I
'm still working on this line of thought.
In order to have free will, you must be able to distinguish good reasons from bad reasons. If a person cannot do this, we typically treat them as mentally ill in some capacity, and believe that they have significantly diminished capacity for moral responsibility. If, however, someone can distinguish good reasons from bad reasons for acting, then they are responsive to reason, which is where "responsibility" comes from. This is why we praise good action and punish bad action -- we are doing our best to strengthen moral habits by presented additional good reasons to do or not do things. If someone is presented with a two alternatives, and recognizes that one has good reason behind it, and the other bad, and chooses the bad one, all we as observers and citizens can do is attempt to give them (and others seeing their example) other good reasons to make the right choice in the future (via punishment, or talking-to, or rewards).
Rewards and punishments are means to the end of inculcating good moral habits, which are in fact their own reward, leading to a peaceful interior life and harmony with one's fellow man. Most of the time, people have a pretty good sense of when they are acting for good reasons or bad reasons, and manage to convince themselves that some perceived benefit from the bad actions can just this time outweigh the more long-term and abstract benefits of good actions. This is a slippery slope.
This account depends on there being, in fact, good reasons and bad reasons to act. If, as the moral nihilist would put it, there are no good or bad reasons, only manipulations and power convincing you that something is good or bad, then there is no such thing as free will. Free will is the freedom to choose the good without being restricted, addicted, or fooled by the bad. It is not the "freedom" to do anything, for no reason at all.