Higher Level Property Causal Relevance
Anthony Dardis
Causal
relevance is a relation on properties. Causation is (based on) a pattern of
regularities that holds among properties. One property is causally relevant to
another iff the pair has a certain kind of place in that pattern. Mill's Methods
constitute a least common denominator framework for what the pattern is like. P
is causally relevant to Q iff P (as a Mackian inus condition) is necessary and
sufficient for Q. If P1 ... Pn supervene on R1 ... Rm, then Mill's methods can't
separate them. The P1 ... Pn aren't independent of R1 ... Rm, so they are not
complete and independent causes; we could just say all those supervening
properties are causally relevant also. But that would let in way too much.
Instead, order the set of properties by supervenience. Run Mill's Methods on the
first group (the properties which supervene on no other properties) and add the
winners to the causal relevance relation. Run Mill's Method's on the second and
each of the remainder of the groups, and at each stage, add the winners of the
Millian competition to the causal relevance relation.