Source
Craver and Bechtel (2007), LIT-1928, pp. 548, 553–554 and nn. 10–11; read in NOTE-1491.
What was actually shown
-
The experiments (pp. 553–554). Four kinds:
- correlating a component’s presence or activity with the whole’s behaviour (microsomes and protein secretion);
- lesioning or retarding a component and finding a deficit;
- stimulating a component (agonists, precursors);
- intervening on the whole and recording components (functional imaging).
“These four kinds of experiment share a form with Mill’s methods, giving the sense that a causal relationship is being tested.” They run in both directions, “scientists seem to be exploiting a symmetrical relation”.
-
The qualification (n. 10, p. 553). A deficit after a lesion “is prima facie evidence that X is a component in the mechanism”, and “It is only prima facie since if the component is not interactively integrated into the system, it could also be construed as constituting a background condition on the operation of the mechanism, not as part of it.” Note 11 (p. 554) says the same of imaging: judgments “depend on discovering that the components are deeply integrated into the system”.
-
The definition (p. 548) makes X a component if it is “one of the entities or activities organized such that S ψs”. It gives no criterion of organization or integration.
-
What could have failed. The paper could have offered an experiment, or a condition on the four, that separates a component from a background condition or from an outside item. It does not. Separating them is left to “interactively integrated”, which is not defined.
What this does not say
- That the experiments are useless. Within a boundary already drawn, a lesion that removes the phenomenon shows the lesioned item is needed for it. What they cannot settle is whether the item is inside.
- That Craver’s constitutive-relevance criterion has the same gap. Mutual manipulability, stated in Explaining the Brain (LIT-832) and “Constitutive explanatory relevance” (LIT-1930), is reported to add a parthood condition to two manipulability conditions. If it does, the boundary is supplied by parthood, which is a further input, not by the experiments. That reading is not checked here, since neither text was reached. The amended account, matched interlevel experiments grounded in causal betweenness (LIT-1931), was not reached either.
- That every outside item passes. A background condition that is not coupled to the phenomenon does not pass. The worry is items coupled to the whole in both directions.
Connections
- CLAIM-522 and CLAIM-530. An outside controller that reads the maintained capacities and supplies their production law is “interactively integrated” in any sense the paper gives the phrase. It passes all four experiments exactly as an internal producer does. That is CLAIM-530. It traces CLAIM-522’s proviso, that the boundary is declared and not given, into the mechanist’s account of constitution.
- CLAIM-194. Interactive integration is a coupling notion. So the one discriminator the paper offers inherits CLAIM-194’s point: it has “no term for who made the coupling”.
- QUESTION-086. The experiments answer “by what intervention” only relative to a boundary drawn beforehand. That is the limit QUESTION-086’s note of 2026-10-26 already records for CASE-161 and CASE-163.
- THEORY-1289. Bechtel and Abrahamsen’s clamping test is a fifth, model-based experiment of this family, with the same limit.