Abstract

On Frankfurt's view of free will, in its simplest form, an agent is free just in case her second-order volitions -- those second-order desires she wishes to be effective -- are in accord with her first-order volitions -- those first-order desires that one actually acts upon. That is, an agent has free will just in case she has the desires she wants to have and they are the desires she acts upon. But now consider an agent who lacks free will because her first-order volitions and second-order volitions conflict. Suppose also that, with enough therapy, she could alter her first-order volitions to be in accord with her second-order volitions. So it is true that she can alter her first-order volitions. Still, there is the question of what she should she do. A natural response that a real self theorist might make is that the agent should change her first-order volitions in order to reflect her second-order volitions. But this is not, in principle, the only option available to our agent. Instead of changing her first-order volitions to reflect her second-order volitions, she might also consider simply revising her second-order volitions to fit with her first-order volitions. In fact, since second-order volitions are arguably more reasons responsive, it could turn out for our agent that changing her second-order volitions is actually the most rational option to take. This applies to even the most recent developments of real self, hierarchical, reasons responsive, and even intention embedded accounts of free will. Indeed, there is empirical evidence that people change their conceptions of themselves quite readily in the face of cognitive dissonance of the kind faced by desire conflicts. But this way of attaining free will does not seem to comport with our conception of a free agent. Free will seems too easy to obtain if all that is required is that our second-order volitions are in accord with our first-order volitions, and I see no particularly principled way of arguing that we shouldn't consider changing our second-order volitions instead of changing our first-order volitions. Secondly, this conception of free will is also subject to other counter examples. A person, for instance, may not be able to respect hierarchies due to some personality trait or other, but she might also be the kind of person for whom even were this trait lacking, still would not want to respect hierarchies. Here we have case in which a person's first order volitions and second order volitions coincide, but we do not want to call her free either. Additional evidence that this cannot be the right conception of free will comes from ideas put forth in the literature on practical rationality: I can be in situations in which the only rational action for me to take is to change my desires (MacIntosh), but if we cannot make irrational decisions, and these are understood as decisions to act in ways that go against our current desires, then being in a situation in which I choose to change my current desires should not be possible, thereby challenging any conception of free will that requires the changing of desires, whether first-order or second-order. I conclude, then, that any compatibilist conception of free will of the kind mentioned will be ultimately unsatisfying.