The Yoneda Lemma

Core Idea

The Yoneda lemma states that for any functor F: C → Set and any object A in C, there is a bijection Nat(Hom(A,-), F) ≅ F(A) that is natural in both A and F. This means natural transformations from a representable functor to any functor F are completely determined by a single element of F(A). The Yoneda embedding A ↦ Hom(A,-) is a fully faithful functor from C to [C^op, Set], showing every category embeds into a presheaf category and that an object is completely determined by how morphisms map into it from all other objects.

How It's Best Learned

Work through the proof step by step: given a natural transformation η: Hom(A,-) ⇒ F, evaluate at A and apply to id_A to get an element of F(A). Verify this map is an inverse to the map F(A) → Nat(Hom(A,-), F) given by the Yoneda construction. Appreciate the slogan 'an object is determined by its relationships with all other objects'.

Common Misconceptions

• The Yoneda lemma is not merely a technical result; it is a fundamental philosophical principle of category theory—objects are determined by their morphisms.
• The bijection Nat(Hom(A,-), F) ≅ F(A) is not a set isomorphism requiring special conditions; it holds for all locally small categories.
• The Yoneda embedding is contravariant when using Hom(-,A); be careful about which version (covariant vs. contravariant) is being used.

Explainer

To understand the Yoneda lemma, start with what a natural transformation from Hom(A,-) to F actually is. Hom(A,-) is a functor that sends each object X to the set of morphisms A → X. A natural transformation η: Hom(A,-) ⇒ F assigns to each object X a function η_X: Hom(A,X) → F(X), subject to naturality squares commuting for every morphism in C. The question is: how much freedom do you have in choosing such an η?

The Yoneda lemma gives a surprising answer: none at all, once you fix one thing. The naturality conditions are so rigid that the entire natural transformation η is completely determined by a single element of F(A) — specifically, by η_A(id_A), the image of the identity morphism on A. Every component η_X is then forced by the functoriality of F and the naturality condition. This means the (potentially enormous) collection of all natural transformations Nat(Hom(A,-), F) is in bijection with the set F(A), which may be much smaller and more concrete.

The bijection goes both ways. Given any element s ∈ F(A), you construct a natural transformation η^s by defining η^s_X(f) = F(f)(s) for any morphism f: A → X. Naturality is automatic from the functoriality of F. The proof that these two constructions are inverse to each other is a short diagram chase that repays close reading — following id_A through both directions reveals why the whole argument works.

The Yoneda embedding takes this further. The assignment A ↦ Hom(A,-) defines a functor from C^op into the functor category [C, Set]. The Yoneda lemma implies this functor is fully faithful: morphisms between objects in C correspond exactly to natural transformations between their representable functors. This means C embeds into [C^op, Set] without losing any structural information — every object is completely determined by how morphisms map out of it into every other object. The slogan is: an object is its relationships.

This principle has far-reaching consequences. Representability becomes a way to define objects by universal properties rather than by internal construction. Limits, colimits, and adjunctions can all be characterized Yoneda-style. In more advanced contexts, the Yoneda lemma underlies the theory of sheaves, stacks, and the entire approach of algebraic geometry via functor-of-points. It is not an isolated lemma — it is the lens through which category theory sees everything.