In 1976, William Unruh showed that an observer accelerating through what an inertial observer calls empty vacuum would detect a thermal bath of particles. The Unruh effect, now a standard consequence of quantum field theory, implies that the number of particles in a region is not an observer-independent fact. Physics has thus complicated one of the most elementary questions an ontology can ask: how many things are there?
Mereology, the formal theory of parts and wholes, faces a parallel complication from the opposite direction. Even if we grant a fixed inventory of fundamental constituents, it remains unclear how many objects those constituents make up. Do the particles arranged in your chair compose one further thing, the chair? Do your left shoe and the rings of Saturn compose an object as well? Common sense answers yes to the first and no to the second. Stating the principle behind that verdict has proven remarkably difficult, and the difficulty is not a lapse of philosophical imagination.
Peter van Inwagen called this the Special Composition Question. The leading answers diverge sharply: everything composes, nothing composes, or only some things do. This article traces the logic of these positions and the argument from vagueness that many take to settle the matter. The force of that argument turns out to depend on what we believe about language, about existence, and about whether the world has a privileged structure that our quantifiers can track.
The Special Composition Question
In Material Beings (1990), van Inwagen separated two questions that had been running together. The General Composition Question asks what composition is. The Special Composition Question asks under what conditions it occurs: when is it true that there is some object the xs compose? Composition is standardly defined so that the xs compose y just in case each of the xs is part of y and every part of y overlaps at least one of the xs.
The obvious answers fail quickly. Contact is one candidate: things compose a whole when they touch. But when two people shake hands, no new object comes into existence. Fastening, cohesion and fusion are other candidates, and van Inwagen showed that each admits clear counterexamples. Scattered objects such as archipelagos, bikinis and solar systems point the other way. We happily count them as single things, though their parts are not in contact at all.
Physics makes spatial criteria worse. At the fundamental level, contact is not well defined. Electrons repel long before anything resembling touching occurs, and quantum states are spread across regions rather than bounded by surfaces. Causal criteria fare little better. A hydrogen atom is a bound state, but so is a gravitationally bound galaxy cluster, and binding energy varies continuously. Any physical relation we might pick out comes in degrees, while composition, on its face, does not.
Van Inwagen's own answer illustrates how far the question can push even a philosopher sympathetic to common sense. On his organicism, the xs compose something only if their activity constitutes a life. Organisms exist, and so do mereological simples. Tables, chairs and mountains do not. What exists in their place are simples arranged tablewise, chairwise and mountainwise. Ordinary talk is preserved through paraphrase, but the ordinary ontology is abandoned.
The stakes go beyond the furniture. How we answer the Special Composition Question shapes how we count objects and how we understand persistence. It also bears on whether the entities of chemistry, biology and psychology are genuine additions to the world or convenient redescriptions of physics. The question appears to concern tables. It is really about what our most basic quantifier ranges over, and whether the world settles that independently of us.
TakeawayEvery physical relation we might use to decide when parts form a whole comes in degrees, yet existence seems all-or-nothing. That mismatch is the engine of the whole debate.
Extreme Answers: Universalism and Nihilism
Universalism holds that any things whatsoever compose something. This principle of unrestricted composition is built into classical extensional mereology, which runs from Leśniewski through Leonard and Goodman. David Lewis embraced it openly, including the trout-turkey: the front half of a trout fused with the back half of a turkey. Lewis argued that such objects are ontologically innocent. If composition is a kind of identity, a whole is nothing over and above its parts, so admitting it costs nothing.
The motivations are theoretical. Universalism draws no arbitrary lines, offers a simple and elegant algebra, and fits well with four-dimensionalism. Theodore Sider uses unrestricted composition to sum temporal parts into persisting objects. On this view, ordinary objects are a small, interest-relative subset of the objects that exist. The restriction lives in our quantifiers, not in the world. The cost is a vast population of gerrymandered entities that common sense never asked for.
Nihilism holds that composition never occurs. Only simples exist. Peter Unger, Cian Dorr and, in later work, Sider have defended versions of it, relying on paraphrases like 'there are particles arranged tablewise.' Trenton Merricks offers a causal argument. If a baseball shatters a window, the atoms arranged baseballwise already suffice for the shattering, so the baseball is causally redundant. Merricks makes an exception for conscious persons, which ties the debate directly to the problem of mental causation.
Physics poses a distinctive threat to nihilism. If matter is gunky, meaning every part has further proper parts, then there are no simples, and nihilism implies that nothing exists. Quantum field theory, which treats fields rather than point particles as basic, does not obviously supply simples either. Jonathan Schaffer turns entanglement into an argument for priority monism, on which the whole cosmos is fundamental and its parts are derivative. That reverses nihilism's direction of explanation entirely.
The two extremes share a structural commitment. Both deny that whether composition occurs depends on how things are arranged physically. Neither lets mereology answer to contingent facts about bonding, contact or function. That independence is what makes them theoretically clean, and it is also why they conflict with common sense. One gives us far more objects than we recognize, the other far fewer. Ordinary ontology sits between them, apparently without principled support.
TakeawayUniversalism and nihilism look like opposites, but both refuse to let physical arrangement decide what exists. The real dispute is whether composition is a substantive fact about the world at all.
The Vagueness Argument
Lewis's argument, stated in On the Plurality of Worlds (1986) and sharpened by Sider, starts from a continuous series of cases. At one end, particles are scattered across galaxies. At the other, they make up a clear example of a single object. Adjacent cases differ only minutely. If composition is restricted, it occurs at one end of the series and not at the other. So either there is a sharp cutoff somewhere along the series, or composition is vague in the cases between.
A sharp cutoff looks intolerably arbitrary, since no physical difference between adjacent cases could ground it. So, if restricted, composition must be vague. Lewis then argues that vagueness is semantic indecision: we have not settled which of many precise meanings a word picks out. A sentence such as 'there are exactly n concrete objects' can be written using only quantifiers, identity and logical connectives. These admit no alternative precisifications. Counting cannot be vague, so composition cannot be restricted.
Restricted compositionalists have several replies. Van Inwagen accepts ontic vagueness: it can be indeterminate whether something exists. That carries heavy costs, including Gareth Evans's argument against vague identity and real difficulty in making sense of partial existence. Ned Markosian instead accepts brutal composition, holding that there are sharp facts about composition that follow no statable principle. The cutoff exists. We simply have no rule for locating it.
A subtler reply targets the quantifier itself. Eli Hirsch's quantifier variance holds that there are several equally good candidate meanings for 'exists'. If so, the existential quantifier can itself be vague, and the counting sentence loses its immunity. Sider answers with ontological realism: one quantifier meaning carves reality at its joints and is therefore privileged. The vagueness argument, it turns out, rests on a substantive metaontology that its opponents are free to reject.
Physics bears on this indirectly. The Unruh effect does not make the quantifier vague, but it does show that particle counts, the most obvious candidates for fundamental counting facts, are frame-relative. Phase transitions are perfectly sharp only in the infinite idealization of the thermodynamic limit. Our best science suggests that sharp boundaries in nature are often artifacts of description. Whether existence is an exception is exactly what the vagueness argument assumes rather than establishes.
TakeawayThe vagueness argument does not settle composition on its own; it converts the question into one about whether existence has privileged, joint-carving structure. Accept that, and universalism follows; deny it, and restriction becomes viable again.
The Special Composition Question looks like a puzzle about tables and trout-turkeys. It is better understood as a test of how far ontology can be read off physical structure. Every candidate criterion grounded in physics, whether contact, binding or causal integration, comes in degrees. Composition, if it is a genuine fact, apparently does not.
Universalism and nihilism avoid the tension by denying that composition depends on physical arrangement. Restricted views keep common sense, but only by accepting brute cutoffs, vague existence or a deflationary account of the quantifier. Contemporary physics, which treats fields rather than particles as basic and makes particle counts frame-relative, puts pressure on the idea of a privileged inventory of fundamental things.
Science may not answer the composition question directly. What it can do is change what the question presupposes. If counting is perspectival even at the fundamental level, then asking when parts make wholes becomes, in part, a question about what kind of structure the world has.