NOTE-149: Chakravartty — Ontological Priority and non-eliminative OSR

Contribution

The chapter gives non-eliminative OSR a precise priority thesis, relation-dependence for the determination of identity (principle (2)). It then maps the possible theories of relata along a thick-to-thin spectrum and shows that no point on the spectrum satisfies the thesis short of eliminativism. The result is a dilemma, stated as the chapter’s conclusion (p. 200): “one cannot deny eliminativism and a role for the intrinsic in the analysis of identity in connection with the relata of scientific discourse; if we deny one, we are stuck with the other”.

Key insight

“Relations are prior to relata” needs a specific dependence relation. The only one general enough is identity-determination. But every coherent relatum has something intrinsic (qualities, dispositions, even a haecceity), and whatever is intrinsic is the natural candidate for fixing identity. Take all of it away and there is no longer anything to stand in the relations. A cat’s grin without the cat is eliminativism under another name.

Assumptions

• (1) Ontological priority as asymmetric dependence: x is less fundamental than y with respect to R iff x depends on y with respect to R and not vice versa (p. 192).
• (2) The relevant R for non-eliminative OSR is determination of identity: relata are less fundamental than their relations iff they depend on those relations for the determination of their identity and not vice versa (p. 193). The chapter argues that this is the common thread in the OSR literature; it does not prove it.
• Concrete structure: “structure” means first-order relations between specific kinds of relata. The ideal gas law is the example (p. 189).
• Broad “object”: anything associated with a group of properties that cohere at a location. This covers mass-noun relata such as plasma and kinetic energy, and not only individuals (p. 189).
• Intrinsic vs extrinsic: the standard parsing, in which intrinsic features are had “independently of accompaniment or loneliness” (p. 201).
• Dispositions as generally intrinsic: the causal powers studied by science are “generally intrinsic properties” (p. 195, fn. 5).
• Orthodox reading of entanglement: granted for argument’s sake. Entangled particles’ relations do not supervene on their intrinsic properties (p. 204).

Key results

• Substance theories (bare substrata, haecceities) are incompatible with OSR, because identity is wholly independent of relations (§10.3, pp. 193–194).
• Bundle theories remove substrata, but the identity-determining properties are still intrinsic to the bundle (p. 194).
• Dispositional essentialism plus bundles (Shoemaker: a property’s identity is “completely determined by its potential for contributing to the causal powers”) individuates properties by their potential for relations. Dispositions are intrinsic and exist whether or not they are manifesting, so the relata still do not depend on actual relations for their identity (§10.4, pp. 194–196). Categoricalism, with quiddities, fails for the same reason as haecceities.
• No intrinsic features at all equals eliminativism: “an object with no intrinsic features at all … is not an object at all” (§10.5, p. 197).
• The graph analogy fails outside Platonism or Pythagoreanism. Asymmetric graphs, where each node has a “unique, purely structural description” (Dipert), give extrinsic identity to locations. When a graph represents a concrete system, its nodes denote things external to the graph, and the labels do not supply their identity criteria (pp. 197–200).
• The no-priority variant (relations and relata on a par) fares no better. Either it has intrinsic properties, which fix identity, or it has none, which is eliminativism (pp. 200–201).
• Asymmetry argument: extrinsic properties cannot by themselves constitute an object, because characterising the bearer extrinsically leads to regress or circularity. This is “at the very least, a serious conceptual puzzle” (p. 201).
• Quantum case: for extrinsic dispositions to help, either all of a particle’s properties would have to be extrinsic, or its intrinsic ones would have to fail to fix identity. Neither holds: mass and charge are “state-independent intrinsic properties”, and a particle must instantiate particular values of them to be the kind of thing it is (p. 204).
• Spacetime case: Esfeld & Lam’s “an object as such is nothing but what bears the relations” invites the question of what the “what” is. If spacetime points have intrinsic features, their identity is not purely extrinsic, whatever their individuation needs (pp. 204–205).

Claims

id	claim	strength	support
C1	The only sufficiently general dependence relation for non-eliminative OSR’s priority thesis is determination of identity.	moderate	informal argument from a survey of the OSR literature, §10.3
C2	Substance, bundle, and dispositional-essentialist theories each leave identity fixed by intrinsic features.	moderate	conceptual argument, §§10.3–10.4
C3	A relatum with no intrinsic features is no relatum, so “purely extrinsic” non-eliminative OSR is eliminativism.	moderate–weak	conceptual argument, including the regress/asymmetry argument of p. 201; the author concedes it may look question-begging
C4	Graph-theoretic extrinsic identity identifies locations in a structure, not the identities of concrete occupants, unless the world is itself mathematical.	moderate	informal argument, pp. 198–200
C5	The no-priority variant of OSR falls to the same dilemma.	moderate	argument by cases, pp. 200–201
C6	Quantum entanglement does not make particle identity extrinsic, because mass and charge are intrinsic and identity-fixing.	moderate–weak	argument from textbook physics. The claim that mass and charge values determine identity (as opposed to kind) is asserted, and the chapter does not engage group-theoretic readings of them
C7	Citing quantum non-individuation for OSR conflates individuation with identity.	moderate	conceptual distinction, p. 204
C8	Nothing here refutes eliminative OSR.	—	the author’s own statement, p. 200

Concepts

• Eliminative vs non-eliminative OSR: eliminative OSR says there are relations but no relata (French’s “façon de parler” view; Ladyman & Ross, fn. 3). Non-eliminative OSR says there are relata, but their natures are fixed by the relations they stand in (Saunders, Cao, Stachel, Ladyman 2007).
• Ontological priority: asymmetric dependence with respect to a named relation, as in principles (1) and (2).
• Concrete structure: first-order relations between specific relata, as opposed to abstract higher-order (isomorphism-class) structure.
• Thick/thin object ontologies: Fig. 10.1’s spectrum, running from substance realism, through object realism and OSR, to eliminativism.
• Extrinsic disposition: a disposition possessed in virtue of relations to something external (McKitrick 2003). It is distinguished from intrinsic dispositions described through extrinsic manifestation conditions, such as solubility.
• Identity vs individuation: what makes a thing the thing it is, as opposed to what distinguishes it from others of its kind.

Connections

The target literature is Saunders (2003), Ladyman (2007) “On the identity and diversity of objects in a structure”, Ladyman & Ross (2007), Esfeld (2003, 2004), Esfeld & Lam (2008) and Floridi (2008). The Floridi reference appears to be the held LIT-152, “A defence of informational structural realism”. Chakravartty places Floridi among the no-priority views (fn. 9), and his dilemma applies to it directly. The SEP entry by Ladyman (LIT-045) is the chapter’s source for the generic definition of OSR, as reference [18], the 2007 edition. Morganti’s metaphysical coherentism (LIT-173) is a later attempt at the no-priority option, and reading it against this chapter is the natural test.

Its argument against the graph analogy bears directly on Leitgeb & Ladyman (LIT-154, read as x8). They defend primitive, structure-internal identity for mathematical places and explicitly leave physics open. Chakravartty’s location/occupant distinction is the reason the move may not transfer. Wallace (LIT-151, read as x5) sidesteps the dilemma: he states OSR as math-first metaphysics, not as a priority thesis about relations over relata, and he classes priority theses as the “second conception” of structural realism, which has drifted from the original motivations. Ladyman & Lorenzetti’s effective OSR (LIT-184) is a later OSR statement that the dilemma should be tested against. The dispositional material in §10.4 connects to Vetter (LIT-109) on dispositions.

Bearing on metaphysics: the chapter defends a thesis about the ontology of objects. Concrete objects have intrinsic natures that constitute their identity, and relations cannot be prior to relata in identity without eliminating them. It presupposes that identity requires something intrinsic (the asymmetry argument). The tag is justified and should lead.

Bearing on the record

It supplies a precise objection that any THEORY or reading in this record endorsing non-eliminative or no-priority OSR must answer. It does not bear on eliminative OSR or on math-first OSR. Nothing here carries over to ML practice or to the Anthology of the SOTA.

Note of 2026-10-16: THEORY-034 now cites LIT-196 as a source.

Limitations

• The argument depends on principle (2), that identity-determination is the dependence relation for OSR. A proponent who states priority in terms of grounding of existence, or of fundamentality as McKenzie does, is not directly addressed.
• The asymmetry argument (p. 201) is presented as a “serious conceptual puzzle”, not a proof. The author concedes that it will look question-begging to some.
• The physics section explicitly declines “a detailed refutation” of the case studies (p. 202). The mass-and-charge point is not engaged with representation-theoretic accounts, on which mass and spin are labels of irreducible representations of the Poincaré group, and with the view that these are structural.
• The chapter treats kind-membership (having a mass value) and identity (being this particle) together in the quantum case, while elsewhere insisting on separating individuation from identity.
• “Graph theory would only provide a helpful analogy … if we were to think of the world as exhaustively comprised of purely mathematical entities” (p. 200). This sets aside, without argument, the views on which physical structure is not mathematical but is still structure (Ladyman & Ross refuse to say what makes it physical, fn. 6).

Open questions

• Can a non-eliminative OSR be stated with a dependence relation other than identity-determination (grounding, fundamentality, or McKenzie-style priority) that escapes the dilemma?
• Is the intrinsic/extrinsic asymmetry sound? Can a coherent account of wholly extrinsic objects be given? That would answer the chapter’s puzzle directly.
• Do group-theoretic or gauge-theoretic accounts of mass and charge make them structural rather than intrinsic, and so undercut §10.6?

Corrections to the seeded skim

• The skim puts Fig. 10.1 in §10.5 and skips §10.2. Fig. 10.1, a spectrum of object ontologies running from “substance realism” (thick) to “eliminativism” (thin), with OSR located toward the thin end, is introduced in §10.2 “Theories of Object Ontology” (pp. 190–192). That section also names the non-eliminative OSR camp: Saunders, Cao, Stachel and Ladyman 2007.
• The skim says “Saunders and Ladyman & Ross are cited as holding that objects are ‘purged of their intrinsic natures’”. That quotation is Ladyman & Ross’s alone (p. 193). Saunders is cited for “a particular body is no more than a particular pattern-position”.
• The skim’s summary says OSR “collapses back into eliminativism”. That applies to non-eliminative OSR, and only if it strips all intrinsic features. The paper says in so many words that “nothing I have said in this essay tells against eliminative OSR” (p. 200). The eliminativist’s problem, making sense of concrete relations without relata, is only said to remain open.
• The skim’s §10.5 bullet leaves out two of that section’s three arguments. The first is against the graph-theory analogy (Dipert 1997; Ladyman 2007, reproduced as Fig. 10.2): graph labels “simply identify locations in a structure”, which is not the same as “constituting the identity of something that occupies that location” (p. 199). Without Platonism or Pythagoreanism, concrete and mathematical objects differ in kind. The second is against the no-priority variant (Esfeld, Esfeld & Lam, Rickles, Pooley, Floridi; fn. 9): if it keeps intrinsic properties, they fix identity, and if it keeps none, it is eliminativism.
• The skim leaves out the §10.5 argument that objects cannot be purely extrinsic (pp. 201). To attribute an extrinsic property is to presuppose a prior grasp of the “it” that stands in the relation; a purely extrinsic characterisation regresses or circles. The author himself flags this as the step some will find question-begging.
• The skim leaves out two §10.6 points. Treating quantum non-individuation as evidence for OSR “conflate[s] individuation with identity” (p. 204). And the same “what is the what?” objection is applied to Esfeld & Lam’s spacetime points. The closing concession, that extrinsic identity may exist in special cases such as membership of a phylogenetic species, but always presupposes an intrinsically constituted bearer (p. 205), is also missing.
• metaphysics tag: justified. It is the chapter’s core subject.
• primary topic: metaphysics. The chapter describes its target as “a provocative metaphysical doctrine” (p. 187). It argues entirely about dependence, identity and intrinsicality, and sets aside the scientific-realism debate and the scope of “lessons from physics” (p. 202). philosophy-of-science is still justified as a second tag, because the frame is OSR as a form of scientific realism. identity is justified as a third.