Backwards counterfactuals

Abstract This paper offers two novel conceptual tools: one concerning the semantics of counterfactuals and what should be held fixed when assessing them (the modal moat), and the other concerning the pragmatics of counterfactual assertions and how to avoid the potential pitfalls of meaning more than we say (antecedent gluttony). These allow us to address existing issues with the assessment of backwards counterfactuals within a framework that applies equally to forwards cases. In addition to solving a thorny problem from the time travel literature, what we learn teaches us something quite general about our evaluation of counterfactuals.