tends to endorse mainstream mathematical practice rather than critique it—but also more systematic, since it aims to provide a philosophical account of mathematics that is continuous with the discipline's own self-understanding.
The rise of powerful computers from the 1990s onward introduced a new dimension to mathematical practice and, with it, a new philosophical challenge. Experimental mathematics—the use of computation for exploration, conjecture generation, and even proof verification—raised questions about the nature of mathematical evidence and the role of traditional proof. When a computer checks billions of cases to verify the four-color theorem (1976) or when it discovers patterns that lead to new conjectures, is the resulting knowledge genuinely mathematical? Does a computer-assisted proof meet the same epistemic standards as a traditional proof?
Experimental mathematics is best understood as a methodological school rather than a full-fledged philosophical framework: it does not offer a comprehensive theory of mathematical truth or ontology, but it does make a distinctive philosophical claim about the scope of legitimate mathematical evidence. Proponents argue that computation can function as a legitimate partner in mathematical discovery and, in some cases, as a component of proof itself. This claim puts experimental mathematics in a productive tension with both Lakatosian and naturalist approaches. Lakatos's heuristic model, with its emphasis on human conjecture and counterexample, did not anticipate the role of automated exploration; Maddy's naturalism, which takes current mathematical practice as normative, must accommodate the fact that many mathematicians remain skeptical of computer-dependent proofs. The experimental mathematics movement thus forces the philosophy of mathematical practice to confront the question: what counts as a mathematical proof, and who—or what—can produce one?
Mathematical pluralism, which emerged around the same time as experimental mathematics, takes the diversity of mathematical systems as a permanent and philosophically significant feature of the discipline. Pluralists argue that there is no single, correct foundation for mathematics; instead, multiple foundational systems—classical set theory, category theory, intuitionistic logic, homotopy type theory, and others—are equally legitimate, each suited to different mathematical purposes. This view contrasts sharply with the foundationalist assumption that one system must be privileged, and it also differs from Maddy's naturalism, which tends to focus on the mainstream set-theoretic practice that most mathematicians actually employ.
Pluralism is not merely a descriptive observation about the existence of multiple systems; it is a normative claim about their equal standing. A pluralist philosopher does not see the coexistence of, say, classical and intuitionistic mathematics as a problem to be resolved but as a resource to be exploited. Different mathematical questions may be best addressed within different frameworks, and the philosopher's task is to understand how these frameworks relate to one another—through translation, interpretation, or mutual enrichment—rather than to reduce them all to a single foundation.
The relationship between pluralism and experimental mathematics is complex. On one hand, the computational turn has made pluralism more plausible by providing practical tools for working within multiple systems; on the other hand, experimental mathematics often relies on classical logic and set-theoretic assumptions, which could be seen as reinforcing a single mainstream framework rather than licensing pluralism. The two frameworks thus coexist in a state of productive tension: experimental mathematics expands the range of acceptable methods, while pluralism expands the range of acceptable systems.
Today, all four frameworks remain active, and together they define the subfield's central debates. There is broad agreement that the philosophy of mathematics must attend to mathematical practice—that foundationalist programs that ignore what mathematicians actually do are no longer tenable. There is also a shared commitment to taking mathematical knowledge as it is, rather than as it ought to be according to some external philosophical theory.
Yet significant disagreements persist. The most fundamental divide is between Lakatos's historical fallibilism and Maddy's synchronic naturalism: should the philosopher of mathematics study the historical development of concepts, or should she describe and justify current practice as it stands? A second divide concerns the status of experimental methods: do computer-assisted proofs represent a genuine transformation of mathematical practice, or are they merely a supplement to traditional proof? A third divide separates pluralists from naturalists: does naturalism's focus on mainstream practice tacitly endorse a monistic picture of mathematics, or can it accommodate the pluralist's insistence on multiple legitimate systems?
These disagreements are not signs of weakness but of a healthy, evolving subfield. The philosophy of mathematical practice has established itself as a distinct area of inquiry, with its own journals, conferences, and research programs. Its central insight—that mathematics is a human activity worthy of philosophical study in its own right—has permanently changed the way philosophers think about the subject. The foundationalist dream of a single, universal foundation may be dead, but the questions that motivated it—about mathematical truth, knowledge, and objectivity—have been given new life by being asked in a new way: from within the practice itself.