We investigate the notion of relevance as it pertains to ‘commonsense’, subjunctive conditionals. Relevance is taken here
as a relation between a property (such as having a broken wing) and a conditional (such as birds typically fly). Specifically,
we explore a notion of ‘causative’ relevance, distinct from ‘evidential’ relevance found, for example, in probabilistic approaches.
A series of postulates characterising a minimal, parsimonious concept of relevance is developed. Along the way we argue that
no purely logical account of relevance (even at the metalevel) is possible. Finally, and with minimal restrictions, an explicit
definition that agrees with the postulates is given.
, vol. 49, no. 2, pp. 137-173, 1998