Contribution
The paper sets out, for a general audience, why structural realism should be an ontic rather than an epistemic thesis. It collects three physics-based reasons to give up an ontology of individuals with intrinsic properties: quantum statistics, non-supervenient entangled relations and the hole argument. It proposes that what science describes, and what survives theory change, is the modal structure of the phenomena. Its distinctive contribution relative to Ladyman 1998 (which it cites for the claim that epistemic SR does not improve on standard realism, p. 61) is the argument against Lewis’s Humean supervenience (pp. 70–73).
Key insight
Standard scientific realism assumes “particularism”. The world’s structure is fixed by individuals and their intrinsic properties, and relations supervene on them. Quantum mechanics breaks this assumption twice. First, the individuality of particles is underdetermined by the physics. Second, entangled relations fail to supervene even on the independent relational properties of the relata. A realism that survives theory change and fits current physics should therefore commit to structure, which Ladyman specifies as modal structure, and not to the objects that bear it.
Assumptions
- Realism is the target. The paper assumes a realist should be able to say what it is to be a realist about non-relativistic QM and not only about a final theory (p. 66).
- The quantum description is complete (“If we assume for now that the quantum description is complete”, p. 65), and the Principle of Statistical Mechanics holds: equipossible states are equiprobable (fn. 7). The PII argument depends on both.
- Indistinguishability postulate: ⟨Pψ₁₂|Q|Pψ₁₂⟩ = ⟨ψ₁₂|Q|ψ₁₂⟩ for every observable Q and every ψ₁₂, where P is the permutation operator (p. 63).
- Lewis’s definitions: internal and external relations and intrinsic properties (p. 71, fn. 6). The argument against Humean supervenience is run on Lewis’s own terms.
- Naturalism about metaphysics: “if we are to be scientific realists … we should surely have our metaphysics informed by our best physics” (pp. 70–71).
- Ladyman (2000) is taken as having shown that constructive empiricism needs objective modality (p. 73). That argument is imported and not given here.
Key results
The results are conceptual.
- The pessimistic meta-induction is stated as a three-step argument (p. 58). Ether and caloric are cases of mature theories with novel predictive success whose central terms do not refer (p. 59). Both standard realist responses, reference-preserving semantics and Psillos’s “only the essential terms” strategy, risk being too easy or ad hoc (p. 59).
- Quantum statistics: with 2 particles and 2 states there are 4 classical configurations (each p = 1/4) and 3 quantum ones (each p = 1/3) (p. 62).
- “Singlet state” (p. 65): the paper prints Ψ₁₂ = 1/√2{|↑⟩₁|↓⟩₂ + |↓⟩₁|↑⟩₂}, checked against the page image. With a ”+” this is the m = 0 triplet state, not the singlet, which takes a ”−”. That is a slip in the paper, but the argument does not depend on it, because under either sign no property of particle 1 cannot also be predicated of particle 2. So, given completeness, either the PII is false and particles have transcendent individuality, or they are not individuals.
- Underdetermination (French 1989): the individuals and non-individuals packages are both consistent with QM (p. 65). Three arguments for non-individuals (PII, surplus structure, van Fraassen’s “empirically surplus factors”) are each rebutted (p. 66). Paraparticle states are one reason the surplus-structure argument fails.
- QFT does not dissolve the problem: particle number is frame-dependent, and Teller’s “quanta” can be aggregated but not enumerated. The individuality question moves to fields and spacetime points (pp. 66–68).
- Spacetime: general covariance and the hole argument (Earman & Norton 1987) tell against point substantivalism. DiSalle and Hoefer locate the fault in “primitive identity” of points (pp. 68–69).
- Against Humean supervenience: Lewis’s external relations, such as spatial separation, supervene on the relational properties each relatum has independently. Entangled relations do not, because “in an entangled state each particle has no state of its own” (p. 72). So they are “much more independent of the properties of the individual particles than spatio-temporal relations between classical objects”, and “this would seem to refute Humean supervenience in so far as the doctrine is supposed to be inspired by science” (pp. 72–73).
- Positive proposal (p. 73): “abandon the attempt to interpret physical theory in terms of underlying objects and properties … about structure and relations directly”. The structure is “the modal structure of the phenomena”. SR differs from standard realism in treating modal relations between phenomena as “ontologically basic” rather than supervenient on unobservable objects.
- Reply to van Fraassen’s collapse objection (“If there is no nonstructure, there is no structure either”): “the phenomena have structure but they are not structure” (p. 74).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Epistemic forms of SR do not significantly improve on standard realism, so SR should be metaphysically revisionary | assertion | refers to Ladyman 1998, p. 61. Not argued here |
| C2 | Permutation invariance and entanglement leave the individuality of quantum particles underdetermined between two metaphysical packages | informal argument | French 1989, French & Redhead 1988, pp. 64–66 |
| C3 | No consideration in the literature breaks that underdetermination | informal argument | three arguments rebutted, p. 66 |
| C4 | Moving to QFT does not remove the individuality problem | informal argument | Teller 1990, Redhead 1996, pp. 66–68 |
| C5 | Entangled relations fail to supervene on intrinsic properties and on the independent relational properties of their relata, so science-inspired Humean supervenience is refuted | informal argument | Teller 1989, Butterfield 1992, Lewis 1986, pp. 71–73 |
| C6 | Scientific theories describe the modal structure of the phenomena, and this explains novel predictive success and counterfactual support | assertion | two sentences, p. 73, leaning on Ladyman 2000 |
| C7 | OSR does not collapse into standard realism, because the phenomena have structure but are not structure | assertion | p. 74 |
Method
Conceptual analysis applied to case studies from physics: quantum statistics, the singlet state, QFT quanta and the hole argument. Lewis’s metaphysical definitions are used against Lewis. There is no formal apparatus beyond the displayed permutation-invariance condition and the singlet state.
Concepts
- Distinguishability vs individuality (Suárez): how we tell things apart, versus in virtue of what they are distinct (p. 64). Candidate principles of individuation are transcendent individuality (haecceity or bare particular), spatio-temporal trajectory, and a set of properties (p. 64).
- Particularism / exclusive monadism (Teller 1989; Dipert 1997): the assumption that structure consists of individuals with properties (pp. 69–70).
- Non-supervenient relations (Teller): relations not fixed by the monadic properties of their relata (p. 71).
- Humean supervenience (Lewis 1986): everything supervenes on local intrinsic qualities at points plus spatiotemporal relations (p. 70).
- Modal structure of the phenomena: the objective modal relations among possible and actual phenomena that SR is realist about (p. 73).
Connections
It presupposes Ladyman 1998, “What is structural realism?”, which was not reachable legitimately. It anticipates French & Ladyman 2003, “Remodelling structural realism”, cited as forthcoming. It is the source of Claim C5 in the SEP entry’s reading (NOTE-038 C5, “moderate”, reported). LIT-045’s §4 is the later, fuller survey by the same author. Its modal-structure thesis is what LIT-184 and r2 (Ladyman 2024) build the real-patterns programme on (“Realism about modal structure is a core commitment of Ontic Structural Realism (Ladyman 1998)”, r2 fn. 6). French’s paper in the same Philosophica issue defends SR about spacetime (fn. 9). The Krause quasi-set programme (fn. 8) is LIT-107’s subject.
Bearing on the record
It supplies the record’s first primary Ladyman text on OSR other than the SEP survey. It pins down what the founder claimed before Every Thing Must Go: realism about modal structure, an underdetermination argument about individuality, and an anti-Humean argument from entanglement, with no eliminativism. It carries nothing for ML practice.
Limitations
- The positive view gets two paragraphs (pp. 73–74). “Modal structure” is undefined, and the argument that realism needs it is outsourced to Ladyman 2000.
- The anti-Humean argument depends on the entangled state being the complete description (C5). A hidden-variables or primitive-ontology interpretation would contest it. The paper assumes completeness “for now” (p. 65) and does not return to it.
- The eliminative vs non-eliminative question is not posed. The closing reply to van Fraassen suggests non-eliminativism about phenomena but says nothing about unobservable relata.
- It is a short journal essay. Much of it (§1, most of §2) surveys existing debates and is not new argument.
Open questions
- If “the phenomena have structure but they are not structure”, what are the relata of the modal structure at the unobservable level? The paper leaves it open. Later answers diverge between French’s eliminativism and Ladyman & Ross’s “no fundamental level” (r2 fn. 9).
- What makes modal structure objective rather than a feature of models? The paper leaves this to Ladyman 2000 and later Berenstain & Ladyman 2012, which was not reachable here.
- Does the non-supervenience of entangled relations survive under interpretations that are not complete in the paper’s sense?
Corrections to the seeded skim
- The record has no reading of the founding papers (Ladyman 1998, French & Ladyman 2003). LIT-045 is its only source for what early OSR said. LIT-045’s list of seven OSR theses puts “eliminativism about individuals” first. This paper, Ladyman’s own statement three years after 1998, is not eliminativist. It recommends “we abandon the attempt to interpret physical theory in terms of underlying objects and properties of which the world is made” (p. 73). It then keeps a structure/non-structure distinction against van Fraassen: “the phenomena have structure but they are not structure” (p. 74). So early OSR in Ladyman’s own hands is (i) anti-particularist and anti-Humean, (ii) realist about modal structure, and (iii) silent on eliminating objects. A reading that files Ladyman as an eliminativist from the start (see NOTE-149’s gloss “Ladyman & Ross, fn. 3”, which follows LIT-196) should date the eliminativist-sounding formulations to later work. This paper does not contain them.
- The argument from quantum non-individuality is stated here as an underdetermination argument, not as a proof of non-individuality (pp. 65–66). French’s two “metaphysical packages” (individuals with restricted states, or non-individuals) are both left standing. Ladyman lists three considerations favouring non-individuals and immediately rebuts all three (p. 66): “the underdetermination is not easily broken”. OSR is offered as the way out of the underdetermination, not as its conclusion. LIT-045 and LIT-173 present the OSR motivation from permutation invariance as an argument for OSR. That is Ladyman’s later framing. In 2001 the motivation is that realism about individuals is metaphysically underdetermined, so realism should be about what both packages share.
- Where “modal structure” enters. LIT-045 says ontic structuralists understand structure as modal structure, and LIT-184 and LIT-170 take modal structure as OSR’s core. This paper is where that identification is first made in print by Ladyman, in two sentences on p. 73: “the structure described by scientific theories is the modal structure of the phenomena”. Its support is an argument by appeal to van Fraassen’s constructive empiricism (Ladyman 2000) and the explanatory role of counterfactuals and novel prediction. There is no account of what modal structure is. This is the gap LIT-170’s “profoundly modal conception of structure” and LIT-151’s math-first reading later try to fill.
- Against LIT-173 (Morganti): Morganti argues that OSR’s symmetry claims are better read as symmetric dependence among object-like entities. The 2001 paper’s actual metaphysical claim is narrower than the “relations are fundamental” thesis Morganti attacks. It is that entangled relations do not supervene even on the mutually independent relational properties of the relata (p. 72). That is compatible with Morganti’s coherentism, so the disagreement is smaller than LIT-173’s framing suggests.
- Against LIT-196 (Chakravartty): the paper concedes the premise Chakravartty uses. Entangled particles keep their “essential” state-independent properties, “charge, mass and so on” (p. 72). Ladyman’s claim is only that the non-supervenient relations are not grounded in them. So the paper is exactly the kind of non-eliminative position whose priority thesis LIT-196 says cannot be made good.
- LIT-151 (Wallace), LIT-170 (French) and LIT-152 (Floridi): no correction. The paper anticipates Wallace’s math-first reading (“mathematics has an ineliminable role to play in theories”, p. 74) but gives no account of theoretical equivalence. It predates the QFT inequivalence problem LIT-170 answers, although p. 67 already raises QFT’s particle-number and interpretation problems as reasons not to evade the individuality question.