Yoneda's Big Idea.
Yoneda’s Lemma is a famous result from category theory. The main idea can be said in plain language:
You can understand what something is by understanding how it relates to everything else.
That’s it. That’s the heartbeat of Yoneda.
More precisely: An object is determined by all the ways other things can connect to it, and by how those connections behave when you chain them together.
This isn’t just “how we learn when we can’t see inside.” It’s deeper than that. Even when you can see inside, what you learn is still a pattern of relationships: parts connected to parts, causes linked to effects, actions linked to results. “Inside” is just more structure, which means more relations.
This essay is a self-contained, jargon-light tour of that idea. We’ll build intuition with everyday examples first, then show why this “relational” way of thinking shows up in modern physics, especially in the view that spacetime and gravity might emerge from deeper patterns of information (“the universe as a information”).
You don’t need any background in category theory or physics. Just curiosity.
- The sealed black box: knowing without opening
Imagine I hand you a sealed box. You can’t open it. You can’t see what’s inside. But you can interact with it:
- Connect it to another device and see what signals pass through.
- Press a button and hear a beep.
- Tilt it and feel something roll.
- Shine a light on it and see it absorb or reflect.
- Connect it to another device and see what signals pass through.
After enough tests, you might be able to say, “This behaves like a stopwatch,” or “This behaves like a radio receiver,” even if you never see the internal parts.
The key is this: the identity of the box becomes recognizable from the pattern of interactions you can have with it.
Yoneda takes this intuition and makes it precise in mathematics: an object is determined by the total pattern of ways it can be related (mapped) to and from other objects. If two things have the same relational signature, same “menu of interactions,” same composition rules—then, in a deep sense, they are the same kind of thing.
- The city analogy: a place defined by routes, not coordinates
Suppose you wake up in a new city with no map. Over time, you learn it by movement:
- how long it takes to get from A to B,
- which neighborhoods connect,
- which routes are blocked,
- which paths are reliable,
- how traffic flows.
You might not know the precise coordinates of everything, but you can still know the city well.
Your “mental map” is basically a network of relationships (travel times, accessibility, connections). In many cases that network is more useful than raw coordinates.
Again: structure from relationships.
- The “person in a community” example: identity by roles and interactions
How do you know who someone is in a community (teacher, organizer, mentor, troublemaker, peacemaker)?
Not by their atoms. By:
- how they interact with others,
- what roles they reliably play,
- how they respond in situations,
- what patterns show up across many contexts.
This doesn’t mean a person is “nothing but” relationships—but it shows how much of identity, in practice, is captured by relational patterns.
Yoneda says: if you had the complete relational pattern (and how it composes), you would have the structural identity.
- The “API” example: what a tool is = what it does in every context
Think of a calculator app. You don’t need to read its code to know what it is. You know it by:
- what inputs it accepts,
- what outputs it gives,
- how it behaves for every possible input,
- how it behaves when used inside larger workflows (“use calculator → then graph → then compare results”).
If two tools behave the same in every situation and inside every workflow, they are essentially the same tool.
Yoneda is the mathematical version of this: an object is known by its full “behavior in context,” including how it behaves when combined with other behaviors.
- What Yoneda is really saying
If we had to say Yoneda’s Lemma in plain language:
To understand what X is, learn everything about the ways other things can connect to X, and learn the rules for combining those connections.
Why “combining” matters:
Because relationships aren’t isolated. They compose.
If A connects to B, and B connects to C, then A connects to C through B. That “through” connection is not a new mysterious thing, it’s the result of chaining two connections.
A huge amount of structure in the world comes from composition rules like these:
- Routes compose into longer routes.
- Functions compose into pipelines.
- Causes compose into chains of influence.
- Tools compose into workflows.
- Words compose into sentences.
- Moves compose into strategies.
Yoneda’s message is: the identity of X is captured by the whole pattern of these composable connections
- The physics example: electrons and “things” defined by what they do
In physics, we often don’t identify something by peeking inside it. We identify it by stable rules:
- how it responds to forces,
- how it appears in experiments,
- what symmetries it obeys,
- what happens when you combine systems.
An electron is recognized by a consistent package: charge, mass, spin, and interaction rules. Even if the deeper theory changes language (particles ↔ fields ↔ excitations), the “electron-identity” is maintained by the role it plays in the whole network of physical relationships.
That’s a Yoneda-style idea: identity is the stable pattern of relations across all contexts.
- "The universe-as-Information" example: spacetime as a relational reconstruction
Now let’s move to a modern idea from quantum gravity that’s been gaining clarity over the last couple of decades.
The idea (in gentle form) is:
Spacetime might not be the deepest layer of reality. It might be an emergent, stable summary of deeper quantum information and relationships.
One of the strongest hints comes from “holography” (a well-studied setting called AdS/CFT). The surprising claim is that the physics of a higher-dimensional region with gravity can be fully captured by a lower-dimensional description without gravity, living on its boundary.
That sounds impossible if you imagine the boundary as a skin holding little chunks of the interior. The modern view is more Yoneda-like:
- The “inside” is not something you need to access directly.
- The inside is an object whose identity is fixed by how it relates to the boundary description—by how bulk questions translate into boundary questions, and how answers compose.
Even more striking: the bulk information is redundantly encoded in the boundary in a way that resembles error correction. This means the bulk’s identity is stable even when you only have partial access to the boundary.
So the interior is “real” in this specific sense:
it is the stable, consistently reconstructable structure that appears across many different relational perspectives.
That is Yoneda’s lesson wearing a physics costume.
In this picture:
- Geometry (distances/areas) can be tied to patterns of quantum correlation (entanglement).
- Gravity can be seen as the rule that keeps the emergent geometric description consistent as the underlying state changes.
- And “what exists” becomes deeply tied to “what can be consistently related, reconstructed, and composed.”
You don’t need to accept all of this as settled truth about our universe. The important point is that modern physics keeps being pushed toward a relational ontology, because that’s what works when direct access to “the inside” is impossible.
one sentence Yoneda:
Something is what it is because of all the ways it can connect to everything else, and the rules for chaining those connections.
Its “turtles all the way down.” Even when you look deeper, you don’t escape relationships, you discover a richer network of them.