Contribution
Morganti names and develops “metaphysical coherentism” as a structure for reality: dependence runs in cycles or holistic networks rather than a well-founded hierarchy. He argues that it gives OSR’s motivating physics (group-theoretic particle kinds, entanglement, AQFT, spacetime structure) an object-based ontology without OSR’s burdens. It does not need to reduce all intrinsic properties to relations, extract concrete tokens from abstract representations (imprimitivity systems, fn. 31), or posit a second category of property-less relata. He claims the view is “arguably new, at least in its detailed formulation” (p. 2).
Key insight
The structuralist lesson of physics is not that relations come first. It is that no object is what it is independently of the others it belongs with. That can be said with objects plus symmetric ontological dependence, which is not a further concrete item, so the inventory can contain only objects (fn. 33). Structure then becomes the pattern of mutual dependence rather than an entity. The standard worry, that symmetric dependence makes explanation circular, is met by the logic of the relation: drop full transitivity, or let only partial dependence be symmetric, and nothing fully depends on or explains itself.
Assumptions
- OSR is well motivated; the question is only what its metaphysics is (§1).
- Quantum theory is the right place to look for a scientifically grounded metaphysics of material reality (§1).
- Only “positive” arguments from physics to relational structure are considered. French’s underdetermination (individuals vs non-individuals) route is set aside, on the grounds that weak discernibility has broken the underdetermination (fn. 1).
- For the sake of argument, symmetry and group considerations are granted as one route to OSR (fn. 5).
- Ontological dependence is tightly connected to metaphysical explanation (Lowe 2003), and is treated as interchangeable with ground for present purposes (fn. 13; the published version adds a caveat).
- Empirical premise: different particle kinds (mass and spin values) correspond to different irreducible representations of the same groups. Which kinds are actual is settled empirically, not group-theoretically (§3.1, pp. 6–7, after McKenzie).
Key results
- §3.1 against strong OSR (French). Ontological dependence presupposes both relata (Wolff 2012) and does not license reduction. Grounding might (McKenzie). Even granting it, some state-independent intrinsic features of actual particle kinds are not structurally reducible.
- §3.2 against moderate OSR (Esfeld & Lam). It posits property-less “bare occupants”, faces Lowe’s asymmetry objection, and faces a collapse dilemma: mutual existential dependence invites elimination of one category, or a single underlying category, which is Esfeld & Lam 2011’s own drift toward “a somewhat old-fashioned substance ontology” (p. 8). The irreducible intrinsic properties sink it anyway.
- §3.3 against contextualist OSR (Ladyman 2007, Stachel). Both premises of the permutation-invariance argument fail. Primitive intra-world identity does not entail haecceitism (counterpart theory, Leibnizian super-essentialism), and permutation invariance does not refute haecceitism, because haecceitistic differences need have no empirical consequences (p. 9). The view also concerns identity only, which is too little for a structuralist.
- §4.1 coherence of coherentism. Irreflexivity is compatible with cycles given antitransitivity or intransitivity (quasi-transitivity: a prior to b, b prior to c, c not prior to a; with a = c, everything depends on everything but itself). Alternatively, full dependence between neighbours plus partial dependence at a distance leaves only partial self-dependence. Or, the preferred option, only symmetric partial dependence, with each entity co-determining the whole that fixes some of its features: “x is P, therefore Str is S, therefore x is R” with P ≠ R (p. 14). An optional quasi-reflexivity principle handles one-object structures by giving the lone object primitive features (fn. 27).
- §4.2 differences from holisms. Unlike Teller’s relational holism, the view is general, not only about entanglement, and uses dependence rather than concrete relations. Unlike Ismael & Schaffer’s common-ground holism, it has no monistic priority of the whole, no third grounding entity, and rejects their Grounding Principle. It is claimed to be “empirically equivalent” to theirs (p. 16).
- §4.3 applications. Entangled systems are individuals that come to depend mutually with respect to their properties (with Calosi & Morganti). AQFT: the dependence holds among local algebras. Particle kinds: “it doesn’t make a difference” whether one reads them as structure-first or mutually dependent objects, so independent advantages decide (p. 18). The Pauli principle is a modal fact about fermion-kind dependence networks, of which actual anticorrelations are instances (speculative, p. 19). Spacetime: only points exist, mutually dependent for existence and metric features, and spatiotemporal relations are dependent. That is closer to substantivalism, and the gap from moderate OSR is small here (fn. 32).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Eliminativist OSR cannot reduce the intrinsic, state-independent properties that distinguish actual particle kinds | moderate | informal argument from the group-theoretic classification (same groups, different irreps; actual kinds fixed empirically), following McKenzie and Wolff, §3.1 |
| C2 | The permutation-invariance argument for contextual identity fails at both premises | moderate | counter-models cited (counterpart theory, super-essentialism, empirically idle haecceitistic differences), §3.3. Morganti defers the details (“we cannot enter the details of it here”) |
| C3 | Symmetric ontological dependence need not entail self-dependence or uninformative explanation | moderate | logical point about the relation’s properties (non-transitive or partial dependence), §4.1; the formal claims are elementary and correct as stated, but no model of the dependence relation is given |
| C4 | Metaphysical coherentism is superior to strong, moderate and contextual OSR | weak | comparative evaluation by theoretical virtues (one category, no reduction of all properties, more than identity), §4.3; the argument depends on C1 and C2 |
| C5 | Coherentism “delivers the same goods as OSR” for the physics (entanglement, particle kinds, AQFT, spacetime) | weak | asserted translations, §4.3. The entanglement case is deferred to Calosi & Morganti, and the Pauli-principle unification is labelled “just a suggestion” |
| C6 | Science does not force an ontology of irreducible physical relations; all structuralist talk can be translated into mutually dependent objects | weak | conclusion (§5), phrased as “it could be contended”; rests on C5 |
| C7 | Coherentism is empirically equivalent to Ismael–Schaffer holism but avoids monism and a third entity | moderate | informal argument (§4.2, fn. 28) |
Method
Conceptual analysis in analytic metaphysics. He classifies OSR (strong, moderate, contextualist), assesses each against physics-derived desiderata (§2), and reconstructs the dialectic as the triad (1) relational structure is basic, (2) so relations are prior to objects, (3) no OSR is satisfactory. He resolves it by rejecting (2) and reinterpreting (1). Then he builds a positive view from the order-theoretic properties of dependence and grounding (reflexivity, symmetry, transitivity; full vs partial) and compares it with rival holisms.
Concepts
- Metaphysical coherentism: dependence structure contains cycles or holistic networks rather than a well-founded hierarchy; object-like entities depend symmetrically, at least partially, on one another.
- Foundationalism / infinitism: well-founded dependence with a fundamental level (pluralist atomism or a Parmenidean One); or infinite descending chains (“relations all the way down”, which structuralists may invoke, Ladyman & Ross 2007, p. 178).
- ODO (after Linnebo 2008): “Each object in D depends on every other object in D.”
- Strong / moderate / contextualist OSR: only relations exist / objects are bare placeholders on a par with relations / only identity facts are structural.
- Partial vs full dependence: depending on something together with other things vs on it alone. In a partial cycle a-b-c-d-a, a is fully grounded by a, b, c, d together (fn. 24).
- Quasi-reflexivity: x partially depends on itself iff it partially depends on something else.
- Holistic coherentism: each entity co-determines a whole (or merely a plurality) that fixes some of its own features.
Connections
Morganti engages the OSR metaphysics catalogued in LIT-045 (SEP Structural Realism): his three OSR types map onto that entry’s OSR theses 1 (eliminativism), 4 (Esfeld’s moderate SR) and 5 (identity grounded in structure). He answers McKenzie (2014, forthcoming) and Wolff (2012, 2014). He contrasts with Teller (1986) and Ismael & Schaffer, and builds on Fine’s (1995) reciprocal dependence, Jenkins (2011) and Correia (2014) on irreflexivity. His AQFT remarks cite French 2013 on unitary inequivalence (compare LIT-170, French on the same problem, not read here). Floridi’s LIT-152 reaches a structurally similar conclusion by another road: its “package hypothesis”, where relata and relations “co-exist as a package”, is close to Morganti’s mutual dependence, though Floridi keeps relations as co-fundamental where Morganti demotes them. Chakravartty’s non-eliminative OSR (LIT-196) is the other priority-based reading. Morganti’s fn. 19 on well-foundedness not being an axiom of mereology bears on LIT-124 (Mereology). Ladyman & Lorenzetti (LIT-184) take the opposite route: they keep modal structure fundamental and make objects scale-relative real patterns.
Metaphysics: the work defends a thesis about the global form of fundamentality (non-well-founded, symmetric dependence) and a one-category object ontology, and presupposes that dependence and ground track metaphysical explanation. It is metaphysics in the tag’s sense, and primarily so.
Bearing on the record
It carries nothing for ML practice. In this record it is the anchor for a dependence and grounding reading of structuralism, and a natural source for any THEORY-level entry comparing relation-first OSR (LIT-045, LIT-184), informational SR (LIT-152) and object-based coherentism. Where the record cites this paper for “symmetric dependence”, it should say partial dependence.
Note of 2026-10-16: THEORY-034 now cites LIT-173 as a source.
Limitations
- The positive view is modelled only by schematic cycles. No formal model of partial ground is given, so C3’s consistency claim is at the level of the properties the relation has, not a demonstrated structure.
- The key anti-OSR premise (C1) is borrowed from McKenzie and Wolff and not argued afresh. An OSRist who treats the particle spectrum as part of the structure is not engaged.
- The applications are sketches. The entanglement case is outsourced to Calosi & Morganti, and the unification via the Pauli principle is labelled speculative. On particle kinds Morganti himself concedes coherentism and OSR make no difference (p. 18), so the case rests on theoretical virtues.
- The draft treats ontological dependence and ground as interchangeable. The published version admits they are not obviously synonymous, and that ground relates facts, not objects. The relata of the coherentist relation (objects or facts) are therefore left unsettled.
- Whether option (i), coherentist OSR, differs from option (ii), replacing OSR, is set aside as “otiose, mostly terminological” (p. 17).
Open questions
- Is there a model of partial ground or dependence with symmetric cycles that satisfies the logic Morganti needs, and does it yield informative explanations? A formal semantics would settle C3.
- Can a structuralist absorb the particle-kind spectrum into structure (e.g. via the full Standard Model gauge group plus couplings) and so block C1?
- If spatiotemporal relations are dependent entities, does the coherentist spacetime survive the hole argument better or worse than moderate OSR?
Corrections to the seeded skim
- The dossier says the draft handles informative symmetric explanation “only briefly (pp. 12–13)” and that the published abstract’s added claims are unchecked. Both of those abstract claims are argued in the draft. Informative explanation under symmetric dependence takes most of §4.1, pp. 12–15: irreflexivity kept through antitransitivity or quasi-transitivity (Sen 1969), cycles of full and partial dependence, symmetric partial dependence only, quasi-reflexivity, and the schema “x is P, therefore Str is S, therefore x is R”. “Best balance of naturalism and consistency” is §4.3’s comparative argument, pp. 17–18. The published endnotes add one note, which concedes that “is grounded in” and “ontologically depends on” are not obviously synonyms and that ground normally relates facts. That concession is not in the draft, and it matters for evaluating the argument, since the argument trades on the two notions interchangeably.
- The dossier’s summary is “symmetric (mutual) ontological dependence”. The operative relation is symmetric partial dependence, which Morganti calls “the most plausible option” (p. 13). Full self-dependence is explicitly ruled out, and a whole may be only a plurality (fn. 25, fn. 28). The view is expressly not monism.
- Precision: the paper assumes OSR is justified and brackets French’s underdetermination argument (§1, fn. 1). Its target is only the metaphysics.
- Precision: coherentism is offered as either a reading of OSR (option i) or a replacement for it (option ii). Morganti prefers ii, “much more convincing as things stand today” (p. 17), so the title’s “from OSR to” means leaving OSR, not refining it.
- metaphysics tag: justified. primary topic: metaphysics (keep). The work is a thesis in the metaphysics of fundamentality and dependence (foundationalism vs infinitism vs coherentism, the logic of ground, monism), with physics as the test case.