Mereology: Main Article

Mereology: Parts, Wholes, and the Self-Determining Structure

Why This Article Exists

Three downstream problems are blocked until we have an account of how parts compose into wholes within the self-determining structure:

1. Open individualism needs an account of how perspectives (parts) relate to the universal experiencer (whole). Saying "there is one subject experiencing from every perspective" is empty without an explanation of how a perspective can be a proper part of a subject. 2. The binding problem at higher levels. The consciousness article solves the local binding problem—why the red, the round, and the looming of an apple are one percept—by noting that they are jointly bound terms in a single relational complex. But this raises a higher-level question: how do experiences themselves compose into the continuous life of a subject? How do substructures combine into larger substructures? 3. The measure problem. Comparing valence across perspectives requires knowing what a perspective is—what counts as one locus of subjectivity, how it is individuated, and how perspectives relate to each other within the whole.

Each of these is a mereological question. This article provides the framework's answers.

The Standard Options and Why They Fail

Classical mereology (unrestricted composition) says any collection of objects composes a further object. This is too permissive: it lets any arbitrary collection—a nose, the number seven, and a distant star—compose a genuine whole. If everything composes everything, composition does no explanatory work. Worse, for our purposes, it would mean that any arbitrary grouping of structural features counts as a genuine part of the self-determining structure, which undermines the idea that the structure has natural joints.

Mereological nihilism says there are no composite objects—only fundamental simples arranged in various ways. This is too restrictive for our framework: it would deny that experiences, subjects, or perspectives are real entities. But the entire project depends on subjects being real structural features, not convenient fictions.

Restricted composition theories say composition occurs only when some condition is met—continuity, causal connection, functional integration. These are closer to right, but they typically treat the composition condition as a rule imposed from outside the parts. Where does the rule come from? If the rule is stipulated, it is arbitrary; if it is derived, it must be derived from the structure of the whole.

The framework's approach is different: composition is not a relation between independently given parts, governed by an external rule. Composition is a feature of the self-determining structure itself. Parts are not prior to the whole; they are aspects of the whole's internal decomposition.

The Core Thesis

Parts are structural features of the self-determining structure, individuated by the structure's own joints.

The self-determining structure determines what is the case—completely, consistently, and without external input. Among the things it determines is its own internal decomposition: which structural features depend on which, which substructures are integrated, which components are functionally distinct. The genuine parts of the structure are not arbitrary slices we impose by drawing boundaries; they are the boundaries the structure draws on itself.

This is not a mysterious claim. It is the familiar insight that a well-designed program has natural modules—not because someone declared them, but because the computation's own dependencies create distinct regions with high internal coupling and low external coupling. The canonical causal diagram of a system (the invariant partial order of informational dependence, extracted by quotienting out definitional and observational equivalence) is a mereological structure. It tells us what depends on what, what is a component of what, what integrates with what. The joints of this diagram are the joints of reality's decomposition into parts.

Three consequences follow.

First: composition is determined, not stipulated. Which parts compose into which wholes is not a question we answer by convention or by rule. It is answered by the canonical causal structure itself. Two features compose a genuine whole iff they are integrated in the canonical diagram—iff their states are informationally dependent, iff they form a coherent substructure with its own internal dependencies. This is a structural criterion, discoverable by analysis, not imposed by fiat.

Second: parts are ontologically dependent on the whole. A part of the self-determining structure does not exist independently and then get combined with other parts. It exists in virtue of its role in the overall structure. The structure is not built from parts; it is decomposed into parts. The direction of ontological priority runs from whole to parts, not from parts to whole. This does not make parts unreal—a vertex of a triangle is a real feature of the triangle. But it means that the vertex does not exist independently of the triangular structure.

Third: there is exactly one correct decomposition. Since the self-determining structure is unique (by the argument in the metaphysics article), and since composition is determined by the structure's own joints, there is one genuine mereology: the decomposition dictated by the canonical causal structure. Different analyses of the same system may draw boundaries in different places, but the gerrymandered analyses are excluded for the same reason gerrymandered models of consciousness are excluded—they smuggle structure into the interpretation rather than extracting it from the thing itself.

The Canonical Diagram as Mereological Structure

The formal method article's central move—quotienting a typed program by definitional and observational equivalence to extract the canonical causal structure—is, at its heart, a mereological procedure. It takes a system described by an arbitrary program and extracts the invariant structure of informational dependence: what is computed from what, which differences propagate where, which components are functionally integrated.

This procedure yields a partial order: a directed acyclic graph in which nodes represent informational states and edges represent dependence relations. Within this partial order, natural clusters emerge—subgraphs with high internal connectivity and defined external interfaces. These clusters are the genuine parts.

A concrete illustration. Consider a visual system that processes color, shape, and motion. In the canonical causal diagram, these may form three tightly integrated substructures, each with internal dependencies (color processing has its own sub-stages: wavelength discrimination, opponent processing, color constancy) and limited cross-dependencies (shape and color are bound in object recognition; motion and shape interact in tracking). The mereological structure captures exactly this: color-processing is a part of the visual system, which is a part of the whole conscious structure. The boundaries are not arbitrary; they reflect the actual dependency structure of the computation.

This gives us the tool the binding problem solution needs at the higher level. How do experiences compose into the continuous life of a subject? Through the canonical structure's own dependency relations. Each experience is a substructure of the subject's larger canonical structure—connected to previous and future experiences through memory, anticipation, and the continuity of the self-model. The "continuity" of conscious life is not a mysterious thread tying discrete moments together; it is the informational dependence between sequential substructures in the canonical diagram.

Mereology and the Consciousness Article

The consciousness article claims that binding dissolves: features of an experience are not separate items requiring glue; they are jointly bound terms in a single relational complex. The mereology article explains why this is true at a deeper level.

Features appear to need binding only when we first (mistakenly) treat them as independently given atoms and then puzzle over what connects them. But features are not atoms. They are structural aspects of a conscious substructure, and they exist only in virtue of their roles in that substructure. The canonical causal diagram shows them as co-dependent: the "redness" of the apple and its "roundness" are not independent nodes that happen to be near each other in the graph. They are aspects of a single integrated processing stream—a node-cluster whose internal dependencies make it one structure rather than two.

The binding problem, stated in mereological terms, is: how do the parts of an experience compose into a whole? The answer: they compose because they are not independently existing parts at all. They are distinguished aspects of a single structure—the conscious substructure—recognized by analysis but not separable in the canonical diagram without destroying the information dependencies that define them. A triangle's angles do not need glue to stay attached to its sides. The angles and the sides are not independently existing items that happen to coincide. They are aspects of a single geometric structure, each defined by its relation to the others.

Mereology and Open Individualism

Open individualism claims there is one subject—the universal experiencer—experiencing reality from every perspective. The mereological account makes this precise.

A perspective is a substructure of the self-determining structure in which the subjectivity property is fulfilled: a world-model, a self-model, and binding between them. Each such substructure is a genuine part of the whole—individuated by the canonical causal structure's own joints, not by our convenience.

The universal experiencer is the self-determining structure itself, considered in its entirety. It is not a separate entity that somehow "contains" perspectives as a jar contains marbles. It is the whole of which perspectives are structural features—the same way a geometric object is the whole of which vertices, edges, and faces are structural features.

This yields a specific mereological picture:

- Each perspective is a proper part of the universal experiencer. - Perspectives are related to each other through their shared embedding in the same self-determining structure. They are not causally isolated; they are connected through the dependency structure of the whole. - The universal experiencer is not the sum of perspectives (as if it were their mereological fusion). It is the structure that determines perspectives—the one structure from which all perspective-structure is decomposed.

This resolves a puzzle the consciousness article leaves open. If each perspective is a locus of subjectivity, and there is only one subject, how can there be distinct experiences? The answer: distinctness is structural, not numerical. Two perspectives are structurally distinct substructures—they have different world-models, different self-models, different binding relations. But they are numerically aspects of the same whole—just as two vertices of a graph are structurally distinct but numerically parts of the same graph. There is no contradiction in one subject having structurally distinct perspectives, any more than there is a contradiction in one graph having structurally distinct vertices.

Mereology and the Measure Problem

The ethics article raises the question: how do we compare valence across perspectives? The mereological framework constrains the answer.

Valence occurs at structural loci—substructures where the subjectivity property and the evaluative narrative are fulfilled. These loci are genuine parts of the whole, not arbitrary groupings. Comparing valence across two perspectives means comparing structural features of two substructures within the same self-determining whole.

This is not a mysterious leap across an experiential divide. It is a structural comparison within a single structure—comparable to comparing two vertices of a graph by their degree, centrality, or connectivity. The comparison is possible because both perspectives are embedded in the same canonical causal structure, and their structural features are commensurable in virtue of sharing that embedding.

What we do not yet have is a precise metric. The framework tells us what we are comparing (structural features of self-modeling substructures: allocation fractions, depth of self-model binding, breadth of memory integration, intensity of prospective content) and why comparison is possible (shared embedding in one structure). The metric itself—how to weight these features, how to aggregate across them—is a technical problem whose solution requires further work on the algebra of quality spaces.

The measure problem, in short, is not a metaphysical obstacle. It is a computational one. We know what kind of thing we are measuring (structural properties of substructures in a determinate whole). We do not yet know the exact function. This is progress.

Objections

"This makes parts too dependent on wholes." If parts exist only in virtue of the whole, how can we ever analyze a system by studying its parts? Response: we can and do. Analysis proceeds by identifying substructures within the canonical diagram and studying their internal dependencies. But this analysis does not presuppose that the parts are metaphysically independent. A biologist studies the heart as a part of the circulatory system without assuming the heart could exist without a body. The direction of ontological priority (whole to parts) is compatible with the direction of epistemic access (we learn about wholes by studying their parts).

"Isn't this just classical mereology in disguise?" No. Classical mereology says any collection of objects composes a further object. Our view says composition is determined by the canonical causal structure—only substructures with genuine informational integration compose wholes. An arbitrary collection (a nose, the number seven, a distant star) has no integrated dependency structure and composes nothing.

"What about the decomposition problem?" If the self-determining structure determines its own decomposition, why does it seem like we can decompose a system in many different ways? Response: we can describe a system in many ways, but only one decomposition reflects the genuine informational dependencies. The gerrymandered decompositions—those that split a genuinely integrated substructure or merge genuinely independent ones—are excluded by the same canonicalization procedure that resolves the Putnam worry in the consciousness article. Finding the genuine decomposition is difficult; that it exists is guaranteed.

"This does not help with the measure problem in practice." Correct. The mereology tells us what a perspective is (a subjectivity-fulfilling substructure of the self-determining whole) and why comparison is possible (shared embedding in one structure). It does not provide a metric. But it does something more important: it tells us what kind of thing a metric would have to measure, which is a prerequisite for developing one.

What Remains

The mereological framework is now explicit. What remains:

1. A rigorous formal account of how the canonical causal diagram's partial order generates a mereology—specifically, a definition of "integrated substructure" precise enough to be computed, not merely gestured at. 2. An application to open individualism: given the formal mereology, an exact characterization of how perspectives compose into the universal experiencer, and how structurally distinct perspectives can be numerically aspects of one subject. 3. The metric: a principled weighting of structural features for comparing valence across perspectives, grounded in the formal definition of the relevant substructures. 4. Connection to the Ruliad and MUH: if the self-determining structure is the Ruliad (or something like it), the mereological question becomes—what are the natural parts of the Ruliad?

Each of these is a technical problem. None is a mystery. The framework tells us what we are looking for and why it must exist.