Loading field map
Category theory began as a language for mathematical structure, but its own development repeatedly forced mathematicians to turn that language back on itself, creating new categorical tools to address problems that earlier categorical tools had revealed. The result is a field organized around thirteen major frameworks, each responding to tensions left by its predecessors, and each still active today in a pluralistic landscape.
The first framework, Natural Equivalence and Functoriality, emerged from Eilenberg and Mac Lane's 1945 paper that introduced categories, functors, and natural transformations. Their goal was to formalize the notion of naturality that had become central in algebraic topology—for example, the natural isomorphism between homology and cohomology. This framework provided the basic vocabulary for all later work, but it did not yet prescribe what kinds of categories were worth studying.
Abelian Categories and Homological Algebra answered that question by axiomatizing the properties needed for homological algebra. Grothendieck's 1957 Tôhoku paper showed that categories of sheaves, modules, and abelian groups all share a common structure: they are abelian categories with enough injectives. This framework narrowed the focus of early category theory to a class of categories that could support exact sequences and derived functors. It coexisted with the functorial language, using it to define homology in any abelian setting.
Adjoint Functors and Universal Constructions, discovered by Kan in 1958, unified the scattered universal constructions—limits, colimits, free objects, tensor products—under a single concept. Adjointness became the organizing principle of category theory, complementing naturality. Where abelian categories had emphasized exactness, adjoint functors revealed a deeper pattern: many constructions are left or right adjoints to forgetful functors. This framework transformed category theory from a language into a systematic tool.
Grothendieck Fibrations and Descent arose from algebraic geometry's need to glue local data. Grothendieck defined a fibration as a functor p: E → B such that morphisms in B can be lifted to cartesian morphisms in E. This provided a categorical framework for descent: a sheaf on a space can be seen as a cartesian section of a fibration. The fibration concept directly fed into the definition of a topos, where the sheaf condition is expressed via descent.
Topos Theory, developed by Grothendieck and Verdier in SGA4 (1960–1963), defined a topos as a category of sheaves on a site. This framework generalized topological spaces, allowing a notion of "space" that is purely categorical. A topos has an internal logic and supports a universe of sets without requiring a global set theory. Topos Theory absorbed Grothendieck fibrations by using them to formulate the sheaf condition, and it later provided a model for categorical foundations.
Lawvere Theories and Monadic Algebra took a different direction. Lawvere's 1963 thesis introduced algebraic theories as categories with finite products, with models being product-preserving functors. This gave a categorical foundation for universal algebra, independent of homological algebra. Monads (triples) later provided an equivalent but more flexible approach. This framework coexisted with abelian categories by focusing on algebraic rather than homological structure, and it directly influenced categorical logic.
Monoidal Category Theory emerged from Mac Lane's coherence theorem for monoidal categories (1963) and the subsequent development of braided and symmetric monoidal categories. A monoidal category replaces the Cartesian product with a tensor product, allowing categories to be "algebraic objects" themselves. This framework provided the infrastructure for enriched category theory and higher categories, and it remains essential for any setting where objects can be combined.
Categorical Foundations was Lawvere's attempt to axiomatize set theory using categorical concepts, beginning with his 1964 "Elementary Theory of the Category of Sets." This framework aimed to replace set-theoretic foundations with category-theoretic ones. It was influenced by Topos Theory, which provided a universe for set-like reasoning (a topos can serve as a model of set theory), and by Categorical Logic, which provided the internal logic of a topos. The two influences were distinct: Topos Theory supplied the categorical structure, while Categorical Logic supplied the logical interpretation. Categorical Foundations remains an active but minority program.
Enriched Category Theory, developed by Kelly and others from 1966 onward, generalized categories by replacing hom-sets with objects from a monoidal category. This allowed categories to be enriched over vector spaces, topological spaces, or any monoidal category. It transformed category theory by making the hom-object itself a structured entity, and it relied on monoidal category theory as its base. Enriched category theory is now a standard tool in representation theory and homotopy theory.
Categorical Logic reinterpreted logical systems as categories with certain structure. Lawvere's 1967 "Adjointness in Foundations" showed that a theory can be seen as a category with finite limits, and models are limit-preserving functors. This framework absorbed Lawvere theories by generalizing from algebraic to logical theories: an algebraic theory is a special case of a logical theory. It also provided the internal logic of toposes, linking Topos Theory and Categorical Foundations. Categorical Logic remains central to the study of type theory and the foundations of mathematics.
Homotopical Algebra and Model Categories was introduced by Quillen in 1967 as a framework for doing homotopy theory in any category. A model category has three distinguished classes of maps (fibrations, cofibrations, weak equivalences) satisfying axioms that allow the construction of a homotopy category. This framework challenged the centrality of abelian categories by showing that non-abelian settings (like topological spaces) could have a well-behaved homotopy theory. It also influenced Natural Equivalence and Functoriality by providing a new context for derived functors, which could now be defined in any model category. Homotopical algebra coexists with abelian categories, each being suited to different problems.
Higher Category Theory emerged from the need to capture homotopy coherence. Bicategories (1970s) and later (∞,1)-categories (1990s onward) extended ordinary categories by including morphisms between morphisms. This framework coexists with homotopical algebra: model categories provide a way to present (∞,1)-categories, and higher category theory extends the adjoint functor framework to higher dimensions. Higher category theory is now the dominant framework in algebraic topology and is increasingly used in geometry and representation theory.
Categorical Semantics of Computation began with Moggi's 1989 work on monads as computational effects. Moggi showed that monads (from Lawvere Theories and Monadic Algebra) could model side-effects, exceptions, and continuations in programming languages. This framework transformed category theory into a practical tool for computer science, providing a semantics for functional programming languages like Haskell. It coexists with categorical logic by offering a semantics for programming languages that parallels the semantics of logical theories. Today, this framework is central to the design of type systems and effect systems.
Today, the leading frameworks are Topos Theory (in geometry and logic), Higher Category Theory (in homotopy theory and algebraic topology), and Categorical Semantics of Computation (in computer science). They agree on the centrality of functoriality and adjointness, but they disagree on foundational commitments: topos theorists often embrace a structuralist foundation, while higher category theorists work within homotopy type theory or set theory. Monoidal Category Theory and Enriched Category Theory remain essential infrastructure, and Abelian Categories and Homological Algebra continue to be used in algebraic geometry and representation theory. The field is pluralistic, with no single framework dominating—each addresses different mathematical pressures, and their interplay continues to drive category theory forward.