There is an all too common view according to which we can only be responsible for actions that we have chosen in some sense freely, or autonomously. If the consequences of our actions matter, it is because we were, or should have been, aware of those consequences. On this view, the consequences of our actions make a difference to our responsibility only insofar as we can or could have foreseen those consequences. This is the picture, then, on which free will or autonomy has primacy; responsibility, in turn, is grounded in it.* I want to argue for a different picture, one according to which the phenomenology of responsibility is intrinsically such that the autonomy or freedom of the agent with respect to an action depends in large part on her future estimation of the rightness or wrongness of its consequences. A deliberative decision—at the time it is made—is largely arbitrary; if it were not, deliberation would have been unnecessary. It ceases to look arbitrary only from the standpoint of one’s future self.
There are doubtless many questions to raise with regard to the condition that we must know or be able to know the consequences of our actions in advance. And there are questions, too, about the related issue of our conative attitudes—how deep must our desires go, how “wholeheartedly” need we accept them, to be held responsible for the execution of action on their basis? Something about this sort of approach certainly seems right. What I want to suggest is that what is right about it, however, is not that there is some unified entity, or even some coherent set of volitional attitudes and beliefs that constitutes the agency necessary for responsibility. That is, I want to question both the common libertarian premise about free will and the common compatibilist one. The libertarian often claims that to be responsible, an agent must directly (or indirectly) and indeterministically choose to perform an action. The compatibilist typically looks for something less demanding, namely, that the action be caused by agency-constituting features, such as rational deliberation, endorsement, cohering attitudes, or desires with which we are fully satisfied.
To say that the compatibilist approach is “less demanding”, of course, is not entirely right. It is only less demanding in the sense that it does not require us to rely on the idea that (1) some events can be caused indeterministically, (2) that the indeterminism in physical processes somehow corresponds to an indeterminism in our deliberations, and in such a way (3) that it is primarily the deliberation, and not the physical indetermistic causation that gives rise to the choice of action. This picture is likely too complicated to be true; and, moreover, it relies on assumptions about the physical world that are, given our current state of knowledge, probably false. In any case, the compatibilist has a further argument: that indeterministic decisions are necessarily arbitrary in such a way that agential responsibility is undermined.
To avoid this last problem, compatibilists typically try to fortify our choices with the aforementioned agency-constituting features. But this, too, ends up being a bit too demanding. I think psychology is slowly uncovering the unlikelihood of our actions being caused through such coherent dispositions; but here I want to pose a phenomenological challenge. It requires noticing that questions of responsibility for an action—and so the accompanying questions of whether the agent was free or autonomous in performing the action—come up only in contexts (sometimes, but not always, moral) where the rightness or wrongness of an action is at issue.
Here is an example: I am taking an exam. I am facing a particularly tough problem, where two answers strike me as possibly right, but I am not sure which one is right and don’t have the time to work it out further. I pick one on a hunch. If all goes well, and the answer is right, I may never think about it at all; or, in thinking about it, I might believe that due to my understanding of the material, my hunches tend to be good ones. But what if I end up getting the problem wrong? I will then blame myself for my stupidity, and for making a bad choice where I should have known better. Although my choice was, for all intents and purposes, arbitrary, whether I praise myself or feel guilt over the bad decision I made depends on the outcome. I would suggest that, in more or less complicated ways, the vast majority of our decisions are of exactly this kind.
A more complex example: I am largely set on graduate school. But the thought that it would be nice to have a decent life—one that has some redeeming features other than the occasional published paper—nags at me; law school becomes a competing option. We can sketch out such a conflict in any number of ways. Perhaps I have different beliefs about what would constitute a good life and different desires regarding my future. If I cannot adjudicate between these sets of beliefs and desires, then my resulting decision is—take your pick—as autonomous as it could be (given the circumstances, i.e., a lack of wholehearted identification with one of the options) and at the same time entirely arbitrary: it could have gone either way. To make this more of a challenge to compatibilist approaches: Let us say that I am wholehearted about going to graduate school; something nags at me, though, urging me to apply to law schools as well. This nagging feeling may appear to me as entirely external—as an affect that I do not identify with—but I humor it and apply anyway. If, in the end, I do not get into any decent graduate schools, but do get into a great law school, it is pretty easy to decide what to do.
So what of this situation? On what grounds can we reasonably say that the decision to go to law school was not autonomous? You will perhaps have guessed that I think there are no such grounds. Imagine that, ten years later, I am miserable with the legal world and wish I could go back and remake my decision to get into it in the first place. Why did I apply to those law schools? Why didn’t I do what my heart told me to do? Why didn’t I stick with the attitudes that defined my “real self”? On the other hand, imagine that, although I would have preferred graduate school, I quickly get into law school—I enjoy having definite problems to solve, clear rules to work with, easily demarcated rules of competition, and the high-paid lifestyle that results. And here I might be glad that I acted autonomously, followed my real self, and went to law school instead of indulging in a post-adolescent intellectualist fantasy of academia, however central that fantasy once seemed to who I am. In evaluating my responsibility for my choice, then, I now look at it not from the perspective of a conflicted self making a largely arbitrary decision, but from the standpoint of my current situation, one in which my satisfaction or dissatisfaction with my choice—its rightness or wrongness—has shaped my perception of what my real, autonomous self should have done.
I suggest, then, that the currently dominant structural theories of autonomy are largely false. There is no real self that is responsible for its decisions. The self responsible is, rather, the future self, the self that is a product of situations brought about as consequences of the initial decision. We cannot eliminate arbitrariness from our choices; we can make them autonomous only by looking back on them. One might, of course, object in the following way: this is a merely phenomenological critique. How we see or experience ourselves has, in the end, nothing to do with whether or not our choices are really the products of some coherent real self. To make this objection work, however, one would need to produce a notion of something like a real self, a coherence of our preferences, or identification with a desire that does not rest on an agent’s self-apprehension. It would be a notion of a choosing self that is understood independently of how this self experiences itself. And that such a picture could be coherent is dubious.
* Strawsonian accounts that base responsibility on our reactive attitudes are not immune from these considerations. Those reactive attitudes depend on our estimation of the agent’s intentions in performing the action for which we judge her. So even if we think that judgments of responsibility derive primarily from the framework of reactive attitudes we have toward others, these judgments still get their appropriateness from considerations about the agent's psychological profile.