This chapter advances a contextual view of evidence, through a reconsideration of Hempel's paradox of confirmation (the ‘ravens paradox’). The initial view regarding Hempel's paradox is that a non-black non-raven does confirm ‘All ravens are black’, but only in certain contexts. The chapter begins by reformulating the paradox as a puzzle about how the same entity can have variable evidential values for a given proposition. It then offers a three-stage solution to the reformulated paradox. (1) The situation makes better sense when we reach a deeper propositional understanding of evidence, recognising that each entity can be represented in multiple observational propositions. (2) Some anti-contextualist intuitions can be defused by distinguishing two different senses of the word ‘evidence’, one applying to objects or events and the other applying to propositions; only the latter is relevant to inference. (3) A fuller understanding comes from analysing the constitution and use of evidence in terms of epistemic action. These reflections on the ravens paradox suggest a general philosophical framework more suitable for understanding the function of evidence in scientific and everyday practices.
A history based approximate epistemic action theory for efficient postdictive reasoning