Abstract

The problem I address is: how are cogent inferences possible? In § 1 I distinguish three senses in which we say that one is “compelled” by an inference: automatic, seductive-rhetorical and epistemic compulsion. Cogency is epistemic compulsion: a cogent inference compels us to accept its conclusion, if we accept its premises and we aim at truth. In §§ 2–3 I argue that cogency is intelligible if we consider an inference as a compound linguistic act in which several component acts are related to one another by a commitment that the premises support the conclusion. Non-automatic inferences are primarily public acts in an intersubjective context. But cogency arises in special contexts described in § 4, epistemic contexts, where the participants care for truth, i.e. are intellectually virtuous. An inference is cogent in an epistemic context if it stands up to all the objections raised in the context. In § 5 I consider three different kinds of cogent inferences. In § 6 I argue that in all three cases cogency is fallible and propose a fallibilist variety of inferentialism. In § 7 I distinguish context of utterance and contexts of evaluation. Cogency is relative to epistemic contexts of evaluation. However, validity, i.e. stable cogency, is transcontextual.