Composition as Dependence: When Parts Make Wholes

1. The composition question

When do some things compose a thing? This is the central question of mereology — the theory of parts and wholes. The answer might seem obvious: things compose a whole when they are "stuck together" or "belong together." But neither spatial proximity nor causal interaction suffices. Your hand and a coffee cup you hold are touching and causally interacting, yet they do not compose a single thing. The particles of a human body are constantly exchanged with the environment, yet the body persists. So what makes the difference?

The standard answers are:

- Universalism: any collection of objects composes a whole, no matter how gerrymandered. Your left shoe and the planet Jupiter compose a scattered object. - Nihilism: composition never occurs. There are only fundamental, partless simples. "Tables" and "bodies" are convenient fictions. - Brutalism: there is a fact about when composition occurs, but it is brute and unexplainable.

Each answer fails in its own way. Universalism makes "object" so cheap that it says nothing — it cannot distinguish genuine wholes from arbitrary collections. Nihilism denies the most obvious facts of experience: that bodies, minds, and organisms are real. Brutalism gives up on the question rather than answering it.

The alternative defended here is a dependency criterion: some things compose a whole when and only when they are not separable — when fully characterizing any one of them requires reference to the others. This is a restricted but principled account: composition is neither universal nor absent, and it is grounded in a specific, characterizable relation.

2. Separability

The key notion. Two items A and B are separable if and only if each can be fully characterized without reference to the other. That is: a complete account of what A is, what it does, and why it has the properties it has can be given without mentioning B — and likewise for B.

When A and B are separable, they are genuinely independent. Combining them in space or coupling them causally does not produce anything new — only two things side by side.

When A and B are inseparable — when fully characterizing A requires reference to B, or vice versa, or both — they compose a genuine whole. The whole is not A plus B placed adjacent; it is A-and-B-as-mutually-constituting.

Two clarifications.

Separability is not spatial proximity. Two objects can touch yet be separable: your hand and a cup you grip. Two objects can be distant yet inseparable: the two ends of a single rope, or the numerator and denominator of a fraction. The dependence in question is constitutive, not spatial.

Separability is not causal interaction. Your hand and the cup interact causally, but neither depends on the other for what it is. By contrast, the left and right ventricles of a heart depend on each other for what they are — each ventricle's function is defined relative to the other and to the heart as a whole.

3. What "full characterization" requires

The criterion relies on what counts as a full characterization of a thing. This might seem to reintroduce observer-dependence: full relative to what interest? But the characterization at issue is not whatever an observer happens to find convenient. It is characterization relative to the thing's nature — what it is and what makes it the kind of thing it is.

Consider a heart. Its nature is to pump blood as part of a circulatory system. Fully characterizing a heart — explaining why it has four chambers, why it contracts rhythmically, why it has the structure it does — requires reference to the circulatory system. The heart is inseparable from the system: remove the reference to circulation and you have not described a heart but merely a piece of muscle.

Now consider a brick in a wall. Its nature — what makes it a brick — is its material composition, shape, and strength. These can be characterized fully without reference to the wall. The brick is separable: it is a brick whether or not it sits in this wall.

This reveals an important asymmetry. Dependence can be one-sided: the wall depends on the bricks (a wall's nature is to be made of them), but the bricks do not depend on the wall (a brick is a brick regardless). One-sided dependence produces a real but asymmetric whole: the wall is a genuine thing, but its parts are more fundamental than it is. Mutual dependence — where the parts depend on each other — produces the strongest wholes, those whose parts are what they are only within the whole.

4. Why this is not mereology of stuff

The dependency criterion is about the structure of characterization, not about physical material. This means it applies equally to:

- Physical objects: a body is a whole when its organs are inseparable — when the heart, lungs, and liver are what they are only in relation to each other. - Mental states: a unified experience is a whole when its components (the redness, the roundness, the motion of a perceived ball) are inseparable — when describing any one fully requires reference to the others. This is the account of phenomenal unity defended in the companion article on consciousness. - Abstract structures: a mathematical group is a whole whose elements are inseparable — each element's identity is constituted by its relation to the group operation and to the other elements. - Social entities: a team is a genuine whole when its members' roles are inseparable — when describing what each member does requires reference to the others.

In each case, the same criterion applies: a whole exists when the parts cannot be fully characterized apart from each other. Nothing about material, substrate, or spatial location enters the criterion.

5. Classical puzzles dissolved

The dependency criterion resolves several perennial puzzles in mereology by relocating identity from material to dependence.

The Ship of Theseus. Plank by plank, every part of a ship is replaced. Is the result the same ship? The puzzle assumes that the ship's identity is determined by its material. But the ship is a genuine whole only if its parts are inseparable — and what makes them inseparable is their joint participation in the ship's form, function, and causal continuity. If these are preserved through replacement, the dependencies are preserved, and the ship persists. If they are disrupted — if the replacements break the functional interdependencies — the ship ceases. The answer depends on whether the replacements preserve the dependence structure, not on whether the wood is the same wood.

The problem of the many. A cloud has many slightly different sets of particles that could constitute it. Which set is the cloud? The dependency criterion says: if the particles are genuinely inseparable — if the cloud's dynamics (condensation, evaporation, internal circulation) bind them — then those particles compose the cloud, and the boundary is drawn by the boundary of the dependence. If no set of particles is genuinely inseparable, there is no cloud — only a convenient label for a loosely correlated collection.

Scattered and gerrymandered objects. Is the set of all left shoes a genuine whole? No: each left shoe is fully characterizable without reference to any other. The shoes are separable, and their collection is a mere aggregate. Is a functioning organism whose parts are spatially separated (a distributed nervous system, a colonial organism) a genuine whole? Potentially yes: if the parts are inseparable — if each part's function is defined by its relation to the others — then spatial distance does not undermine composition. The criterion is structural, not spatial.

Tib and the statue. A statue made of clay: is the statue identical to the clay? They differ in persistence conditions (the statue is destroyed by squashing; the clay is not), which suggests they are not the same thing. The dependency criterion says: the statue is a genuine whole to the extent that its parts are inseparable in their statue-role — the arrangement, the form, the aesthetic and functional properties. The clay is a genuine whole to the extent that its molecules are inseparable in their clay-role. These are different wholes with different dependence structures, sharing some parts. There is no deep puzzle, only two different dependence relations over overlapping material.

6. The self-determining structure as maximal whole

The dependency criterion connects directly to the metaphysics of self-determination.

The companion article on foundations argues that reality is the unique self-determining structure: one in which every feature is determined by every other, with no arbitrary parameters, no brute facts, no free constants. A self-determining structure has no separable parts — no feature can be fully characterized without reference to the whole, because every feature is what it is only in virtue of its role in the total structure.

This means the self-determining structure is the maximal whole: the limit case of the dependency criterion, in which nothing is separable from anything else. It is not a collection of independent things that happen to fit together; it is a structure in which the fitting-together constitutes every part.

Does this collapse all distinction? No. The self-determining structure is one whole, but it contains subregions with varying degrees of internal dependence. Within the total structure, some subsystems are tight wholes — their parts are deeply inseparable. Others are looser — their parts are separable at some levels of characterization. The dependency criterion applies recursively: within any whole, its parts may themselves be wholes, depending on whether their sub-parts are inseparable.

This gives a principled account of levels of organization. A molecule is a whole (its atoms are inseparable), which is part of a cell (whose molecules are inseparable relative to cellular function), which is part of an organism (whose cells are inseparable relative to organismal function). At each level, the same criterion applies: is there genuine dependence among the parts, or only juxtaposition?

7. Objections and replies

Circularity objection. The criterion defines composition in terms of dependence between parts. But dependence is a relation between things that are already parts. So the criterion presupposes what it is trying to explain.

Reply. Dependence is not a relation between pre-existing parts. It is a relation between aspects or features of a region that constitutes those aspects as parts. The whole and its parts arise together from the dependence structure. There is no temporal or logical priority: the parts are parts because they depend on each other, and they depend on each other because they are parts of one whole. This is not circularity but constitution — the same phenomenon seen from two angles.

Vagueness objection. The criterion allows for degrees of dependence, which means composition itself admits of degrees. But composition is supposed to be all-or-nothing: either the parts compose a whole or they do not.

Reply. The demand for sharpness everywhere is a theoretical prejudice, not a datum. Many real cases of composition are genuinely borderline — vagueness in composition is a feature of reality, not a defect of the theory. A criterion that forces sharp boundaries in every case would misrepresent these cases. The dependency criterion is honest about where the boundaries are sharp (a functioning heart vs. a non-functioning one) and where they are not (a dying organism, a dissolving team).

Objection from physics. At the fundamental level, everything interacts with everything (gravity, at least). Does this mean the whole universe is a single inseparable whole by the criterion?

Reply. Interaction is not the same as constitutive dependence. Two particles can interact gravitationally without either being characterizable only in terms of the other — gravity is a relation they bear, not a feature that constitutes their identity. The dependency criterion requires more than causal coupling: it requires that the part's nature — what it is — depends on its relation to the other. Fundamental particles, on most physical theories, do not have this property: each is fully characterized by its intrinsic properties and state, independent of distant particles.

8. Conclusion

A whole is not a collection of things placed together. It is a collection of things that depend on each other for what they are. The dependency criterion says: composition occurs when and only when the parts are inseparable — when characterizing any part fully requires reference to the others.

This criterion is principled without being universal. It dissolves classical puzzles by relocating identity from material to structure. It connects to the metaphysics of self-determination by identifying the self-determining structure as the maximal whole in which nothing is separable from anything else. It applies uniformly across domains — physical, mental, abstract, social — because it is about the structure of characterization, not about any particular substrate.

And it provides the foundation for the companion account of consciousness: a unified experience is what it is because its components are inseparable — because describing the redness of the ball requires reference to its roundness and motion, and vice versa. Consciousness is a special case of composition, and composition is dependence.