illustrate, if the actor is thinking about shooting his victim, he can still change his mind. However, once the actor has fired the gun, unless he is Superman (“faster than a speeding bullet”), he can no longer prevent the result from occurring or reduce the risk thereof.

It may be objected that our theory of control is overand underinclusive. The actor may not be able to control his reasoning and may be able to control the result. For example, he may not have complete control over whether he will decide to engage in the action – perhaps he will suddenly become distracted – and he may have control over whether the result occurs, as when he can snuff out the fuse that he has lit. But there is an important difference between these cases; this difference lies in the ability for reason alone to affect whether the outcome occurs. In the former case, although the actor may (fortuitously for him) forget his criminal plans, these plans were still subject to revision through practical reasoning; however, in the latter case, although the actor may decide to take another action in order to try to avert the harm, merely changing his mind is not alone sufficient to prevent the harm from occurring.2

This brings us to our third point: the law should not punish an actor on the basis of a prediction that the actor will ultimately engage in criminal conduct. When determining whether an actus reus formulation is sufficient to demonstrate that the actor has acted culpably, the law cannot focus on what the action reveals about what the actor might do; rather, the action itself – what the actor has done – must ground blame and punishment. For the reasons discussed here, we believe that only the last-act formulation punishes the actor for what she has done.

B. INTENTIONS

The starting point along the continuum is with the actor’s criminal intention. In our view, intentions cannot themselves constitute culpable

2T he actor who lights the fuse in order to burn down another’s property or blow up an inhabited building is culpable at the moment he lights it because, even if he changes his mind, he may not be able to snuff out the fuse. He may black out, or be prevented by another, etc. Even Superman commits a culpable act by firing a gun because the sudden appearance of kryptonite may prevent him from stopping the speeding bullet. And, of course, when the lit fuse passes the point that the actor believes puts it completely beyond his power of recall, his culpability is at its zenith.

and hence punishable acts. In the first part of this section, we argue that the formation of an intention is an act. In the second part of this section, we explain why, despite the fact that the formation of an intention is an act, it cannot be a culpable act. Specifically, intentions are difficult to distinguish from fantasies and desires; intentions may involve differing levels of commitment to the goal; the future conditions in which the intention will be acted upon are opaque; and intentions are revocable.

1. Are Intentions Acts?

A preliminary question is whether forming intentions is an action at all – and thus a potentially culpable one – or are intentions merely the states of mind that accompany other acts? There are, of course, many kinds of mental acts that are performed intentionally. Mathematical calculations, silent prayers, and attempts to remember fall into this category. Are intentions about future conduct like this, so that they can be criticized, not just for the traits of character they reveal – as would an involuntary flash of anger at seeing one’s spouse acting flirtatiously around others – but as culpable acts in themselves? Are intentions subject to direct voluntary control in the way that acting on intentions is subject to direct voluntary control?

The view that an intention might itself be a culpable act rests on two assumptions. First, this view assumes that forming an intention alters the world in some way that is material to the criminal law’s concerns. That assumption is surely met, at least under many standard philosophical accounts of intentions. Under those accounts, intentions alter the balance of reasons for the actor. Before he forms the intention, he has reasons A, B, and C in support of doing the act and reasons X, Y, and Z against doing it. After he forms the intention, he has a new reason for doing the act, namely, the intention itself, a reason that makes the act more eligible and hence, if the actor is rational, more likely.3

Second, this view assumes that forming intentions not only changes the world in the way indicated but also is something we do intentionally.

3See Michael E. Bratman, Intentions, Plans, and Practical Reason 80 (1987); Joseph Raz,

Practical Reasons and Norms 65–71 (1975). For a somewhat different view that holds decisions to act to be actions, but not reason-altering actions, see Thomas Pink, The Psychology of Freedom 125–128, 137–165 (1996).

If this assumption is granted, then we can say that forming a culpable intention – an intention to commit a future culpable act – is itself a culpable act.

The argument we are considering here is not that it is the intention itself that is the culpable act but rather the formation of the intention.4 To illustrate, suppose a person has a certain belief-desire set. Ordinarily, this person will form an intention to act on the basis of his evaluation of his beliefs and desires and a decision regarding what is the best course of action to take in light of them.5 In pursuance of this decision, the actor forms an intention, a plan to engage in that course of action. Hence, although intending, by itself, is not an act but rather a mental state, the mental act of deciding what to intend is potentially a culpable act.6

2. Why Intentions Are Not Culpable Acts

Even if intentions are acts, we must still ask whether they are culpable acts about which the criminal law should be concerned. In one respect, it would make the burden of our argument in Chapter 4 with respect to culpable aggressors (CAs) easier were forming the intent to act culpably itself culpable. (We instead located the culpability of CAs in their unleashing the risk of creating fear.) Nevertheless, despite making defense against CAs easier to justify, we believe that there are a number

4See Bratman, supra note 3, at 103 (implicitly acknowledging that forming an intention is an action while discussing the Toxin Puzzle, in which a Genie offers you a million dollars if and when you form the intention to drink a very unpleasant but harmless potion after you have received the money: “[T]he million-dollar reason is a reason for a present action of causing yourself so to intend; this reason is relevant to deliberation about whether so to act now, not to whether to drink the toxin later”). For a different view of the Toxin Puzzle, see Pink, supra note 3, at 137–165.

5See Michael S. Moore, Act and Crime: The Philosophy of Action and Its Implications for Criminal Law 140 (1993) (noting that “[i]n the face of conflict between prima facie desires, there seems to be a resolution when the actor decides which of the alternative courses of action he is going to pursue”).

6T he triggering of this decision need not be an intentional action itself. Otherwise the objection would be that of regress: one must intend to intend, will to will, decide to decide. Cf. id. at 115 (discussing Gilbert Ryle, The Concept of Mind 67 [1949]). Our everyday lives present us with many situations where we make choices without deciding that we should first think about choosing. Thus, a belief, a thought, or desire, none of which is controllable, might trigger the deliberations. Nevertheless, the actor does have control over his deliberations and knows right from wrong at the point at which he decides what he plans to do. If John is confronted with his wife sleeping with his best friend, he might suddenly think that he wants to kill her, but he still controls whatever decision he might make in light of that desire.

of reasons to reject the view that forming the intention to impose risks should be sufficient for criminal liability.

a. Distinguishing Intentions from Desires: The first concern is one of line drawing. Gerald Dworkin and David Blumenfeld make the point that the lines between intending, on the one hand, and fantasizing, wishing, desiring, and wanting, on the other, even if philosophically clear, are quite difficult to draw as a practical matter, even for the actor himself:

This . . . objection has two aspects, the difficulty of the authorities distinguishing between fantasying, wishing, etc. and even more importantly the difficulties the individual would have in identifying the nature of his emotional and mental set. Would we not be constantly worried about the nature of our mental life? Am I only wishing my mother-in- law were dead? Perhaps I have gone further. The resultant guilt would tend to impoverish and stultify the emotional life.7

This objection, of course, has no purchase when we are dealing with completed crimes and completed attempts. Nor does it apply to cases where one solicits or encourages another to commit a crime, which we later argue should count as recklessness toward the victim even in the absence of a purpose to harm if the criteria for recklessness are otherwise present. It does apply forcefully, however, to incomplete attempts, where it is the actor’s attitude toward his own future conduct that is at issue. The actor may have difficulty distinguishing a mere flight of fantasy from a settled intention.

There are two worries here, one epistemic and one normative. The first is that because forming an intention is an act of will, but wishes, desires, and fantasies are not, an actor may fear being punished for something that is not itself an “act,” and most certainly not a culpable act, because he fears the state will not be able to distinguish between the two. The second concern discussed by Dworkin and Blumenfeld is that if actors fear being punished even for these thoughts, they will be constantly policing themselves in a way that diminishes the overall quality of actors’ lives. Although this concern is external to retributive justice, it still presents a significant worry. If in our efforts to reach the culpable, we also significantly infringe upon the freedom of the nonculpable

7Gerald Dworkin and David Blumenfeld, “Punishment for Intentions,” 75 Mind 396, 401 (1966).

(those who have only fantasies and never intentions), then the achievement of retributive justice may come at too great a cost.

b. The Conditionality of Intentions and the Opacity of Future Circumstances:

In almost every conceivable circumstance, when we form an intention to engage in some future conduct, our intention to do so is conditional. That is, we intend to engage in that conduct only under some conditions and not under others. We intend to drive to the store at three o’clock – but only if the roads are not flooded, the store is open today, we do not find that we already possess the items we intended to purchase, and so forth. Some of the conditions may negate the culpability of the act we intend: we intend to shoot Susan – but only if she is about to blow up a bus full of children. Sometimes the conditions may not negate the culpability of the intended act but make the probability that the intended act will occur and also be culpable extremely unlikely: Roger intends to steal the Mona Lisa, but only if it is in the Met (not the Louvre), and only if it is not heavily guarded. Sometimes the conditions are present in the actor’s mind when he forms the intention, so that his conscious intention is to do X on condition Y. Often, however, the conditions are not conscious but nonetheless operative: for example, almost no thief who forms an intent to steal a television set would do so after having inherited a billion dollars, even if that latter possibility never occurred to him.

Now the conditionality of intentions makes assessing their culpability less straightforward than assessing the culpability of the intended acts themselves. As we have argued, the latter are assessed by asking what risks of what harms did the actor believe he was imposing and what reasons did he believe (with what degree of certainty) existed that might justify that risk imposition. But before that act, the actor cannot be certain that he will so act, nor can he be certain about what culpability-enhancing or culpability-mitigating facts will obtain at the time of the act.

So suppose Roger will be culpable at level C if he takes what he believes to be the Mona Lisa from the Met. And suppose that he presently intends to do so, but only if the Mona Lisa is actually in the Met (quite unlikely), only if it is not heavily guarded (quite unlikely), only if he does not inherit a billion dollars tomorrow, and so forth. What is Roger’s culpability now, the time when he forms that conditional intention to steal the Mona Lisa?

Or suppose Maggie intends to drive to her child’s school to pick him up, and she intends to get there in ten minutes. If traffic is moving freely, the road is not icy, and so on, Maggie can get there in ten minutes without recklessly endangering others. And her intention to get there in ten minutes is not unconditional: she would not drive through a Boy Scout parade on the road, mow down pedestrians crossing at crosswalks, and so on. Still, there are some circumstances, quite unlikely but still possible, in which Maggie would drive recklessly to some degree, to which she may or may not be adverting when she forms the intention to get to the school in ten minutes. How should we assess Maggie’s culpability for so intending?

Perhaps there is a solution to this problem of the conditionality of intentions. For example, perhaps we should assess the culpability of the actor’s intention by asking what probability did he attach to his acting on the intention – or, what is the same thing, what probability did he attach to the conditions obtaining that governed whether he would act? (This, of course, will be difficult and perhaps impossible for those conditions to which the actor did not consciously advert but nonetheless were present dispositionally.) Then we could ask what culpability-enhancing and culpability-mitigating conditions did the actor believe would exist at the time of the intended act, and with what probabilities. The culpability of intending the act would be the average of the culpability of the various scenarios the actor envisioned might obtain at the time of the intended act, weighted by the probabilities he assigned each scenario, and discounted by the probability he attached to his actually engaging in the intended act (the probability that the defeating conditions would not exist).

If this “solution” appears impossibly complex – and it is – the problems of assessing the culpability of intending a (possibly) culpable future act are only beginning. For the actor may assess the probabilities differently at different points in time. Roger may, at the time he forms the intention to steal the Mona Lisa, believe that it is quite unlikely that the Met possesses it. Later, he may assess that likelihood as higher (or lower). Similarly, over time he may alter his assessment of the likelihood of its being heavily guarded. Thus, Roger’s estimate of the likelihood that he will ever execute his intention to steal the Mona Lisa will vary over time. Is Roger’s culpability dependent on the point in time at which he is arrested? Could he have been more or less culpable if he had been arrested sooner (or later)? And is culpability assessed just at the moment of arrest, or is it the average of his culpability from the moment he

formed the intention until the moment of arrest, which, too, will vary with the time of the arrest?

What goes for the actor’s estimate of the probability that he will execute the culpable intention also goes for his beliefs about the culpabil- ity-relevant facts that will obtain at the time of the intended act. Those beliefs may change during the time in which the intention is held.

For instance, assume that Heidi decides to blow up a football stadium to protest the allocation of funds to the football team instead of a new law school building. At that point in time, she may not have yet considered whether she will blow up the stadium when it is full of people or when it is empty. She may be debating whether to wait until there is a home team present, or whether to ensure that it is a day when no one is in the area. But even if Heidi has resolved to blow up the stadium at a specific time, her beliefs about the attendant risks and possible justifying reasons may change several times between the time she forms the intent and the time she detonates the bomb: she may believe at one time that the stadium will be full of fans, at another that it will be empty, at another that it will be occupied only by a terrorist cell, and so on. If she were to detonate the bomb at any of these times, her culpability for doing so would vary enormously – from extremely culpable to possibly not culpable at all in the case of the terrorist cell. Again, is Heidi’s culpability determined by her beliefs at the time she is arrested – in which case it may be high if she is arrested at t1 but nonexistent if arrested at t2 – or is it determined by the average culpability of her intention from the time of its formation to the time of her arrest? In either case, because her beliefs about risks and reasons will change over time, her culpability level will depend on when she is arrested. (And how does one “average” a culpable intention, as when Heidi believes the stadium will be full of fans, and a justifiable one, as when she believes the stadium will contain no one but a dangerous terrorist cell?)

Indeed, not only may Heidi’s beliefs about risks and reasons change over time, but she simply may have formed no belief whatsoever about the degree of risk that she will ultimately impose, leaving this “detail” to be filled in later. T he ultimate risks and reasons for the intended action and therefore its culpability will be indeterminate at the moment of forming the intention.

Or take another case where an intention exists at t1, but its justifiability, and thus its culpability, fluctuates over time. Leo looks at his

to kill Sally next January 1. On July 2, Hank has held the intention to kill Sally for over six months, whereas Harry has done so for only a day. Is Hank more culpable than Harry?

i.Duration Does Not Affect Culpability: One view would be that the duration of the intention is immaterial to its culpability. On July 2, Hank and Harry are equally culpable. And Hank is as culpable on January 1 as he is on July 2 (barring changes in beliefs about conditions and circumstances that we discussed in the previous subsection).

But suppose that Hank now renounces his intention to kill Sally, whereas Harry does not. What effect does his renouncing it have on his culpability, and does it matter why he renounces it? (If he renounces the intended act because he views it as imprudent as opposed to immoral, should that deprive his renunciation of any effect on his culpability?)

If forming an intention to commit a culpable act is itself a culpable act, then it is difficult to understand how the culpability that exists at that time can be expunged by later revoking the intention. The past is fixed and contains the culpable act. It cannot be altered. If Hank was culpable on January 1 for intending to kill Sally, he remains culpable for having so intended on January 1 even if he no longer intends to kill her, whether owing to prudence or conscience. Any view to the contrary would suggest that our concern was not with past culpable acts but only with present character.

If the culpability of intending cannot be expunged by revoking the intent, and if the duration of the intention is immaterial, then Harry is just as culpable as Hank even if Harry revokes his intention after thirty seconds, whereas Hank has held his intention firmly for more than six months. Or if Harry is indecisive and forms the intent to kill Sally, then thirty seconds later revokes it, then forms it again, then again quickly revokes it, Harry has committed several culpable acts of intending to kill whereas Hank has committed only one. But it seems odd to deem indecisive Harry more culpable and deserving of more punishment than steadfast Hank.

ii.Duration Does Affect Culpability: Alternatively, we might say that Hank’s holding the intention to kill Sally for six months makes him more culpable than Harry, who has held the intention for a much

shorter time. The problem is explaining why the duration of the intention should matter. We give an explanation in Chapter 7 for why the duration of a perceived risky act matters: it matters because duration affects the degree of risk. (Driving for one minute while intoxicated is less risky than driving for an hour in that condition.) But the six month’s duration of Hank’s intention to kill Sally does not make the risk to Sally higher on July 2 than the risk from Harry’s one day’s intention.

d. The Revocability of Intention: There is a final problem with deeming intentions to be culpable and punishable. As the discussions that both precede and follow this section reveal, this concern is decisive for us. This is the problem that intentions are open to reconsideration and thus revocable. On the one hand, if it is not necessarily rational to reconsider one’s intention,9 the decision to do wrong may be the point at which the balance of reasons has shifted for the actor, and he has committed the culpable – unreasonably dangerous – act. The formation of the intention might then be analogous to the lighting of a long fuse where, although the actor may still exert control over whether the harm does materialize, the risk to the victim has nonetheless increased. One might thus argue that from the actor’s perspective, the risk to the victim has increased because, even from the actor’s perspective, the actor’s balance of reasons has shifted. Thus, if this risk is being imposed for insufficient reasons, the actor has committed a culpable act in forming the intention itself.

However, when one is planning to commit a crime, it will always be rational to reconsider.10 Indeed, many intentions are formed with the proviso that there can always be later reconsideration. Although intentions may serve to guide our future decisions, they are not irreversible, nor may they be carried out without any further effort on our part. The risk from the actor’s point of view – objectively risk is always zero or one – may have increased, because the actor believes he will act as he now intends; but the actor still remains in total control of whether this risk will be unleashed, and he knows this.

9Id. at 64 (“nonreflective (non)reconsideration of a prior intention is the upshot of relevant general habits and propensities”).

10Cf. id. at 67 (“[I]t seems plausible to suppose that it is in the long-run interests of an agent occasionally to reconsider what he is up to, given such opportunities for reflection and given the stakes are high, as long as the resources used in the process of reconsideration are themselves modest”).

Now, it may be said that there is a distinct difference between rationality and culpability. After all, an individual who has lit a fuse will also be under rational pressure to stomp out the fuse, so the pressure to reconsider cannot itself suffice to make the act of forming the intention nonculpable. Even if the actor may be able to change his mind and “revoke” his intention, how does any irrationality in not reconsidering render the initial choice less culpable?

Unlike a lit fuse, however, which the actor may find himself unable to put out even if he now wishes to do so, the actor who only intends to unleash a risk in the future knows that he is still in control of his actions. And, indeed, just as crucially, through the exercise of reason and will alone, he may “stop” the harm from occurring. The common law’s focus on locus penitentiae is best thought of not as an opportunity to abandon or renounce but as an opportunity to continue deliberating and change one’s mind. Just as the criminal law seeks to influence the reasons for which one acts, the criminal law should allow room for deliberation, even including room for the formation and renunciation of an intention. The actor may still deliberate and think the better of his plan.11

Moreover, often when actors intend an act that may in some circumstances be culpable, the actor will not have considered whether he will really go through with it in those circumstances. Or he may have considered it, believes that he will, but in fact he will not. He cannot know until he actually commits the act in the culpability-creating circumstances (or fails to commit it).

In our view, even if there is rational pressure to stick with one’s intention – pressure that in some way increases the risk of harm – there is also rational pressure to abandon one’s criminal plan, pressure that decreases the risk of harm. Hence, rationality cuts in both directions and cannot therefore point to an increase in risk. One may know that forming the intention, or buying the gun, does in some sense make one “more committed” to the crime, but one also knows that one can and should think the better of one’s plan, and this influence serves as

11See also R. A. Duff, Answering for Crime: Responsibility and Liability in the Criminal Law

104 (2007) (“If the state is to treat its citizens as responsible agents who can be guided [who can guide themselves] by reasons, it should be slow to coerce them on the ground that they are likely to commit a wrong if not thus coerced, since that is to treat them as if they will not be guided by the reasons that should dissuade them from such wrongdoing”).