Source
Baumgartner and Casini (2017), LIT-1939, §§4–5, pp. 12–18 (NOTE-1490); the criterion it develops is Baumgartner and Gebharter’s (FH) and (CR), LIT-1934, pp. 20–22 (NOTE-1493).
What was actually shown
The definition (B&C p. 15). Φ₁(X₁) is constitutively relevant to Ψ(S) iff some variable set V contains Ψ(S) and a set Φ(X) = {Φ₁(X₁),…,Φₙ(Xₙ)} such that: (1) Parthood: every complex instance of Φ is part of an instance of Ψ; (2.i) every cause of Ψ in V is a common cause of Ψ and at least one Φᵢ; (2.ii) no proper subset of Φ satisfies (2.i); (3) No De-Coupling: the coupling “cannot be broken (invalidated) by expanding V.”
Why redundancy-freedom. Without (2.ii), a sterile effect of a constituent (the car’s air conditioning, run by the engine) or a detached part damaged by the same cause (the ashtray in a crash) would join the set, since every cause of the phenomenon reaches them too, through a constituent. Dropping them leaves the coupling intact, so (2.ii) excludes them (B&C p. 13).
Why abductive. Data from any finite V are reproduced by a pure common-cause model (THEORY-1288). The two models differ in what they predict when V is expanded. The common-cause model gives no reason why a surgical cause of Ψ should never appear; the constitutive model makes it impossible. When repeated expansions fail to break the coupling, the constitutive model is preferred “because it exceeds the latter in explanatory power” (B&C pp. 14–15). B&G give the same argument for (FH) (pp. 20–21).
Logical form (B&C p. 16). (3) is a negative existential, so NDC constitution can only be “inductively corroborated”, like causal irrelevance. A surgical cause of Ψ falsifies one candidate set conclusively, but not the constitutive relevance of its members, who may belong to another set (pp. 17–18).
Extrinsic, not relativized (pp. 16–17). Whether Φᵢ is a constituent depends on whether it belongs to some constituting set, not on a chosen V; across expansions the relation holds throughout or not at all.
What this does not say
- How to test it. “For reasons of space, we must postpone a methodological operationalization of NDC to another occasion” (p. 17); how much expansion suffices has “no universal rule” (footnote 10).
- That practice proceeds this way. “We have done none of that sort for NDC here” (p. 19).
- That the constituting set is unique, or what follows if Ψ has two.
- Where the boundary is. Condition (1) takes parthood, a thin spatiotemporal containment (footnote 1), as given.
- Anything about constitution by relations in general. The relation is a part’s behaviour realizing a whole’s behaviour, not relations fixing what a relatum is.
Connections
- QUESTION-284 asks, for a stated kind of entity, what evidence tells constitution from characterization and from causal modification. For part–whole constitution in mechanisms, this is one answer: a second-order pattern, the persistent failure to decouple, and nothing less. Whether a like route exists for relations constituting relata is the question’s harder half.
- CLAIM-477: constitutive relevance here is extrinsic to the pair and fixed by membership in a set, a case of identity-of-role fixed by configuration. Same structure, not shown same relation.
- CLAIM-405: set-relative and not relativized, so a verdict, once the phenomenon is fixed, can be false.
- CLAIM-522: the outside controller of CLAIM-522 plays the part of NDC’s de-coupling cause, something that moves the capacities while running through nothing inside the declared boundary. Same structure, not shown same relation: CLAIM-522’s production is diachronic.
- CLAIM-515: on this account, “realized” in the synchronic sense is inferred by abduction, not shown by discriminating interventions (CLAIM-528).