Coping Rationally with Unpreferred Preferences

In this paper the "two-selves" models of self-imposed constraints are shown to carry normative implications only when the present-self acts to restrict the freedom of a future-self whose ability to act rationally has been clearly compromised. Conversely, when this future-self, if unrestrained, would act in a way which by any account would have to be called "rational," the normative power of this model is severely compromised. Metapreferences are demonstrated as an approach to such internal conflict that has the advantage of overcoming this normative incoherence. The act of ruling out certain future choices is shown to leave the agent better off if such a prohibition succeeds in altering the agent's preferences in a way that the agent prefers.