Why Explanation Comes in Layers

Introduction

Why are there layers of explanation? Why is it possible to explain a pot of boiling water in one moment by molecular motion, in another by vapor pressure and atmospheric conditions, and in another by saying that someone wanted ramen?

That question matters because our answer shapes how we think about knowledge itself. If every higher-level explanation is merely a shorthand for a deeper one, then explanation is fundamentally a matter of reduction. We may speak in terms of agents, goals, organisms, markets, or ideas, but only as a temporary convenience. On that view, real understanding always lies below.

I think that view is mistaken.

The mistake is not in thinking that lower-level mechanisms are real. Of course they are. The mistake is in assuming that lower means better explanatorily. That assumption often feels natural because deeper descriptions seem more fundamental. But explanatory depth and ontological depth are not the same thing. A more fundamental description is not automatically a more illuminating one.

My conjecture is that explanatory layers exist because finite minds cannot use reality in its fully uncompressed form. Reality, as encountered by a knower, is too informationally dense and too relationally entangled to be directly intelligible. Explanation therefore requires compression. But compression does not merely shorten a description. It selects structure. It throws away vast amounts of information while preserving what matters for understanding. That is what allows higher-level explanatory objects to appear: pots, ants, colonies, goals, markets, reasons, selves.

There is a second conjecture that goes with this one. As we move lower in the explanatory hierarchy, the burden of connectedness does not merely increase. It regresses toward holism. Lower-level descriptions are not merely more detailed. They are more dependent on wider relations, more sensitive to context, and less separable into self-contained units. Going lower demands more of reality, not less, until at the deepest levels the context implicates the whole.

If that is right, then reductionism misses something central. It carries the higher-level practice of dividing phenomena into neat components down into levels where connectedness matters more than separation. In doing so, it can destroy the very relations required to understand the phenomenon at all.

A Note on Method

Explanation and causation are not opposites, and this essay does not treat them as such. Tracing causes is frequently how we explain things. The essay does not deny this.

The more precise point is that causation operates at multiple levels. Molecules cause phase transitions. Intentions cause actions. Markets cause prices. All of these are genuine causal claims. The question the essay is asking is not whether causation is real at each level. It is which level of causal description makes a given phenomenon intelligible.

Reductionism assumes the answer is always the substrate level. This essay argues that it is not. The most illuminating causal description is frequently at a higher level, and the hierarchy of explanation exists precisely because finite minds must select the level at which to describe the causes that matter.

One term needs flagging before proceeding. “Level of emergence” is used here in Deutsch’s sense: an explanatory domain where phenomena explain one another without requiring reduction to constituent parts. This differs from the standard complexity-theory usage, which refers specifically to bottom-up causal emergence. The two are independent concepts. Conflating them is one of the easier ways to misread what follows.

The Temptation of Reductionism

Reductionism, as I am using the term, is not simply the claim that things have parts. Of course they do. Nor is it the claim that lower-level processes are real. They are. Reductionism is the epistemological mistake of assuming that the best explanation must come from analyzing things into components, and that higher-level explanations are therefore less fundamental or less real.

That view has intuitive appeal. If a thing is made of parts, then surely understanding the parts should explain the whole. If water is made of molecules, then molecular description seems closer to the truth than talking about boiling pots or people making dinner. If an ant colony is made of individual ants, then it seems that the real explanation must be the behavior of the ants themselves. The higher-level description can appear secondary, as though it were only a verbal summary of what is really happening underneath.

But this view quietly assumes something that is not obviously true. It assumes that breaking a thing into components reveals the explanatory structure relevant to the phenomenon. That is precisely what needs to be argued, not assumed.

A thing can be composed of lower-level parts without being best explained at the level of those parts. In fact, once we care about understanding rather than mere inventory, the opposite is often true. The relevant structure may exist in patterns among the parts, in relations across time, in boundary conditions, in collective behavior, in history, or in the interaction between a system and its environment. In such cases, decomposition may preserve causal ingredients while destroying explanatory form.

That is the pressure point. Reductionism often mistakes explanatory decomposition for the structure of reality itself.

Compression and the Problem of Finite Minds

Compression is not mere abbreviation. It is the act of representing a phenomenon with less information while preserving what matters for explanation.

This matters because finite minds cannot reason over reality in fully uncompressed detail.

A complete microphysical description of even a mundane event would be hopelessly vast. But the problem is not only size. The deeper problem is that a microphysical description is not self-interpreting. It does not arrive with its important abstractions already marked. It does not tell us which cluster of particles is a pot, which pattern is a person, which temporal sequence counts as cooking, or which behavior expresses a goal. Those are not written on the microstate. They are abstractions over it.

That means knowledge does not arise simply by accumulating more detail. A complete inventory is not yet an explanation. It becomes explanatory only when structure is selected, compressed, and organized into a form a finite mind can use.

This is why layers of explanation are not optional conveniences. They are what explanation looks like for finite knowers in a connected world. We do not first possess reality in its pure informational fullness and then choose whether to simplify it. We begin in the opposite condition. Reality must be made intelligible through abstraction. Compression is not a stylistic aid added after understanding. It is part of what understanding is.

This also explains why higher-level explanatory objects are not unreal. They are not hallucinations produced by laziness. They are the result of successful compression. A higher-level object is real in the explanatory sense when it preserves stable structure better than an ocean of lower-level detail does for the task at hand.

Connectedness and the Cost of Going Lower

Compression alone does not complete the picture. Another idea is needed: connectedness.

At higher levels of explanation, things appear fairly separable. The pot, the stove, the ant, the colony, the person, the kitchen. That separability is useful and often profoundly explanatory. But it is an achievement of abstraction, not a primitive feature of reality.

As we move lower in the explanatory hierarchy, the burden of connectedness does not merely increase. It regresses toward holism. To describe a molecule in boiling water, you must specify temperature and pressure, which are collective properties of millions of molecules. You must specify container geometry, atmospheric pressure, the external heat source, the gravitational field. Go lower still, to quantum mechanics, and the situation deepens. Particles are entangled across arbitrary distances. The wave function of a system is global, not local. You cannot describe a particle in isolation without reference to relations that may span large regions of space. In general relativity, local spacetime geometry is determined by the distribution of mass and energy across the entire universe. Every boundary condition expands. Every attempt to isolate a local description pulls in more of the world.

Following this descent to its conclusion, you do not arrive at simple, self-sufficient units. You arrive at a description in which nothing can be specified without specifying more, until the context implicates the whole of reality. The lower limit of reductionism is not simplicity. It is total interconnection.

Lower-level description does not only add detail. It restores dependence. It reintroduces the relations that higher-level abstraction had screened off. And at the deepest levels, those relations are not merely numerous. They are non-local, historical, and in some cases unbounded.

This is why higher-level explanation is not merely convenient. Each explanatory layer draws a boundary around what is relevant and treats everything outside that boundary as fixed background. That severing of connections, managed disconnection, is what makes tractable explanation possible at all. Compression is not just shortening a description. It is actively quarantining the vast web of relations that would otherwise make every local event dependent on the state of the universe.

Why Higher-Level Explanations Are Real

At this point a natural objection arises. If higher-level explanations are products of how minds must engage with reality, does that not make them less real, artifacts of cognition rather than features of the world?

No. But the objection points toward a clarification worth making explicitly.

Layers of explanation are required by minds. They are not features of reality that would exist whether or not any mind encountered it. A hypothetical intellect without the limitations of finite minds, with unlimited processing, unlimited memory, and the ability to hold all relations simultaneously, would not need layers. It would encounter reality as one connected whole.

But minds are real features of reality. Humans are minds. The structures that minds must impose to make reality intelligible are therefore real features of how reality is encountered. Not fundamental to the substrate, but not invented or arbitrary either. This is a more modest claim than saying higher-level objects exist independently of all minds. It is also the more defensible one, and it follows directly from the essay’s own starting premises.

What makes a higher-level explanation real in the relevant sense is that it has explanatory reach: it illuminates multiple related phenomena, answers why-questions, preserves counterfactual structure, and does so without requiring the full lower-level substrate to be reconstructed each time. That reach is not illusory. It is a genuine property of the explanation, grounded in genuine structure.

This is why “Buddy wanted ramen” is a valid explanation of boiling water. It is not a denial of physics. It is an explanation at a level where agency, preference, and intention organize the event more economically and more informatively than a particle-by-particle history would. To reject it because it is not microphysical is not rigor. It is a confusion about what explanation is for.

A colony is not unreal because it is made of ants. A market is not unreal because it is made of transactions. A self is not unreal because it is implemented in matter. Higher-level explanations are real when they carve reality at joints that matter for understanding, and minds are real enough to make that mattering real.

From Boiling Water to Wanting Ramen

Consider again the familiar example of boiling water.

One explanation says that water boils because molecules have enough energy to enter the gas phase. Another says that water boils when the vapor pressure of the liquid equals the surrounding atmospheric pressure. A third says that Buddy wanted ramen and turned on the stove.

These are not rival explanations in the sense of mutually exclusive competitors. They are explanations at different levels of emergence.

A level of emergence, in the Deutschian sense, is a set of phenomena that can be explained well in terms of each other without being analyzed into their constituent entities. That matters because it identifies a real explanatory domain. Vapor pressure, boiling point, and atmospheric pressure belong together in a way that supports explanation without forcing us down to atomic detail. Human wanting, planning, cooking, and eating belong together in another such domain. Neither is made illusory by the existence of molecules.

What would it take to derive “Buddy wanted ramen” from a pure microphysical description? Far more than a snapshot of matter. One would need a large temporal window, stable identification of entities across time, abstractions for persons, beliefs, intentions, and artifacts, and a way of distinguishing relevant structure from meaningless detail. In other words, one would need a great deal of the world and its history already organized conceptually.

That is why reductionist triumphalism here is misplaced. The lower-level account is not automatically the superior explanation simply because it sits closer to the substrate. The higher-level account may preserve the structure that matters far better. It compresses what is explanatorily central while ignoring oceans of irrelevant microvariation.

This is not a failure of rigor. It is what rigor looks like when explanation is the goal.

The Reductionist Mistake

We can now state the reductionist mistake more clearly.

Reductionism assumes that going lower means breaking a phenomenon into components and explaining the whole from them. But this program does not terminate where reductionism expects. It does not arrive at simple, separable units that explain everything above them. As the connectedness argument shows, it arrives at holism, a description of total interconnection where nothing can be specified without specifying more. The reductionist pursues simplicity and finds its opposite.

This is the deepest problem. Reductionism is not merely incomplete. It is self-undermining. It promises a foundation and delivers an infinite regress. The practice of decomposition works well at higher levels precisely because abstraction has already screened off most connectedness. We can isolate a pot from a kitchen, an ant from a colony, a buyer from a market, because managed disconnection has already made those isolations tractable. The success of decomposition at higher levels is not evidence that reality is fundamentally composed of separable units. It is evidence that higher-level abstraction has done its job.

Carrying decomposition downward, into domains where connectedness has not been screened off, destroys the structure required to make the event intelligible. This is the epistemological error: treating decomposition as a universal method even in domains where it strips away the very relations the description depends on.

Ant colonies make this vivid. Individual ants follow local behavioral rules, many mediated by pheromones. But the colony exhibits food-finding, bridge-building, collective defense, and adaptive organization that no individual ant plans or understands. The colony-level pattern has explanatory autonomy. Insisting that only the ant is real in the explanatory sense, and the colony merely derivative, means missing the phenomenon entirely.

The same mistake occurs in discussions of mind. Predictive processing may describe something real about lower-level cognitive architecture. But it does not follow that creative explanation, abstraction, criticism, or agency are reducible to it in the explanatory sense. A lower-level mechanism can be real without supplying the best explanation at a higher level.

The broader lesson: components matter, mechanisms matter, but explanation is not a contest to find the smallest pieces. It is the search for the right managed disconnection in a deeply connected reality.

Conclusion

Explanations come in layers because finite minds cannot reason over uncompressed reality directly, and because deeper descriptions do not lead to simpler, more separable units. They lead toward holism. A complete lower-level inventory is not yet an explanation. It becomes explanatory only through compression, managed disconnection, and the selective preservation of structure that matters.

This is why higher-level explanations are not merely conveniences. They are the only forms in which reality becomes intelligible to finite knowers. They do not float free of lower-level reality, but neither are they dissolved by it. They are real, not because they exist independently of minds, but because the minds that require them are real features of the world.

Reductionism misses this by assuming that breaking things into components is always the path to better understanding. But the deeper we go, the more connectedness must be preserved. Decomposition, carried far enough, erases what matters and arrives at the holism it sought to escape.

If this is right, then the hierarchy of explanation is not an embarrassment to be eliminated. It is a feature of reality as encountered by finite knowers. The world is not given to us as explanation-ready. Explanation requires compression and managed disconnection. Together they create layers. And those layers are not a retreat from truth. They are among the forms in which truth becomes knowable.

Appendix A: Working Definitions

Compression: representing a phenomenon with less information while preserving the structure relevant to explanation.

Connectedness: the extent to which a phenomenon depends for its intelligibility on relations to other entities, conditions, histories, or environments.

Explanatory reach: the ability of an explanation to illuminate many related phenomena, answer why-questions, and preserve relevant counterfactual structure without reconstructing the full substrate each time.

Level of emergence: a set of phenomena that can be explained well in terms of each other without being analyzed into their constituent entities.

Managed disconnection: the active severing of relations that are not relevant to a phenomenon, treating everything outside the explanatory boundary as fixed background. The mechanism by which higher-level layers make tractable explanation possible in a deeply connected world.

Reductionism: the epistemological mistake of assuming that explanations must always come from analyzing things into components, and that lower-level explanations are therefore more fundamental or more real. Reductionism is self-undermining: followed to its conclusion, the program of decomposition does not arrive at simple, isolated units but at total interconnection. Its apparent successes at higher levels are products of the managed disconnection that higher-level abstraction has already performed.

Appendix B: A Short Conjecture

Reality is deeply connected, and that connectedness does not diminish as descriptions go lower. It intensifies, regressing toward holism. Finite minds cannot reason over that full connectedness directly. Explanation therefore depends on compression and managed disconnection: abstractions that sever most relations while preserving those relevant to the phenomenon. Higher-level explanatory layers are real because the minds that require them are real. Reductionism fails not merely because it mistakes decomposition for understanding, but because it is self-undermining: followed honestly, it does not arrive at the simple foundation it promises. It arrives at the holism it sought to escape.