Higher Category Theory

Ordinary category theory is built on arrows between objects, and on the rule that arrows compose: an arrow from A to B followed by one from B to C gives an arrow from A to C.

9 frameworks, from 1965 to today · Category Theory · Mathematics

still a working view ended · each leads to its entry below

Overview

Ordinary category theory is built on arrows between objects, and on the rule that arrows compose: an arrow from A to B followed by one from B to C gives an arrow from A to C1. Higher category theory asks what happens when there are also arrows between arrows, arrows between those, and so on upward2. The motivating examples come from topology and logic. A space has points, paths between points, homotopies between paths, homotopies between homotopies; a proof has statements, derivations, and transformations between derivations. In both cases the higher arrows are not merely extra data — they carry coherence conditions, and the central problem of the field has been to say precisely what those conditions are. The history of higher category theory is largely a history of competing answers to that question, and of the slow realization that several of them describe the same mathematical objects.

Bicategories and the first weakening (1960s)

The founding step was taken by Jean Bénabou in the mid-1960s. In an ordinary category, composition of arrows is strictly associative and identity arrows act strictly1. Bénabou observed that in many natural examples — categories whose objects are themselves categories, with functors between them and natural transformations between those — composition of 2-arrows holds only up to invertible 2-arrows, and the associativity and unit laws hold only up to coherent isomorphisms. He introduced the bicategory: objects, 1-arrows, 2-arrows, with composition associative and unital only up to specified invertible 2-cells, subject to coherence axioms23. The coherence conditions were the novel part; they ensure that any two ways of bracketing a composite are related by a canonical isomorphism, so the weakness never causes ambiguity.

The prototype is the bicategory of categories, functors, and natural transformations243. Bénabou's work, circulated in the 1960s and published in a widely read 1967 account, established the pattern that would recur: at each dimension, laws that were equalities one level down become isomorphisms one level up, and one must impose coherence. John Gray's 1974 book on formal category theory developed the theory of 2-categories and their homomorphisms and made the subject usable; the strict 2-category, where all laws hold on the nose, became the standard working setting for formal category theory.

Strict n-categories and the failure of strictness (1970s–1980s)

The obvious generalization is to iterate: a strict n-category has k-arrows for every k up to n, with strict associativity and interchange at every level2. Strict n-categories can be defined by iterated enrichment — a strict 2-category is a category enriched in categories, and so on2. This was understood early and is technically clean.

The trouble is that strictness is too rigid for the motivating examples. The fundamental n-groupoid of a space — points, paths, homotopies, higher homotopies — is only a weak structure, and it was known by the 1980s that not every weak 3-groupoid is equivalent to a strict one. So a definition of weak n-category was needed, and the field spent roughly two decades producing candidates. The problem is combinatorial: one must specify, at every dimension, which composites exist, which coherence cells relate them, and which coherence laws those cells satisfy, without either demanding too much (strictness) or losing control (an infinite regress of coherence data).

The homotopy hypothesis and higher groupoids

The organizing conjecture of the subject is the homotopy hypothesis: weak n-groupoids (higher categories in which every k-arrow for k ≥ 1 is invertible) should model homotopy n-types, the spaces that have no homotopy above dimension n. The idea goes back to Alexander Grothendieck's 1983 manuscript Pursuing Stacks, which proposed that homotopy types be defined as ∞-groupoids and sketched a program for proving it. The hypothesis gives the field its compass: it says that higher category theory is, at least in the groupoidal case, a combinatorial model for homotopy theory, and it turns every proposed definition into a testable claim — does it recover the right homotopy types?

Grothendieck's manuscript also proposed a specific combinatorial model, opetopes, higher-dimensional polytopes whose cells are shaped by iterated operation. The opetopic approach, developed in the 1990s by John Baez and James Dolan and later by Tom Leinster and others, defines weak n-categories as algebras for an operad-like structure of opetopic shapes. It was one of several competing definitions, and its influence was as much conceptual — it made precise the idea that higher categorical composition is governed by a globular or opetopic geometry of cells — as foundational.

Globular, operadic, and other definitions (1990s–2000s)

The 1990s and early 2000s saw a proliferation of definitions, each with a different geometric shape for cells and a different bookkeeping for coherence. Globular definitions take cells to be globes: a 0-cell is a point, a 1-cell an interval with two endpoints, a 2-cell a disk with a boundary circle, and so on; composition pastes globes along boundaries5. Simplicial definitions take cells to be simplices, so that an n-category is a simplicial set with extra structure — this lineage leads directly to quasicategories6. Operadic definitions, following the opetopic lead, describe higher composition as an algebra over a operad of cell shapes. Penon and Batanin gave definitions in terms of globular operads and contractible operads respectively, the latter making precise the idea that coherence data should itself be contractible.

The plurality was not chaos: a recurring theme was that the definitions should be equivalent in a suitable sense, and comparisons were proved case by case. The practical lesson, absorbed by the 2000s, was that the choice of shape is a matter of convenience, and that what matters is the homotopy-theoretic content.

Quasicategories and ∞-category theory (1970s–2000s)

The definition that came to dominate was not one of the weak n-category definitions at all, but a simplicial one. In 1973 Michael Boardman and Rainer Vogt introduced weak Kan complexes in their work on homotopy-invariant algebraic structures; André Joyal, from the early 1980s, developed the theory of what he called quasicategories — simplicial sets in which every inner horn has a filler — as a model for ∞-categories. The key insight is that a quasicategory encodes higher composition implicitly: a composable pair of 1-simplices has a filler 2-simplex witnessing a composite, and higher simplices witness the coherence, without any explicit algebraic composition law6. This sidesteps the combinatorial explosion of the globular definitions.

Jacob Lurie's Higher Topos Theory (2009) and Higher Algebra (2017) turned quasicategories into a working language for derived algebraic geometry and stable homotopy theory, and the theory of ∞-categories became the default. The price is that quasicategories are a model, not a definition of the Platonic object; the field's working philosophy is that ∞-categories form an (∞,1)-category of models, and one works inside a chosen model.

Simplicial and Segal-space models

Parallel to quasicategories, other simplicial and topological models were developed. Segal spaces, introduced by Graeme Segal in the late 1960s in the context of Γ-spaces and by Charles Rezk in 2001 as a model for (∞,1)-categories, encode composition via the Segal condition: the space of n-simplices is equivalent to the space of composable n-tuples of 1-arrows. Complete Segal spaces add a completeness condition ensuring that the higher cells really are invertible where they should be7. Simplicial categories — categories enriched in simplicial sets — give yet another model, and the comparison theorems of Julie Bergner and others in the mid-2000s established that these models are Quillen-equivalent, i.e. describe the same homotopy theory. This body of work, sometimes called the homotopy theory of homotopy theories, is what makes it legitimate to speak of "the" (∞,1)-category of (∞,1)-categories.

Homotopy type theory and higher-categorical logic (2000s–2010s)

A separate strand runs through logic. The Curry–Howard–Lambek correspondence identifies proofs with arrows and propositions with objects; Martin-Löf's dependent type theory, from the 1970s, gives a formal language in which this correspondence is native. In the 2000s, Vladimir Voevodsky, working on univalent foundations, observed that Martin-Löf type theory has a natural interpretation in which types are spaces, terms are points, and identity types are path spaces — so that the higher structure of homotopy theory is already present in the syntax. This became homotopy type theory (HoTT), developed in a collaborative 2013 book. Its central new principle, univalence, asserts that identity of types is equivalent to equivalence of types, which is a higher-categorical statement about the universe of types.

HoTT gave higher category theory a syntactic, machine-checkable home and connected it to proof assistants such as Coq and Agda8. It also raised foundational questions — whether univalence is consistent with the rest of type theory, and how to add higher inductive types — that remain active.

The present

Today the field has a stable core and live edges. Quasicategories and complete Segal spaces are the standard models for (∞,1)-categories, with comparison theorems making the choice a matter of taste; ∞-category theory is a working tool in algebraic geometry, homotopy theory, and mathematical physics. The theory of (∞,n)-categories for general n is less settled: several models exist (including iterated Segal and Θ-space models), and their comparison is an active project. The homotopy hypothesis is proved in low dimensions and in various model-dependent forms, but a fully general statement remains open. And the interaction with logic — HoTT, higher inductive types, and the search for a synthetic theory of ∞-categories — is where the field's foundational questions now concentrate.

References

  1. Category theory, Wikipedia
  2. Higher category theory (Strict higher categories, Weak higher categories), Wikipedia
  3. 2-category, Wikipedia
  4. Category of small categories, Wikipedia
  5. Globular set, Wikipedia
  6. Simplicial set (introduction, History and uses of simplicial sets), Wikipedia
  7. Simplicial space, Wikipedia
  8. Univalent foundations (History), Wikipedia

The frameworks of Higher Category Theory

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1965-Present

    Bicategories and 2-categories

    What laws govern arrows between arrows?

    Fly to its tower in Nousfarer

  2. 1970-1990

    Strict n-categories

    How do higher arrows compose strictly?

    Fly to its tower in Nousfarer

  3. 1980-Present

    Quasi-category

    How can higher composition be encoded implicitly?

    Fly to its tower in Nousfarer

  4. 1980-2010

    Weak n-category

    What coherence laws define weak higher arrows?

    Fly to its tower in Nousfarer

  5. 1983-Present

    Homotopy hypothesis

    Do higher groupoids model homotopy types?

    Fly to its tower in Nousfarer

  6. 1990-2010

    Opetopic and operadic definitions

    What cell shapes govern higher composition?

    Fly to its tower in Nousfarer

  7. 1995-Present

    Operadic and Globular Definitions

    Which operadic coherence makes higher categories?

    Fly to its tower in Nousfarer

  8. 2001-Present

    Segal-space and simplicial models

    Which simplicial models describe higher categories?

    Fly to its tower in Nousfarer

  9. 2006-Present

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.

Bicategories and 2-categoriesStrict n-categoriesQuasi-categoryWeak n-categoryHomotopy hypothesisOpetopic and operadic definitionsOperadic and Globular DefinitionsSegal-space and simplicial modelsHomotopy type theory

Beyond the field

Where Higher Category Theory's ideas came from in other fields, and what they led to there: the strongest 12 of 40 links.

A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Higher Category Theory 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 Higher Category Theory

Related fields

Fields its ideas are linked to

More in Category Theory

See where Higher Category Theory sits on the Map of Human Knowledge