What are the normative implications of behavioral economics? This paper proposes a general and simple theoretical framework to answer this question. Although we show that revealed preferences cannot, in general, underpin welfare, we offer conditions validating such a link. We assess the scope of the existing normative criteria used elsewhere and propose a normative criterion based on individual autonomy as a refinement. We use our autonomy criterion to justify a class of public policy interventions we label soft-libertarian, which offers theoretical grounds for empowerment policies. Our theory is falsifiable and the testable implications of behavioral and standard decisions are different. Our approach unites a variety of seemingly disconnected positive behavioral models, and it allows for preferences to be not necessarily complete or transitive and for action sets to be not necessarily convex. We use our model to study the interaction between the decision-maker's initial disadvantages and her capacity to aspire.