Abstract

Traditionally Kant's theory of punishment has been seen as wholly retributive. Recent Kantian scholarship has interpreted the theory as more moderately retributive: punishment is deterrent in aim, and retributive only in so far as the amount and type of penalty is to be determined by retributive considerations (the ius talionis). But it is arguable that a more coherent Kantian theory of punishment can be developed which makes no appeal to retribution at all: hypothetical contractors would have no good reason to endorse punishment distributed retributively. This position is first sketched behind Rawls's neo‐Kantian ‘veil of ignorance’, and it is suggested that the same theory will emerge from Scanlon's more relaxed neo‐Kantian position.