The Shape Beneath the Surface: On Emergence and Real Structure
“Nature is not only more complex than we think, it is more complex than we can think.” — J.B.S. Haldane
When we look at a murmuration of starlings swirling in the sky, we don’t see thousands of birds following local rules, we see a single living form in flight. A wave curling onto the shore is not the story of trillions of molecules; it is simply a wave. A sentence, once understood, no longer feels like a string of sounds, it becomes a thought. These are glimpses of emergence: the appearance of structure, pattern, or behavior that arises from the interaction of simpler parts.
Meanwhile, at the prison…
Inmate: “It isn’t fair that I’m locked up! I just read Sapolsky (2023). The universe is deterministic. I couldn’t help but commit the crime!”
Guard: “Fair enough. But if the universe is deterministic, then I can’t help but keep you locked up.” (shrugs)
This exchange may sound glib, but it points to a deeper issue. The inmate is appealing to physics, the guard to society. Both descriptions are valid, but they belong to different levels of explanation. At the most fundamental level, physics describes evolving states through mathematical laws. No particles ever say “must” or “because.” Causation itself is an emergent notion, a way we make sense of regularities at scales where prediction and intervention matter. At the social level, our causal vocabulary shifts again, and we talk about choice, intention, and responsibility.
The paradox arises only when one vocabulary is mistaken for the other. Physical determinism should not be taken to erase human responsibility, and moral judgments should not be projected downward onto atoms and synapses. Recognizing that each level carries its own indispensable concepts is the key to understanding emergence. Higher-level descriptions are not illusions layered on physics, but real structures that organize how the world works at their scale.
With this in mind, we can now turn to some varieties of emergence. What they all share is this: every emergent level brings its own vocabulary, a language suited to the entities and relationships that are real and useful at that scale.
Two Kinds of Emergence
Not all emergence runs equally deep. Some emergent patterns, what I will call descriptive emergence, simply give us a more efficient way to describe a system without introducing fundamentally new kinds of complexity. A center of gravity makes a mechanical system easier to reason about, but it is just a mathematical shorthand. In fluid dynamics, we average over the chaotic motions of particles and speak instead of pressure, temperature, and flow.
Another example is the principle of least action in physics. Instead of tracking every step-by-step force and acceleration, we can summarize an entire trajectory as the one that minimizes (or in technical terms, extremizes) the quantity called “action.” Think of action as the cost of each part of the trajectory. When you sum up all of these costs, nature will usually have picked the one where that sum is as small (or as stable) as possible.
This principle feels almost teleological: systems behave as if they “know” the end and take the most efficient path. But it is simply another way of describing the same underlying mechanics. Like pressure or center of gravity, it is a higher-level law that emerges as a remarkably stable and predictive shorthand.
These are stable, useful regularities that make prediction and explanation possible, but they do not add new structures or agents to the world. Because these descriptive properties are stable and measurable, we can build entire theories around them. We don’t need to know the momentum of every molecule to predict how a gas exerts pressure on a container, or track every atom in a pendulum’s arm to calculate its swing. Higher-level quantities like pressure or motion are so consistent that they become the natural language for reasoning at that scale.
Other cases run deeper. This is what I will call constructive emergence. Here, new patterns of interaction give rise to new forms of stability: self-contained entities with their own properties and dynamics. A molecule can exist only if atoms can bond. A cell can function only when molecules self-organize into membranes and metabolic cycles. Thought can take hold only when neurons form circuits. In each case, the system crosses a threshold and exhibits properties that arise from the interaction of its parts, properties that cannot be found in the parts alone.
Physicist P. W. Anderson made this point forcefully in his classic 1972 paper More Is Different, arguing that at each new level of complexity, the whole “becomes not only more than but very different from the sum of its parts” (Anderson 1972, 393).
Constructive emergence is how complexity is built in the first place, through successive layers of stability that make further elaboration possible. Deep complexity never appears all at once, It grows in stages. Stability is the scaffolding of complexity.
Waypoints in the Climb
Because complexity builds upon stable layers, we can often discover those layers. They become waypoints, places where we can pause, name, and model what’s happening. These emergent levels are not arbitrary human conveniences; they reflect real, constructed structure built up through the system’s own dynamics.
Each level, whether descriptive or constructive, has its own language, shaped by what is stable and causally relevant at that scale. In physics, we speak of fields, particles, or waves depending on context. In biology, we talk of cells, enzymes, and blood flow, not atoms. In neuroscience, we talk about neurons and circuits. In psychology, we invoke attention, memory, and emotion. In society, we shift again to roles, norms, laws, and institutions.
These vocabularies are not optional shortcuts. They reflect real regularities that persist at their respective scales. Long before the discovery of molecules, scientists could make accurate predictions in fluid dynamics using only pressure and temperature. The vocabulary of thermodynamics was sufficient because those properties are stable at that level. You don’t need to track every molecule to explain how steam drives a piston; the language of pressure and temperature suffices.
Medicine offers another striking example. A physician treating pneumonia works with the language of lungs, airways, inflammation, and immune response. They do not need to invoke the chemistry of oxygen binding to hemoglobin, even though those processes are real. At the clinical level, the vocabulary of airflow and infection suffices. If a doctor insisted on explaining your cough in terms of molecular interactions, they would be not only unhelpful but incomprehensible.
Mixing concepts across levels often produces confusion. Speaking of “selfish genes,” for example, is a useful metaphor at the level of evolutionary biology, but becomes misleading if taken to imply that molecules themselves possess motives (Dawkins 1976/1989). Likewise, trying to explain consciousness in terms of individual neurons overlooks the emergent dynamics of networks. Each level demands its own vocabulary because each level stabilizes its own set of causal patterns. This use of different languages at different scales is not a weakness of science but its strength. It allows us to focus on what is real and manageable at a given scale, without being lost in every microscopic detail below.
The Challenge of Mapping Complexity
One might ask: if everything is made of particles, can’t we just describe the world from the bottom up? In theory, yes. In practice, no. The problem is mapping complexity. This is a term I will introduce here to capture the difficulty of connecting different levels in a tractable way, that is, in a way that can actually be managed or reasoned through.
Some mappings are short and relatively straightforward. For example, eye color is largely determined by a single gene that controls the production of pigment. The relationship is close to one-to-one: the gene changes, and eye color changes in a predictable way. Other traits, however, are vastly more tangled. Height, for instance, is influenced by hundreds of genes, each making a small contribution, and by environmental factors such as nutrition. Intelligence is even more complex, shaped by many genes interacting with one another and with cultural and developmental contexts. Although levels are causally linked, the pathways between them can be so intricate that direct derivation becomes impractical.
This is because many mappings are rarely linear or one-dimensional. A small change at a lower level can yield a large change at a higher level… or none at all. Variables interact in nonlinear ways, creating cascading influences that defy simple cause-and-effect tracing. Explaining traits like height or intelligence from genes alone means confronting a dense web of causal interactions.
Mapping complexity is why we need higher-level concepts in the first place. Just as a city map needs neighborhoods and routes, not the position of every grain of sand, science needs conceptual handholds at each stable level. These abstractions are not arbitrary; they are the only workable way to reason, predict, and act at that scale.
Real Transitions, Not Just Perspectives
It’s tempting to think of molecules, cells, minds, and societies as cognitive conveniences imposed by limited human minds. And in part, they are. But the deeper truth is that we are discovering, not inventing, these levels. They correspond to real transitions in stability and coherence.
We don’t talk about quark clusters in anatomy because their influence has already been absorbed into more stable configurations: atoms, molecules, cells, and so on, up through the layers to organs. At the level of organs and tissues, what matters is not the arrangement of fundamental particles but the integrated function of biological modules.
These stable units: neurons, genes, proteins, economic agents are the load-bearing elements in the structure of complexity. Each marks a plateau of stability from which further complexity can grow.
The Architecture of Complexity
The world is not flat, not in space, and not in structure. It is layered. It builds itself stage by stage, with each layer resting on the stability of the one below. This is what we mean by hierarchy: a structure of emergence, systems built from systems, each level bringing new regularities into being.
At each layer, new patterns stabilize, new entities cohere, and new ways of understanding become necessary. These are more than merely viewpoints; they are discoveries. Our languages change because reality changes. What becomes meaningful at each level is what remains stable, structured, and causally effective.
While every emergent level is grounded in lower-level dynamics, it cannot be reduced without loss. Reduction becomes useless in the face of mapping complexity. Higher-level concepts are indispensable; they are the only way we can effectively interface with reality at that level.
This is the central theme of this book: complexity is built in stages, and each stage requires stability before the next can arise. To understand complex systems is to recognize these waypoints of emergence as real structure. The rest of this book explores how this principle plays out: in physics, biology, cognition, and culture. It begins here, with a simple idea: emergent levels exist in the world, and we can learn to see the shape beneath the surface.