Contribution
It turned the exclusion debate from an a priori dispute into a question about particular causal systems. It gave necessary and sufficient conditions for upward exclusion (the realizer is the cause) and downward exclusion (the higher-level property is), within an explicit counterfactual semantics.
Key insight
Ask whether the effect depends on how the higher-level property happened to be realized. If it would still have happened under a different nearby realization, the realizer makes no difference and the higher-level property is the cause.
Assumptions
- Causation as difference-making: F causes G iff F → G and ¬F → ¬G, evaluated by Lewis-style spheres with weak centring (the innermost sphere may hold worlds other than the actual one).
- Deterministic causation, binary properties, no overdetermination clauses.
- Supervenience may be contingent, holding in worlds relevantly like ours.
Key results
- Compatibility result (§IV). (1a) M → B, (1b) ¬M → ¬B, (2a) N → B and (2b) ¬N → ¬B hold together iff B holds in all closest M-worlds, fails in all closest ¬M-worlds, and fails in all closest ¬N-worlds that are M-worlds.
- Upward exclusion (§V). This holds iff N is a difference-maker for B and either B fails in some closest M-worlds that are ¬N-worlds, or B holds in some closest ¬M-worlds outside the smallest ¬N-permitting sphere.
- Downward exclusion (§V). This holds iff M is a difference-maker for B and B holds in some closest ¬N-worlds that are M-worlds: the relation is realization-insensitive.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Kim’s exclusion principle is false under difference-making | strong | counterexamples (pigeon, Andersen’s monkeys) within the semantics |
| C2 | The revised exclusion principle is contingent, holding in some systems and not others | strong | compatibility result, proved |
| C3 | A higher-level cause excludes its realizer iff the relation is realization-insensitive | strong | theorem, §V and appendix |
| C4 | Special-science causal relations are typically realization-insensitive | weak | an observation about practice, after Woodward |
Concepts
- proportionality — a cause specific enough and no more specific than needed for the effect (Yablo).
- realization-insensitive — the effect persists in some closest worlds where the higher-level property is realized otherwise.
- downward exclusion — the higher-level cause excludes its realizer.
Connections
It answers Kim, builds on Yablo’s proportionality and Woodward’s interventionism and his “insensitive” causation, and uses Andersen’s neural-decoding work as its example. Pernu (LIT-147) argues that example supports identity rather than autonomy.
Bearing on the record
Source of THEORY-133. It bears on THEORY-048: if psychological functions are realization-insensitive causes, biology implements them without replacing them. That is this paper’s downward exclusion. No relation is declared, since that theory’s sources do not use the framework.
For the owner’s essay (§5.3), it supplies the missing criterion. Northstar’s organizational property explains the outcome in its own right if the outcome would survive re-realizing it: different staff in the roles, a different route through the same authority structure. That is a testable question, and its answer can go either way.
Limitations
Binary properties and deterministic causation. Everything turns on which worlds count as closest, which the authors make context-dependent and discipline-relative without fixing.
Open questions
How to settle the similarity ordering empirically for a given system, and whether the result survives probabilistic causation.