Punishment as Contract

This paper provides a sketch of a contractarian approach to punishment, according to a version of contractarianism one might call “rational contractarianism,” by contrast with the normative contractarianism of John Rawls. Rational contractarianism suggests a model according to which rational agents, with maximal, rather than minimal, knowledge of their life circumstances, would agree to the outlines of a particular social institution or set of social institutions because they view themselves as faring best in such a society governed by such institutions, as compared with a society governed by different institutional schemes available for adoption. Applied to the institution of punishment, a rational contractarian approach maintains that members of society would reach broad agreement with one another concerning the outlines of a system of punishment, based on the fact that they would regard themselves as benefitting from the deterrent effect of such a system. But they would balance the deterrence benefits of such a system with the incursions any scheme of punishment makes into personal liberty. Rational agents would adopt that scheme of punishment that maximizes marginal deterrent benefit without unduly burdening individual liberty. The paper also suggests that a rational contractarian approach is able to capture the best insights of the two leading alternative theories of punishment: deterrence theory and retributivism. On the one hand, rational contractarianism shares the deterrence view that the guiding aim of any punishment scheme must be the deterrence of crime, where a crime is an action that violates the background social contract. On the other hand, rational contractarianism solves the central problem associated with pure deterrence theories—the problem that punishment on this view involves “using” individuals for the sake of achieving the general social goal of deterrence. It does so by maintaining that the way in which the aim of deterrence is incorporated into punishment theory is not