Source
List & Menzies (2009), LIT-722, read in NOTE-560.
What was actually shown
List & Menzies (LIT-722) take F to cause G when, in relevantly similar situations, G follows F and fails without it. In that framework they prove three things:
- A property and its realizer can both be causes of the same effect, but only under special conditions: the higher-level relation must be realization-sensitive, so the effect would vanish under small changes of realizer.
- The realizer is the cause and the higher-level property is not (upward exclusion) when the effect tracks the particular realizer.
- The higher-level property is the cause and the realizer is not (downward exclusion) exactly when the effect would still follow in some nearest situations where the property is realized differently.
The result could have come out the other way: on Kim’s principle, upward exclusion is guaranteed whenever the realizer is sufficient. The paper’s examples are a pigeon trained on red and Andersen’s monkeys, whose reaching intentions are read from aggregate firing rates realised by many neural patterns.
What this does not say
- Not that higher-level causes are always autonomous. That depends on which worlds count as nearest, which the authors leave context- and discipline-relative. Their claim that the special sciences typically require realization-insensitivity is an observation, not a theorem.
- Not settled by the neuroscience example. Pernu (LIT-147) argues neuroprosthetic decoding supports identifying intentions with neural states rather than their autonomy.
- Not specifically about minds. The framework applies to any levels. Applying it to organizations is the record’s suggestion, not the authors’.
Connections
- Biology and psychology (THEORY-048). Biological findings implement psychological functions without replacing them, which is downward exclusion under this account. No relation is declared, since that theory’s sources do not use the framework.
- Group agency (THEORY-127). A rational group’s judgments supervene on members’ whole sets of judgments without being fixed issue by issue. Whether a group-level attitude is then a cause of what the group does is this account’s question. Its answer turns on whether the outcome would survive a different distribution of members’ judgments realizing the same group attitude.
Derivation (2026-10-20, promotion audit batch 047)
The framework: F is a cause of G iff F is a difference-maker for G, that is F □→ G and ¬F □→ ¬G. Counterfactuals are Lewis–Stalnaker: true iff the consequent holds at every nearest antecedent-world, with strong centering and the limit assumption. At the actual world the realizer P, the property F and the effect G all hold, and P entails F, so every ¬F-world is a ¬P-world.
By centering, F □→ G and P □→ G are true. So F causes G iff ¬F □→ ¬G, and P causes G iff ¬P □→ ¬G.
Result. Suppose F causes G. Then P is not a cause of G iff some nearest ¬P-world is an (F ∧ ¬P ∧ G)-world: F realized differently, G still present.
- If. Such a world is a nearest ¬P-world where G holds, so ¬P □→ ¬G fails.
- Only if. If P is not a cause, some nearest ¬P-world w has G. Suppose w were a ¬F-world. Because the ¬F-worlds are among the ¬P-worlds, no ¬F-world is closer than the nearest ¬P-worlds, so w, at that distance, is a nearest ¬F-world. Then ¬F □→ ¬G puts ¬G at w, a contradiction. So w is an (F ∧ ¬P ∧ G)-world.
Corollary. F and P are both causes iff every nearest ¬P-world lacks G. Which case holds is fixed by which situations are nearest for the system in question, not by the levels.
The title before this note said the effect “would still follow” and did not require F to make a difference. Without ¬F □→ ¬G, G surviving a change of realizer does not make F a cause, and the condition the proof needs is about some nearest ¬P-world, not all.