Determinism is the view that every event that occurs, including human decisions and actions, is necessitated by earlier events and the laws of nature. So, for instance, according to determinism the words and phrases that I use to express myself right now as I write this essay are chosen by me due to mental processes and events that occur as they do because they are determined by prior events and those events were in turn necessitated by even earlier events and so on. Compatibilists believe that determinism is compatible with the existence of human freedom and responsibility. That is, even if all events in the universe are necessitated by earlier events then we can still exhibit freedom in our decisions, choices, and actions and we can still be responsible for them. Libertarians, on the other hand, believe in the existence of human freedom and responsibility, but they believe it is inconsistent with determinism. According to libertarians, if there is human freedom and responsibility, then at least some of the decisions, choices, and actions we commit must not be necessitated by earlier events.

There are different kinds of libertarian theories of human freedom and responsibility. Two of the main varieties are the agent-causal theories and the event-causal theories. According to the agent-causal theories of human freedom and responsibility, free human decisions and actions are caused not by earlier events but by the agent himself, who in acting as the cause of a decision or action acts as an uncaused cause. According to event-causal theories, free human decisions and actions are caused by earlier events but these earlier events are themselves causally indeterminate so that they don’t necessitate our free decisions and actions. Event-causal libertarians reject the idea of there being special agent-causal powers involved in human free decision and action.

Libertarian theories are often criticized on the ground that they provide a conception of human free decision and action in which such decision and action is merely the product of luck and that consequently on these views we cannot make sense of agent control and responsibility for such decisions and actions. This is problematic since free human decisions and actions are the very actions over which we are supposed to exert control and for which we are supposed to be responsible.

In a recent essay Meghan Griffith (2010) has argued that agent causal libertarian theories can be adequately defended against this kind of objection but that event-causal theories cannot be adequately defended against it. In supporting the latter point, she focuses on the event-causal libertarian theory of Robert Kane (1996, 2002, 2007) arguing that his theory does not have the resources to answer the problems presented by the luck argument. I disagree with Griffith’s assessment of Kane’s theory. In this essay, I will explain Kane’s theory and his reply to the luck argument, and I will go on to show why Griffith’s criticisms of Kane’s position fail.

Kane’s Event-Causal Libertarian Theory

According to Robert Kane, many of the decisions and actions for which we are responsible are causally necessitated by our character, by our beliefs and desires. In a famous example that he borrows from Daniel Dennett (1984), Kane talks about Martin Luther’s famous proclamation “Here I stand, I can do no other.” Luther is reported to have said this in making his break from the Roman Catholic Church. Kane says that even if it was literally true that Luther could not have done and/or chosen otherwise than to break off from the Roman Church, Luther could still rightly be said to have acted freely and responsibly in so doing. In other words, even if Martin Luther’s character at the time of his decision necessitated his choice of breaking off from Rome, he could still rightly be viewed as freely making this choice and he could rightly be viewed as responsible for his choice.

At this point one might wonder how it is that Kane is a libertarian if he thinks our decisions can be necessitated by our character and still be free. The answer has to do with the fact that on Kane’s view in order for Martin Luther’s decision to be free and to be one for which he is responsible then Martin Luther must act from a character that he has freely constructed, and in freely constructing his character there must have been times in the earlier stages of Luther’s life when he made decisions which were not causally necessitated and which shaped his character. Kane calls such undetermined, character shaping actions “self-forming actions (SFAs)”. He says that these actions occur in those difficult moments in life when we are torn between competing conceptions of what we want to do or to become. They occur when we are faced with conflicts between such things as duty and self-interest or between the desire for immediate pleasure and later rewards for current sacrifice, etc. When we make decisions in these contexts our decisions are caused by the mental events leading up to our decision, but since the mental events leading up to the decision are not themselves causally determined our ultimate decisions in these contexts are not causally determined.

According to Kane, even if Martin Luther’s decision to break off from the Roman Church was necessitated by his character, he was still freely making that break and he was responsible for that break as long as he engaged in causally undetermined SFAs in the past that led him to have the character he had in later life and that subsequently caused him to make the decision he made. Kane believes that if all of our decisions, choices, and actions were causally necessitated then we would never make free decisions and we would lack responsibility for our choices. Thus, he rejects compatibilism. However, he does believe that human beings often engage in these causally undetermined, character shaping decisions (SFAs) and that when we do so we act freely and responsibly. Further, these freely performed SFAs that we have committed in the past do in turn shape our character and because of this we are responsible for actions necessitated by our character in later life.

The Problem of Luck

It has been argued that Kane’s libertarian theory, like other libertarian theories, suffers from a problem of luck. In subjecting Kane’s theory to the luck argument, critics make much of the fact that according to Kane there is causal indeterminacy in the agent’s mental states in the moments leading right up to a decision. In one of his favorite examples, Kane talks about a business woman who is on her way to an important business meeting when she notices an assault taking place in an alley. Upon seeing this, she is then torn between the desire to (A) go on to her business meeting and the desire to (B) help prevent the assault. He says that if her decision in this situation is an SFA, then her decision must not be causally necessitated—must not be causally determined—by her beliefs and desires. Critics note that, according to Kane, in SFAs there is causal indeterminacy involved right up to the moment of choice; and they exploit this aspect of his view in generating the luck argument against it.

One way in which the luck argument is commonly generated is through the use of possible worlds examples. Possible worlds are different possible ways in which our world, the actual world, might have been or might be different from the way that it is. Some commentators have us suppose that in the actual world the business woman (BW) chooses to help prevent the assault. They note that, according to Kane, if this choice is an SFA then right up to the moment of her choice she could have chosen differently. This means that in another possible world there is another woman, BW*, who faces the same choice and who is just like BW in all respects, possessing an exactly similar past, and who has all the same mental states right up to the moment of choice, and she chooses differently, opting to go on to her business meeting.

Critics point out that given these facts it follows that there is really nothing about the mental states of the business woman in the actual world that explains her choice. Since there is another woman, BW*, in another possible world with the exact same mental states leading up to the moment choice who chooses differently, then there is nothing about the actual business woman that explains why she chooses to help rather than to go on to her meeting. If there is nothing that explains her choosing in this way rather than the other, then her choice is really just a matter of luck, meaning she does not exert the kind of agent control over her choice that is necessary for her choice to be free. Further, since she does not really control her choice and her choice is not free, then her choice is not really a choice for which she is responsible. [My articulation of the problem of luck is here shaped largely by my reading of Haji (2005). But the problem of luck has been presented by various contemporary critics of libertarian theories. See, for instance, Waller (1988), Haji (1999), Mele (1999, 2006), van Inwagen (2000), Strawson (2000), Clarke (2002), and Allen (2005).]

Kane’s Reply to the Luck Argument

To solve the problem presented by this kind of argument, Kane has invoked the concept of dual efforts, or dual willings. The basic idea is that in the context of performing SFAs, where the agent is torn between doing two different acts, A and B, and she cannot do both, the agent actually exerts effort, wills, to do both acts. Further, it is due to the fact that she cannot do both acts but is trying to do both that her ultimate choice of one over the other is causally indeterminate. So, according to Kane, when our business woman is faced with the choice to go on to her business meeting and to help prevent the assault, she actually ends up willing to do both of these things; she wills to go on to her meeting and at the same time she wills to help prevent the assault. As a consequence of this dual willing, it is causally indeterminate as to whether she will go on to the business meeting or help to prevent the assault. In the process of deliberating, one of these willings wins out in the competition between them and this is ultimately reflected by the decision that she makes. So, for instance, if she ends up stopping to prevent the assault then this is a consequence of the fact that her will to prevent the assault has defeated her will to go on to the business meeting. Again, it is important to remember that, according to Kane, in the context of SFAs it is causally indeterminate as to which of two competing efforts will win out.

Here critics will say it is still a matter of luck as to which of her dual efforts wins out. Thus, it is not really up to her what she does in this situation. So, her choice is not really in her control and she is not responsible for it. But Kane disagrees with these conclusions and argues that they just don’t follow from his theory. He draws attention to the fact that in SFAs the agent wills to do both of two incompatible actions. So, for instance, our business woman wills to (A) go on to her business meeting and she wills to (B) stop the assault. He also notes that when we will to do something we do it purposefully and intentionally. Thus, on his view, we are in control over what we will to do. This allows Kane to say that whatever the business woman does—whether A or B—she does what she was willing to do. Thus, whatever she does she will be in control of what she does and she will, thus, be responsible for what she does.

Critics might say that even if she wills to help prevent the assault and ends up doing so she still is not in control of doing so, because there was nothing about her character which determined this choice. She could have just as easily ended up going on to her business meeting. In reply Kane argues that just because what we do is not determined does not mean that we are not in control of nor responsible for what we do. Here he uses important arguments from analogy. One of them proceeds as follows:

A husband, while arguing with his wife, in a fit of rage swings his arm down on her favorite glass-top table top intending to break it. Again, we suppose that some indeterminism in his outgoing neural pathways makes the momentum of his arm indeterminate, so that it is undetermined whether the table will break right up to the moment when it is struck. Whether the husband breaks the table is undetermined and yet he is clearly responsible, if he does break it. (It would be a poor excuse to offer his wife, if he claimed: “Chance did it, not me.” Though indeterminism was involved, chance didn’t do it, he did.) (Kane 2007, 27)

Here the husband is clearly trying to break the table by striking it, but it is causally indeterminate as to whether the table will break. If it does break, the husband is clearly responsible for its breaking. Kane is arguing that if the husband can be responsible for this causally indeterminate breaking of the table then so can persons involved in causally indeterminate decisions made in the context of SFAs.

Again, suppose our business woman chooses to (A) help prevent the assault and suppose her decision to do so is the consequence of a causally indeterminate decision made between doing this or (B) going on to her business meeting. According to Kane, the woman in this situation actually wills, exerts effort, to do both acts, and, consequently, it is causally indeterminate what she ends up doing. He contends that if she ends up choosing A then even though her choice of this was not determined by her prior mental states she is still responsible for this choice because she willed the doing of A. When the husband willfully strikes the table, it is causally indeterminate as to whether the table will break. But in willfully striking it, he knows or should know that it might break and so he is responsible for breaking the table. Analogously, when the woman wills A prior to choosing it, she knows or should know that this may result in her choosing A. Thus, like the husband, she is responsible for doing A when she makes that choice even though the occurrence of that choice is not determined.

For all of the above stated reasons, then, Kane does not feel that his libertarian theory suffers from a problem of luck. The free acts performed in SFAs, which are the basis for all of our freedom and responsibility, involve the dual willing of both acts considered in our deliberations leading up to choice. Because of this we are responsible for whatever we end up choosing to do in these situations, even if our choices are not causally determined.

Griffith’s Critique of Kane’s Theory

Meghan Griffith argues that event-causal libertarian theories do indeed suffer from a problem of luck. In making her case for this she focuses on the views of Kane in particular, and she attacks the argument from analogy presented in the previous section. She writes:

While the agent can be said to cause the “action”, she seems not to have control over the crucial element for which she is responsible; that she has decided to A rather than B. This can be compared to the husband. Although the husband causes the table’s breaking, he does not completely control whether the table breaks. In this sense, its breaking happens to him. He is held responsible for this nonetheless, but that is because he intentionally performed an action (bringing his arm down upon it) that he knew could lead to its breaking. The decision case is different. Although free will theorists often claim that responsibility for a current decision can be traced back to prior decisions, there must be decisions that stop an infinite regress. In fact, it is the regress stopping decisions that must not be determined, according to the event-causal libertarian. (Griffith 2010, 50)

Kane has argued that causally undetermined decisions made in the context of SFAs are decisions for which we are responsible because they are sufficiently akin to the situation of the husband who is responsible for his causally indeterminate breaking of the table. But Griffith is arguing the analogy here is not strong enough to support the conclusion. She thinks there are significant differences between the cases.

In Defense of Kane’s View

One unfortunate aspect of Griffith’s critique of the argument is that it is not entirely clear what she thinks the significant differences between the cases are. In what follows, I will suggest a couple of interpretations of her point and argue that either way her argument is problematic.

Maybe Griffith’s point is that in the table-breaking case the man can be responsible for his causally indeterminate breaking of the table because he was in clear control of his striking of the table and he clearly knew (or should have known) that this might break the table. But in the context of a Kanean event-causal free decision between A or B and where the agent chooses A it’s not clear that the agent is in control of doing anything in the moments leading up to the decision which makes her responsible for it. Thus, Kane’s argument from analogy does not show how agent control and/or responsibility are present in Kanean event-causal free decisions.

In defense of Kane, it can be said that in the moments prior to the choice of A the agent wills the doing of A and she wills the doing of B. We are in control of what we will to do. Our willings don’t just happen to us; we will things intentionally and with a purpose. Thus, in the moments leading up to the choice of A the agent was in control of willing A. Her simultaneous willing of B makes her ultimate choice of A causally indeterminate, but when she chooses A she is still responsible for the choice because she was in control of willing A and had she not willed A she would not have chosen A.

In her defense Griffith might contend that I have not really captured the point of her criticism. She might say that in Kanean free choice the agent decides to A rather than to B. She might say there’s nothing about the agent which explains why she decides to A rather than B. On the Kanean view the agent wills the doing of each, causal indeterminacy ensues, and a decision occurs; but there’s neither an explanation for nor a reason why A is chosen over B. In this sense, the choice of A over B just happens to the agent and is not controlled by her. Griffiths might also add that if the agent does not control the decision to A instead of B then the agent cannot be responsible for this decision. In contrast, in the case of the husband’s breaking of the table there’s no indeterminacy in the choice to strike the table. Rather, in the latter case he simply formulates the intention to break the table and strikes the table with his arm. The indeterminacy lies not in the making of a choice of doing one thing instead of another, but in the attempt to break the table. This is why in the husband’s case we can plainly see that he is responsible for breaking the table. Since there is no indeterminacy in his choice to strike the table, he has clearly chosen to do so and, thus, he is in turn responsible for its breaking.

Perhaps, we are now closer to the point that Griffiths is trying to make, but I believe that even on this interpretation of Griffith’s argument Kane can still defend the view that his event-causal theory of free choice can make sense of agent responsibility. In fact, I think that Kane would and should concede that on his view agent control is significantly diminished in the free choices made in the contexts of SFAs, but that despite this agents involved in these choices are still responsible for the choices they make. Before considering how responsibility is preserved and accounted for on the Kanean view, consider the following passages taken from one of Kane’s more recent writings on the subject:

But can’t we say that it is a “matter of chance” whether one of these efforts leading to SFAs succeeds or not? For isn’t it true that whether or not an effort succeeds in producing a choice depends on whether certain undetermined neurons involved in the agent’s cognitive processing fire or do not fire (perhaps within a given time frame)? And whether these neurons fire or not is by hypothesis undetermined, is it not, and therefore not under the control of the agent? Well, yes, we can say all of these things: whether an effort succeeds does depend upon whether certain undetermined neurons fire or not; and whether these neurons fire is not under the control of the agent; and we can consequently say it is a matter of chance whether the efforts leading to SFAs succeed or not.

But the really astonishing thing is that, while all these things can be truly said, it does not follow that the agent is not responsible for the choice, if the effort succeeds. (Kane 2007, 37)

Here Kane begins to acknowledge that in Kanean event-causal undetermined free choices there is a diminished control over the choice but no loss of responsibility. The point is developed further in the next passage.

But does not the presence of indeterminism or chance at least diminish the control persons have over their choices or actions? ... The answer is yes, again. But the further surprising point worth noting—a point that I think is so often missed—is that diminished control in such circumstances does not entail diminished responsibility when the agents succeed in doing what they are trying to do. (Kane 2007, 38)

Kane would say that when the business woman from our earlier example is torn between the desire to (A) stop and prevent the assault and the desire to (B) go on to her business meeting and when she wills to do both of these things, this makes her choice of doing one or the other indeterminate. Supposing again that she chooses to do A, Kane’s point is that the indeterminacy involved in this choice does, indeed, reduce her control over the choice but it does not reduce her responsibility for stopping to prevent the assault.

Kane believes that to be genuinely self-forming agents (creators of ourselves) there must be times in our lives where we genuinely could have done otherwise than we in fact do and that when we decide and act in these contexts our decisions have an impact in shaping our character. But, he also believes that in these situations agent control over our decisions is diminished without any resulting loss in responsibility. The reason why there is no loss of responsibility despite the diminished control is that in SFAs the agent is in control of willing both A and B. Again, suppose the agent ends up ultimately doing A. Then, since she was in control of willing A and since she knew (or should have known) that she might consequently choose A, she is responsible for choosing A. Furthermore, had the agent chosen B, then, for the same sorts of reasons, she would have been responsible for that choice as well. In this way Kane can argue that in free decision-making between two options, A and B, we have diminished control of choosing one over the other, but we may nonetheless be responsible for our choices in these situations due to our prior willing of both A and B.

More Concerns About Kane’s Views on Control and Responsibility

In the preceding section I argue that Kane would and should maintain that (1) there is diminished agent control in SFAs, but (2) there is no loss of agent responsibility for what the agent ultimately chooses and does in these contexts. In Griffith’s article, right after critiquing Kane’s argument from analogy, she cites a passage from Kane which suggests he would not be willing to give up on agent control in the ways that I have suggested.

One might object and insist that the agent on such views does control what she needs to (i.e., it does not just happen to her). Kane, for instance, claims that “it does not follow that, because one cannot determine which of a set of outcomes will occur before it occurs, one does not determine which of them occurs when it occurs” (1999, p. 238). His claim is that although there is indeterminacy involved in the process of deciding so that the decision is not predetermined, the agent still determines the choice when she decides, thus controlling it. (Griffith 2010, 50)

Here she cites a passage from Kane (1999). A similar point is made in Kane (2007).

Of course, for undetermined SFAs, agents do not control or determine which choice outcome will occur before it occurs. But it does not follow, because one does not control or determine which of a set of outcomes is going to occur before it occurs that one does not control or determine which of them occurs, when it occurs. (Kane 2007, 29–30)

Griffith might argue that my defense of Kane is fundamentally unKanean, because Kane says we do exercise control over our decisions in the context of SFAs. She might additionally contend that my suggestion that we should acknowledge some lack of control is fundamentally problematic. She might say that in the context of an SFA where the agent is torn between the desire to A on the one hand and the desire to B on the other and where the agent chooses A, then the agent must have control of choosing A over B. For, otherwise, the agent cannot be responsible for her choice.

In response to the point that my defense of Kane is unKanean, I would simply note that this is hard to know. Maybe my claims are not in the spirit of what Kane wants to say, but it is hard to know because Kane’s discussion of agent control is not entirely clear. In earlier parts of this paper I cite passages in which Kane maintains we have diminished control over our choices in SFAs. Despite this, Kane clearly thinks there is no diminished responsibility for the choices we make in these situations, and, as Griffith notes, he clearly thinks that in some sense agent control is preserved in these situations. All of this raises questions about precisely in what sense Kane thinks control is diminished and in what sense he thinks control is preserved. Kane is ultimately unclear on these issues. In what follows, I will suggest a way of reading Kane on these issues that is plausible and at the same time renders his position immune to Griffith’s critique.

Ultimately, I think we have to be careful with how we interpret Kane when he says things like:

Of course, for undetermined SFAs, agents do not control or determine which choice outcome will occur before it occurs. But it does not follow, because one does not control or determine which of a set of outcomes is going to occur before it occurs that one does not control or determine which of them occurs, when it occurs. (2007, 29–30)

I want to suggest that when Kane says this, his point is not that in deciding the agent controls his choice of A over or instead of B or vice versa. Notice he doesn’t say this. Rather, he says that in the context of an SFA the agent doesn’t control whether A or B will occur before it occurs but when, say, A does occur the agent controls its occurrence. Notice that in the actual wording of the relevant passages from Kane there is no mention of the agent having control over a decision to A rather than to B. In earlier parts of this paper I cite passages where Kane says we exhibit diminished control in the choices we make in SFAs. But, he still thinks that in an SFA when we ultimately end up doing A and not B, we are in control of doing A. In such a case we are in control of doing A because we willed the doing of it and we wouldn’t be doing it had we not willed it. Thus, Kane might plausibly be interpreted as saying that, even though we do not control the choice of A over or instead of B and even though our doing of A is causally undetermined, we are in control of doing A when we do it and we are thereby responsible for doing it. We might say that, according to Kane, in the context of an SFA where A is chosen over B, the agent has no control over the occurrence of the contrastive fact that A is chosen over or instead of B, but when A occurs the agent is in control its occurrence.

The preceding considerations give us an interpretation of Kane’s view that (1) is consistent with what he actually says and (2) makes sense. But, as noted above, even if I am right about this, Griffith is likely to say that the Kanean position I am defending is still problematic.

In talking about decisions made in the context of SFAs, she says:

While the agent can be said to cause the “action”, she seems not to have control over the crucial element for which she is responsible: that she has decided to A rather than to B. (Griffith 2010, 50, my italics)

She thinks that in the context of an SFA if the agent does A instead of B, then it is “crucial” that the agent have control over the contrastive fact that A is chosen instead of or rather than B. Later she states:

It is already granted that the event-caused libertarian decision is caused by the agent. The problem is that the agent is not able to control which decision is made. The extent to which the agent is causally involved leaves this open.

So the agent in such a picture does disappear to an important extent. And this problem is not constituted by the unavailability of an explanation. It is constituted by a lack of control. It just happens to the agent that A occurs instead of B, since the agent has no involvement beyond those states or events that leave it causally open. So in the end, the event-causal libertarian cannot give the agent the control she needs. (Griffiths 2010, 51)

Notice how she says that on event-causal libertarian views like Kane’s “It just happens to the agent that A occurs instead of B…So, in the end the event-causal libertarian cannot give the agent the control she needs.” Clearly, Griffith thinks that there is something fundamentally important about an agent controlling the relevant contrastive fact that A is chosen instead of B or vice versa. While it is not explicitly stated why this kind of control is so crucial or so needed, it seems clear that she thinks it is needed for responsibility.

However, for reasons that I have already expressed at various points in this essay, this is false; and I think Kane would agree with me on this. Even if an agent does not control the contrastive fact that he does A instead of B, he may still be in control of A-ing when he does A because (1) prior to A-ing he willed to A while he knew (or should have known) that he might consequently end up A-ing and (2) had he not willed to A he would not end up A-ing. Further, this control of his A-ing establishes his responsibility for A-ing.

Conclusion

For all of the above state reasons, Griffith’s critique of Kane’s event-causal libertarian theory is problematic. She does not exhibit any weakness in his argument from analogy to the causally indeterminate breaking of a table top. Further, she mistakenly assumes that in the context of an SFA to be responsible for the doing of either A or B one must be responsible for the contrastive fact that one chooses A instead of B or vice versa. Kane’s view seems to be that, even if an agent doesn’t exert control over such contrastive facts, he can still be in control of A-ing when he does it because of his prior willing of A. In this way, the agent can be responsible for A-ing even if his doing A is causally undetermined.

In concluding, I would note that in her article Griffith argues that agent-causal libertarian views do not have the same problem making sense of agent control that event-causal libertarian views have. In particular, she does not think the agent-causal views fall prey to the luck objection, whereas event-causal views do. In the article, this point is made as though it gives us some significant reason to favor the agent-causal variety over the event-causal variety. If my arguments in this essay are sound, then they may contribute in significant ways to tipping the balance of reason in favor of the event-causal libertarian theory. Assuming for the moment that (1) both the agent-causal view and the event-causal view can provide adequate response to the luck argument and (2) that there is something metaphysically spurious about the concept of agent causation and (3) that other things are equal, we now have reason to prefer the event-causal brand of libertarianism. But, of course, these are some big assumptions, and so, for the moment at least, it would be hasty to conclude that the event-causal variety of libertarianism is superior to the agent-causal variety.