Contribution
After the negative results (NOTE-1493, NOTE-1490), it shows that an intervention-based sufficient condition for constitution exists after all, if the intervention is restricted the right way and three metaphysical principles are granted. It also shows why the obvious repair, admitting Woodward’s “permissibly fat-handed” interventions, is circular.
Key insight
A constitutionally revealing intervention may be fat-handed across levels but must be surgical within each level. If it changes a part and the whole at once, the whole’s change must be realized at that part’s level by some constituent. If the part were not a constituent, the intervention would have to change a second behaviour at the same level, and so would not be horizontally surgical. Simultaneity does the work. Because changes that happen together cannot be causally related, the second change cannot be downstream of the first.
Assumptions
- Parthood and levels are given. “We will henceforth assume clarity on the spatiotemporal parthood relations obtaining among analyzed behaviors” (p. 5). Levels are local, defined by direct proper parthood relative to the phenomenon, after Eronen 2013; “merely instrumental” (footnote 6). The Parthood principle is taken from Leuridan’s p. 410.
- Non-gerrymandered variables are assumed without a criterion (footnote 5).
- (UC) Universal Constitution (p. 5): “Every (change in a) state of any (non-fundamental) phenomenon Ψ is necessarily and simultaneously realized by (a change in) the state of at least one constituent Φᵏᵢ of Ψ on every lower level, such that Ψ ≠ Φᵏᵢ.” No gaps in constitutional hierarchies.
- (SIM) Simultaneity (p. 6): “Simultaneous changes cannot be causally related.”
- (TR) Transitivity of co-occurrence, at non-relativistic distances (p. 10).
Key results
- Systematic Fat-Handedness (p. 6). From UC and SIM: a cause Σ of Ψ changes some constituent simultaneously with Ψ; the path Σ → Ψ → Φ is excluded by SIM; so Σ is a common cause. “Every cause of a phenomenon is also a cause of some constituent or other of that phenomenon on a different causal path (Romero 2015; Baumgartner and Gebharter, 2016).”
- pFAT is question-begging (pp. 7–8). Whether a common cause of Ψ and Φ is “permissibly” fat-handed under Woodward 2015 “depends on whether Φᵏᵢ belongs to the supervenience base of Ψ, which, in turn, depends on whether Φᵏᵢ is a constituent of Ψ.”
- FAT is a non-starter (p. 8). Sterile effects (the heat sink warmed by the first transistor) and detached parts (a disconnected heat sink smashed by the same hammer as the transistor) share common causes with the phenomenon. “Only simultaneous changes in parts and phenomena are revealing of constitution”, and a revealing intervention “must not directly target multiple parts … on the same level” (pp. 8–9).
- (H) (p. 9). I is a horizontally surgical intervention variable on a part Φᵏᵢ of Ψ w.r.t. Ψ iff (i) I causes Φᵏᵢ; (ii) if I changes both Φᵏᵢ and Ψ, the changes are simultaneous; (iii) I directly causes at most one behaviour on every level lower than Ψ’s.
- (SUF) and its proof (pp. 9–11). SUF: some possible horizontally surgical intervention on Φᵏᵢ changes both Φᵏᵢ and Ψ. Then Φᵏᵢ is a constituent. Reductio: assume (1) such an intervention and (2) Φᵏᵢ ∉ Φ. By H the changes are simultaneous (3). By UC some constituent Φʰⱼ on the same level realizes Ψ’s change simultaneously (4). By TR, Φᵏᵢ’s and Φʰⱼ’s changes co-occur (5). By SIM they are not causally related, so both are direct effects of I (6). So I directly causes two behaviours on one level (7), contradicting H(iii). Checked: valid on UC, SIM and TR. Example: a needle that changes hippocampal map generation and the oxygen concentration in a vessel, but not neural firing, cannot exist if oxygenation is a sterile effect.
- hFAT (pp. 11–12). Making SUF also necessary yields a definition. It is not paradoxical, as MM was, but “a matter of scope”, and shares interventionism’s doubt about whether ideal interventions exist for every dependence (the moon and the tides).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Given UC and SIM, every cause of a phenomenon is a common cause of it and some constituent | strong | p. 6, deduction |
| C2 | Defining constitution by Woodward’s permissibly fat-handed interventions is circular | strong | pp. 7–8 |
| C3 | One horizontally surgical intervention that changes a part and the phenomenon suffices for the part’s being a constituent | strong, on its premises | §4 reductio, valid given UC, SIM, TR and settled parthood |
| C4 | SUF can be upgraded to a definition (hFAT) without MM’s paradox | weak | pp. 11–12, argued by analogy with interventionism; necessity not argued |
Concepts
- Horizontally surgical intervention — fat-handed across levels, surgical within each: at most one directly caused behaviour per lower level, and simultaneous change of target and phenomenon (p. 9).
- Permissibly fat-handed intervention — a common cause whose parallel effects are related by supervenience, after Woodward 2015 (p. 7).
- Constituting set — all and only the constituents responsible for Ψ in its mechanistic context (p. 4).
Connections
- Takes Systematic Fat-Handedness from Romero (LIT-1933), Baumgartner & Gebharter (LIT-1934) and Baumgartner & Casini (LIT-1939), and presents itself as the interventionist alternative to the latter’s abductive NDC (p. 9).
- Footnote 3: Krickel’s separate route (LIT-1925), “MM may be saved if the phenomenon is represented by multiple variables standing for different temporal phases”.
- Footnote 2 reports that Craver acknowledges the difficulty (Craver & Tabery 2017) and holds it a matter of formulation.
Bearing on the record’s own argument
- CLAIM-522 and CLAIM-194: what has to be given. The one positive intervention result in this debate needs parthood and a hierarchy of levels as inputs (p. 5) and outputs constitution. It does not output a boundary. That is CLAIM-522’s “relative to a declared boundary” and CLAIM-194’s “a coupling statistic has no term for who made the coupling”, found in a literature that set out to analyse constitution by intervention. Same point, reached for part–whole constitution.
- CLAIM-522’s interventions are not horizontally surgical, and need not be. H(ii) requires the target and the phenomenon to change simultaneously. A production knockout in CASE-189 changes the capacities only over time. So SUF neither applies to nor is needed for CLAIM-522’s conclusion, which is about where a cause sits, not what constitutes what (ARG-050).
- CLAIM-480: Simultaneity is a property of a description’s time scale. SIM and H(ii) make the line between causation and constitution the line between delayed and simultaneous change. CLAIM-480’s chain has an interior x₂ whose response to the boundary is instantaneous when it equilibrates fast (E_eff, memoryless) and delayed when it relaxes slowly (a memory kernel). So whether the interior’s change counts as simultaneous with the boundary’s, and so whether the dependence is constitutive by SIM, depends on the time resolution of the description. The paper’s TR is restricted to non-relativistic distances; it has no parallel restriction to time scales. CLAIM-528 part 3 states this. Krickel’s temporal-phase move (footnote 3) points the same way and has not been read here.
- CLAIM-515’s “realized”. SUF is the one route by which a single intervention shows realization in the synchronic sense. It needs UC, SIM, TR and the level hierarchy. An organization’s capability is rarely given with a level hierarchy of parts, so for the record’s organizations the route is mostly closed, and the realization question falls back to abduction (NOTE-1490).
- QUESTION-284. A positive answer to “what evidence would tell constitution from … ordinary causal modification”, for part–whole constitution: a simultaneous change in whole and part under an intervention that touches nothing else at that level. Its price is QUESTION-284’s second half: the boundary and the levels must already be fixed.
- TERM-088. The paper’s levels are local parthood levels relative to a phenomenon (footnote 6), neither the term’s senses nor CLAIM-526’s levels of description.
Limitations
- UC is strong. Constituents “on every lower level”, with no gaps, and realization necessary and simultaneous. For organizations whose behaviour is realized by processes that take time, it is unclear what UC asserts.
- SIM is assumed, not argued, beyond Craver and Bechtel 2007. It is sharp only at a fixed time resolution.
- Parthood is assumed clear, and Leuridan’s endosymbiosis case (LIT-1929) is not discussed. Gerrymandering is set aside without a criterion (footnote 5).
- Whether horizontally surgical interventions exist in any real system is not shown; hFAT’s necessity half is only suggested.
Open questions
- In a dynamical model with explicit relaxation times, is there an intervention that is horizontally surgical at one time resolution and not at another? CLAIM-480’s chain is a candidate to compute.
- Does SUF survive replacing SIM with “changes within the relaxation time of the faster process are not causally related”?