Loading field map
Homotopy theory began with a simple question: when can two continuous maps between spaces be deformed into one another? This question quickly led to a deeper one: how can we classify spaces up to continuous deformation, ignoring rigid geometric shape and focusing only on the 'shape' that survives stretching and bending? The history of homotopy theory is a story of successive frameworks that each redefined what counts as a basic invariant, what methods are legitimate for computing it, and what kind of spaces are worth studying.
The first systematic framework for homotopy was the fundamental group, introduced by Henri Poincaré in the 1890s. The fundamental group assigns to a pointed space a group whose elements are loops based at that point, considered up to continuous deformation. This invariant could distinguish spaces that homology could not—for example, the circle and the sphere have the same homology groups but different fundamental groups. Covering space theory provided a geometric counterpart: a covering space of a space encodes information about its fundamental group, and the classification of coverings became a powerful tool for computing fundamental groups. This framework dominated the first decades of the twentieth century, but it had a limitation: the fundamental group captures only one-dimensional information. Spaces that are simply connected (with trivial fundamental group) are invisible to this invariant, yet they can be topologically distinct.
In the 1930s, Witold Hurewicz introduced higher homotopy groups, which capture information in every dimension. For a pointed space, the n-th homotopy group consists of maps from an n-dimensional sphere into the space, considered up to deformation. These groups are abelian for n ≥ 2, a fact that follows from the Eckmann–Hilton argument. Higher homotopy groups are much harder to compute than homology groups, but they are also more discriminating. The theory of fibrations, developed by Hurewicz and later by Jean-Pierre Serre, provided a crucial computational tool: a fibration is a map with the homotopy lifting property, and it gives a long exact sequence relating the homotopy groups of the total space, base space, and fiber. This framework coexisted with the fundamental group approach, but it shifted attention from low-dimensional invariants to a full sequence of invariants, one for each dimension. The difficulty of computing these groups, even for simple spaces like spheres, became a central pressure driving later developments.
In the 1940s, J. H. C. Whitehead introduced CW complexes, a class of spaces built by attaching cells of increasing dimension. CW complexes are flexible enough to model most spaces of interest in algebraic topology, yet rigid enough to allow inductive arguments. Whitehead's framework transformed homotopy theory by providing a combinatorial model for spaces, making it possible to prove theorems by induction on the number of cells. The cellular approximation theorem and Whitehead's theorem (a map between CW complexes that induces isomorphisms on all homotopy groups is a homotopy equivalence) became foundational. This framework did not replace higher homotopy groups; rather, it gave them a natural setting where computations could be organized. CW complexes also made precise the relationship between homotopy groups and homology groups, via the Hurewicz theorem and the Serre spectral sequence. By the 1950s, homotopy theory had a standard language: spaces are CW complexes, invariants are homotopy groups, and computations rely on fibrations and spectral sequences.
A major turning point came in the 1950s with the realization that homotopy groups become more regular after suspension—the operation of wedging a space with a circle. The Freudenthal suspension theorem showed that for a connected CW complex, the homotopy groups in a range stabilize as the suspension coordinate increases. This led to the concept of the stable homotopy groups of spheres, which are the homotopy groups of spheres after enough suspensions. Stable homotopy theory studies phenomena that are invariant under suspension, captured by the stable homotopy category. In this category, the sphere spectrum replaces the sphere as the fundamental object, and the stable homotopy groups of spheres become a graded ring. This framework absorbed earlier work on homotopy groups by providing a setting where many computations become systematic. The Adams spectral sequence, introduced by J. Frank Adams, became the primary tool for computing stable homotopy groups. Stable homotopy theory remains active today, with deep connections to algebraic K-theory, cobordism, and chromatic homotopy theory.
By the 1960s, homotopy-theoretic ideas had spread beyond topological spaces to other categories, such as chain complexes and simplicial sets. Daniel Quillen, in his 1967 book Homotopical Algebra, introduced model categories as an axiomatic framework for doing homotopy theory in any category that has a suitable notion of weak equivalence, fibration, and cofibration. A model category is a category equipped with three distinguished classes of maps satisfying axioms that allow one to define the homotopy category by formally inverting weak equivalences. This framework did not replace stable homotopy theory; instead, it provided the infrastructure that made stable homotopy theory rigorous and generalizable. The stable homotopy category itself can be constructed as the homotopy category of a model structure on spectra. Model categories also enabled homotopy theory to be applied in algebra, giving rise to derived categories and the theory of simplicial rings. Today, model categories are a standard tool, though they coexist with other approaches such as ∞-categories, which offer a different axiomatization of the same underlying ideas.
In the 1970s, Dennis Sullivan and Daniel Quillen independently developed rational homotopy theory, which studies spaces after ignoring torsion information. The idea is to tensor homotopy groups with the rational numbers, or equivalently, to consider the homotopy theory of spaces up to rational equivalence. Rational homotopy theory is dramatically simpler than integral homotopy theory: rational homotopy groups of simply connected spaces are graded vector spaces, and the rational homotopy type of a simply connected space is encoded by a commutative differential graded algebra (the Sullivan minimal model) or by a differential graded Lie algebra (the Quillen model). This framework narrowed the focus from all homotopy information to the rational part, but it made that part completely computable for many spaces. Rational homotopy theory coexists with stable homotopy theory; it is a special case where computations become algebraic and tractable. It remains active, with applications to the study of mapping spaces, configuration spaces, and the homotopy theory of operads.
The most recent framework on this timeline is A¹ homotopy theory, introduced by Fabien Morel and Vladimir Voevodsky in the late 1990s. This framework extends homotopy theory from topological spaces to algebraic varieties, using the affine line A¹ as a stand-in for the unit interval. In A¹ homotopy theory, two maps between varieties are considered homotopic if they can be deformed using the affine line, and the resulting homotopy category captures deep algebraic invariants such as motivic cohomology and algebraic K-theory. This framework is a radical extension of classical homotopy theory: it imports the entire apparatus of model categories, stable homotopy theory, and rational homotopy theory into algebraic geometry. The stable version, motivic stable homotopy theory, produces a motivic sphere spectrum and motivic homotopy groups that encode information about the arithmetic and geometry of fields. A¹ homotopy theory does not replace earlier frameworks; rather, it applies their methods to a new class of objects, creating a bridge between homotopy theory and algebraic geometry.
Today, several frameworks remain active and serve different purposes. Stable homotopy theory is the workhorse for computations in algebraic topology, especially through the Adams spectral sequence and chromatic methods. Model categories provide the foundational language for homotopy theory in any categorical setting, though they are increasingly supplemented by ∞-categories, which offer a more flexible approach. Rational homotopy theory remains the go-to tool for studying the rational aspects of spaces, and it has found new applications in geometric topology and mathematical physics. A¹ homotopy theory is a vibrant area of current research, connecting homotopy theory to number theory and algebraic geometry.
These frameworks agree on the basic principles: homotopy theory is about studying objects up to weak equivalence, and invariants should be derived from functors that preserve weak equivalences. They disagree on what the most natural objects are (topological spaces, spectra, simplicial sets, or motivic spaces) and on what level of generality is most useful. Stable homotopy theory and rational homotopy theory are more computational, while model categories and A¹ homotopy theory are more structural. The tension between computation and abstraction continues to drive the field forward, with each framework offering a different balance between the two.