Title: New grounds for the semantics of counterfactuals

Abstract: The semantic views of counterfactuals that are dominant in philosophy are versions of Stalnaker/Lewis closeness accounts. These accounts make essential use of an ordering of worlds: roughly, counterfactuals quantify over antecedent worlds that are closest to the actual world, according to this ordering. Yet it can be showed that any semantics that relies on a fixed ordering of worlds is inadequate. I explore how to solve these difficulties via an extended analogy with the causal modeling framework developed by Judea Pearl and others. It turns out that a promising semantics for counterfactuals will involve shifts of ordering triggered by the antecedent. This has an important impact on the logic of counterfactuals and seems to shed some new light on the general notion of a counterfactual supposition.