abstract. The intuitive idea that failed attempts to complete crimes are often themselves crimes belies the complexity and confusion surrounding the adjudication of criminal attempts. This Article offers an account of the grounds for the criminalization of attempt that provides the courts with sorely needed substantive guidance about precisely which kinds of behavior constitute a criminal attempt. The Article focuses on three well-known problems in the adjudication of attempt that have been particularly baffling both to courts and to commentators: specifying the line between solicitation and attempt; determining the conditions under which an “impossible” attempt is still criminal; and identifying the relevance of abandonment to responsibility for and sentencing of attempts. The Article proposes specific doctrinal recommendations for adjudicating all three kinds of attempts; these recommendations are implied by the conceptual framework developed here for thinking about attempt.

author. Professor of Law and Professor of Philosophy & Psychology, Yale Law School. So many people have helped with the development of this project over the years that it is absolutely impossible to thank them all. I start by thanking the authors of the many published critical commentaries on my earlier work on attempt. Their work has benefited the present project immensely: Larry Alexander, Mitch Berman, Michael Bratman, David Brink, Jan Broersen, Antony Duff, David Enoch, Leora Dahan-Katz, Doug Husak, Alfred Mele, Michael Moore, Thomas Nadelhoffer, Steven Sverdlik, and Alec Walen. Thanks are also owed to my former colleagues at the University of Southern California, conversations with whom forced me to figure out what I really think, and to re-think that. Thanks in this respect are especially owed to Steve Finlay, Greg Keating, Andrei Marmor, Jacob Ross, Mark Schroeder, Scott Soames, and Gary Watson. And, finally, thanks to my new colleagues at Yale Law School, especially Bruce Ackerman, Tracey Meares, Robert Post, Jed Rubenfeld, and Scott Shapiro, for encouraging me to seek a broader audience for this work than I am naturally wont to do. In addition, invaluable help was provided by Ben Eidelson. To describe his contribution as that of a “research assistant” is to radically understate the case.

Introduction1

For good reason, attempts to commit crimes are themselves crimes in every mature legal system. A bungled robbery, a missed shot, a beating that fails to kill despite the perpetrator’s best effort, a would-be rape fought off by the intended victim, a smuggling stopped at the border, and many more failed efforts besides possess the marks of wrongful conduct to which the state should respond with criminal penalties. And yet courts and commentators have consistently failed to explicitly offer a coherent theory of this fundamental area of criminal law. Struck by the difficulty of discovering—and the darkness surrounding—principled solutions to adjudicatory problems about attempts, Jerome Hall wrote in 1940:

Whoever has speculated on criminal attempt will agree that the problem is as fascinating as it is intricate. At every least step it intrigues and cajoles; like la belle dame sans merci, when solution seems just within reach, it eludes the zealous pursuer, leaving him to despair ever of enjoying the sweet fruit of discovery.2

Despair no longer. This Article offers a framework for thinking about attempts that solves important problems of adjudication—problems to which we currently lack principled solutions despite the great frequency with which defendants charged with criminal attempts appear in courtrooms.

The simple intuitive appeal of the idea that attempts are to be punished belies the complexity and confusion that surround their adjudication. Some cases are black and white, to be sure, but a startling percentage are not. We have a much less clear idea than we need of what, exactly, we have criminalized in criminalizing attempt. It is therefore often very difficult to tell if a particular defendant has committed a criminal attempt; the courts do not know exactly what they are looking for.3 This confusion manifests itself, for instance, in the many and various descriptions of the conditions that must be met in order for the defendant’s conduct to constitute more than “mere preparation,” several of which are metaphorical (“direct movement towards” completion, for instance).4 But it also comes up in many other places, often in contexts in which the problems seem, at first glance, to be more tractable than courts have actually found them to be. Consider three well-known problems.

First, the problem of specifying the line between solicitation and attempt: Ronald Decker paid Wayne Holston $5000 to kill Decker’s sister. When Holston asked Decker if he was sure this is what he wanted, Decker replied, “I am absolutely, positively, 100 percent sure, that I want to go through with it. I’ve never been so sure of anything in my entire life.”5 Unfortunately for Decker, Holston was not a hitman but an undercover cop.6 Decker clearly solicited murder, a crime for which he could be sentenced for up to nine years in prison in California, where he lived.7 But did Decker attempt murder? If so, he could be sentenced to life in prison. When does asking someone to commit a crime amount to attempting it?

The court in Decker noted that a long string of decisions in California use the term “slight acts” to refer to conduct in furtherance of a criminal intention that suffices for attempt of the intended crime.8 Reasoning that since Decker made a down payment, he engaged in such “slight acts,” the court convicted Decker of attempted murder. But, of course, the question of whether an act is “slight” or less than slight (whatever that might mean) is no easier to answer than the question of whether Decker tried to kill his sister. The justices’ problem was that while they were convinced that Decker tried to kill his sister, they were powerless to explain why that was true, and so they used a bit of entirely uninformative legal terminology to hide their confusion.

Second, the problem of so-called “impossibility”: the defendant in United States v. Crow had multiple conversations in an Internet chatroom with someone going by the name of “StephieFL.”9 During the course of their conversations, StephieFL claimed to be a thirteen-year-old girl. In fact, the messages were written by an undercover (adult) police officer. Crow was charged with attempting sexual exploitation of a minor because he tried to convince StephieFL to send him sexually explicit photographs of herself.10 The completed offense requires a showing that the person exploited is indeed a minor. Did Crow attempt sexual exploitation of a minor, or does the fact that it was an adult he was actually in contact with show that he did not? After all, given that Crow was chatting with an adult, there was no chance at all that his conduct would succeed in sexually exploiting a minor. Under which conditions do the circumstantial elements of the completed crime need to be in place for the attempted crime? And what mental state need the attempter have with respect to those elements when they are absent?

On appeal, Crow noted that the jurors had not been instructed that for guilt they must find that the person Crow was attempting to sexually exploit was in fact a minor.11 As this is an essential element of the completed crime of sexual exploitation of a minor, Crow claimed that it was also an essential element of the attempt, and so the trial verdict could not stand.12 Crow was raising a general question to which an answer is required: do circumstantial elements of completed crimes need to be in place for attempts of those crimes (and if not, why not)? But the court, having no idea how to answer this question, did not even try to give a reason for its answer, simply asserting that Crow’s argument failed.13 The judges’ problem was that they were quite certain that Crow was trying to sexually exploit a child in the sense of relevance to criminal responsibility. What they were ill equipped to explain was how that is consistent with the fact that the only person Crow was trying to sexually exploit, namely the one he was chatting with, was an adult.14 The result is that the court lacked the tools it needed to explain why it decided the case as it did.

Third, the problem of determining the relevance or irrelevance of change of mind to attempt: George Taylor forced his way into the apartment of a stranger and, “threatening her with a knife, he made aggressive sexual advances.”15 The court describes what happened next:

Because of her fear of the knife, [the victim] sought to dissuade him—rather than fighting him or screaming—by “trying to make him believe he could be [her] boyfriend and he did not have to do it this way.” Despite these efforts, he carried her into the bedroom where he continued to touch and rub himself against her and tried to pull down her pants. After [the victim] “told him he could come to [her] house anytime,” he relented and they “went back to the living room and started talking.” He took off the surgical gloves he had been wearing during the attack, saying that he was “not going to be needing these anymore.”16

A bit later, the victim convinced Taylor to accompany her to a liquor store where they could get a bottle before returning to her apartment. On the way out, she ducked back into the apartment and locked the door behind her, leaving Taylor in the hall. Taylor knocked on the door and tried without success to get her to open it. She then called the police.17 Does the fact that Taylor changed his mind matter to the case? Does it relieve him of guilt for attempted rape? Or does it provide a reason for mitigation of sentence? Or neither?

The court in Taylor is in the same lamentable position as the courts in Decker and Crow. Although the judges are confident that in whatever sense Taylor changed his mind it was not the sense that matters to attempt, they have no idea in what sense change of mind does matter. Hiding their confusion with a legal term, they insist that Taylor’s change of mind, while “voluntary,” was not “complete.”18 The court then appeals to a definition of “complete” offered by the Model Penal Code, and adopted in New York, according to which renunciation is incomplete if the defendant chose merely to wait till a later time to commit the crime.19 The problem is that there is no reason at all to think that when Taylor stopped his attack and removed his surgical gloves, he was planning to rape the victim later. He seemed convinced that they would have consensual sex. But the court ignores this glaring fact—perhaps because it can see no other ground on which to reject Taylor’s abandonment defense, and the judges are convinced (with good reason, as will be shown in this Article) that it should be rejected.

Judges in the domain of attempts appear to be behaving in the way that legal realists have for years taken to be endemic to judicial behavior: the judges seem to decide first and rationalize later by appealing to legal concepts like slight acts and incomplete renunciation, or by simply rejecting arguments for no stated reason at all. There is little uniformity across jurisdictions, or even within them, in how courts deal with cases of any of these three sorts. Moreover, even courts that handle such cases in a consistent manner seem to have no idea which principles, if any, support their approach.

But as this Article will show, we need not bow to the powerful impulse to describe what is happening in the law of attempt in the terms of the legal realist. There are valid principles on the basis of which to decide attempt cases; they have simply been overlooked. Quite often judges are cottoning on to those principles, even if they are not articulating them. In fact, as will be suggested here, it is no surprise that judges are catching on, for the relevant principles are entrenched in ordinary moral thought of the sort that informs many commonplace interactions between people outside of the legal domain. Those principles are incorporated into the law, it will be suggested, whenever we proscribe completed conduct.

In ordinary life, and in the law, we implicitly prohibit a set of failed efforts, or tryings, whenever we prohibit various forms of completed conduct. So the question of what criminal attempts are—the question of what complexes of conduct and mental states ought to count as attempts for purposes of criminal law—is the question of what is necessary and sufficient for implicit prohibition of the relevant sort. As we will see, the answer is that we have implicitly prohibited conduct guided by an intention that commits the agent to all of the conditions involved in success. To engage in such conduct is to try in the sense of relevance to criminal law. Exactly what this means will become clear as we move forward. Furthermore, as we will see, this deceptively simple set of principles ought to underlie, and to some extent does underlie, the doctrinal criminal law of attempt.

As noted below, the principles motivating the criminalization of attempts have the following doctrinal implications. First, these principles imply that a solicitation is an attempt if the element of the crime that the defendant asks another to produce is a result element of the crime, but not if it is an act element. Second, they imply that circumstantial elements of completed crimes need not be present for the attempt if the defendant believes or intends that they are in place, but they do need to be in place if the defendant is merely reckless in this respect. Third, they imply that while mitigation of sentence is sometimes warranted on grounds of abandonment, it is never appropriate to grant an affirmative defense of abandonment.

To adopt these three doctrinal recommendations would not only provide much greater uniformity across jurisdictions (and within them) in the adjudication of attempt. It would also bring our law in line with the fundamental principles in light of which attempts deserve to be crimes in the first place. Demonstrating these claims is this Article’s fundamental aim.

Part I identifies the simple and intuitive grounds for criminalizing attempts. It also argues that neither of two ordinary notions of what it is to try to act can be the sense of relevance to criminal law, given the grounds for the criminalization of attempt. Parts II and III offer an alternative account of the nature of attempt of relevance to criminal law—an account that is motivated by the observations offered in Part I. Part II offers a description of the nature of intention, which is the central component of attempt. The description of intention builds on and develops recent work in the philosophy of action. Part III uses this account of intention to explain what it is to try in the sense of relevance to the criminal law. Parts I, II, and III therefore comprise the fundamental theory of criminal attempt offered in this Article. Parts IV, V, and VI use that theory to solve the problems of solicitation-as-attempt, impossibility, and change of mind, respectively. Thus, the Article first provides a conceptual framework for thinking about attempt and then explains how that framework helps with the construction of justifiable doctrine.

I. the source of an attempt’s criminality: the criminality of completion

A. The Transfer Principle

The confusion in the courts can potentially be remedied by an account of what, exactly, the crime of attempt is. If we know exactly what it is to attempt, then we can check to see whether someone like Decker or Crow or Taylor did indeed attempt the crimes they were charged with attempting. Theorists of attempt who have tried to give such accounts are typically divided between the “subjectivists” and the “objectivists.”20 Struck by the fact that attempts are often harmless—the bullet misses, the child is not abducted, no drugs cross the border—subjectivists conclude that it must be that attempts are properly punished thanks to their mens rea elements. From this point of view, attempts are thought crimes. The fundamental challenge for subjectivists, then, is to explain why it is not monstrous for a liberal society to punish attempts. The task of meeting this challenge is typically undertaken by offering an explanation for how the mental states involved in attempt differ from other thoughts that it would be monstrous to punish (for example, they involve resolute intention of a sort that is not idle, but is manifested in action).

Objectivists, by contrast, start with the thought that if attempts were thought crimes, then it would be monstrous to punish them, and so it must be that attempts are properly punished thanks to the conduct that they involve; the emphasis is on actus reusrather than mens rea. The challenge for objectivists is therefore to explain what it is about the conduct involved in attempt thanks to which it is punishable despite its harmlessness (for instance, it risks harm, is “proximate” to harm, or would result in harm if not prevented).

Neither subjectivists nor objectivists have taken seriously the idea that to attempt is to try. Trying, like all the forms of action that statutes criminalize, consists of mental states and conduct. So the fact that trying involves these two parts does not distinguish trying from any other form of action. Neither subjectivists nor objectivists have tried to explain the criminalization of attempt through appeal to what distinguishes trying from other forms of conduct. This simple fact suggests a middle way worthy of exploration. Perhaps attempts are crimes because of the peculiar thing they are, namely tryings, and not because they involve something else (bad thoughts, bad conduct) that there are independent grounds for criminalizing.

As a first step toward developing this admittedly abstract idea, consider something important that we find in ordinary morality. The father tells the child not to jump on the sofa; no ice cream if she does. Moments later, the child is charging toward the sofa with the intention of jumping on it. The father stops her and says, “That’s just what I told you not to do!” Imagine that the precocious child replies, “No, you told me not to jump on the sofa, you didn’t tell me not to try to jump on the sofa. But all I managed to do was try.” She speaks the simple truth. But she still would deserve to lose out on ice cream. In promising to penalize completion, we also, just like that, promise to penalize attempt. And so it is in the law: typically, criminalization of attempts is accomplished automatically through the criminalization of completion.21 There is an ordinary notion of trying that we take to be worthy of censure by the state whenever completion is worthy of such censure. It can seem as though we need not saywhat it is to attempt, for all that needs to be said is said already in describing completion. In this respect, the logic of the law mirrors the logic of everyday morality.

The point can be made in a different way. Imagine that you are asked why it is a crime to attempt murder. In answer you will cite those features of completed murder that make it worthy of criminalization. Chances are, you won’t say a word about attempt at all. What this implies is that we take the features of a form of completed action in virtue of which it is properly considered a crime to somehow transfer to the attempt. We criminalize attempts under the following principle, which I will call “The Transfer Principle”: if a form of conduct is legitimately criminalized, then so are attempts to engage in that form of conduct.

Under the Transfer Principle, the criminality of the completed crime spreads only to attempts to perform that crime. The criminality of attempted battery derives from the criminality of battery and not from the criminality of, say, theft. The Transfer Principle supports criminalizing an attempt, then, only if a description that applies to the attempt—words that correctly describe what is attempted—is also an apt description of a kind of conduct that is legitimately criminalized thanks to the fact that it meets that description.

So the Transfer Principle has an important implication, which tells us where we should start in thinking about attempt. Determining what is properly criminalized as an attempt requires determining what ordinary notion of trying is implicated in the Transfer Principle. In what sense of the term “try” is trying to jump on the couch implicated in the proscription against jumping on it? Or, to put the question in the legal context, in what sense of the term “try” is trying to commit a crime implicated in the proscription against committing it? In short, we need to know the necessary and sufficient conditions for trying in the way that inherits criminality from that which one is trying to do.

B. The Wide and Narrow Senses of “Try”

We might think this problem is easy to solve: just appeal to our ordinary notion of trying. We could then assess whether particular legal doctrines concerning attempt sort defendants as they ought to do, by seeing whether our ordinary notion of trying sorts them in the same way. However, things are not so simple. There are several “ordinary” notions of trying. In fact, a quick glance at ordinary usage of the term “try” suggests that there are at least two senses of the ordinary term that are different from the sense of relevance to the law. And this leaves us wondering in what ordinary sense trying is implicitly prohibited whenever we prohibit completion. We can see that these two ordinary conceptions of trying are inadequate to our task by seeing that trying, in those conceptions, sits uneasily with the Transfer Principle.

Under one ordinary usage of the term “try,” what might be called the “wide” sense of the term, anything that would be true of the person’s act were he to succeed in doing as he intends is part of what he is trying to do.22 We particularly find this usage in cases of mistake. But to see this, start with an example that does not involve mistake. Consider someone who is paid a sum to carry a pound of white powder into the United States from abroad. He is quite certain that the powder is cocaine, but he is not motivated by that fact. So long as he gets paid, he does not care if the stuff is cocaine, or heroin, or talcum. Is this person trying to smuggle cocaine? Or to put the same question another way, does the phrase “to smuggle cocaine” correctly describe what this person is trying to do? In the wide sense of “try,” this is an attempted cocaine smuggling not because of the person’s state of knowledge, but because the stuff is actually cocaine. It is an attempt to smuggle cocaine because if the intended smuggling had come to pass, it would have been a smuggling of cocaine; and this counterfactual is true because the stuff is actually cocaine. In this respect, the wide sense of trying yields a result that is appropriate to the criminal law. Conduct like that of this hypothetical person is of the sort implicitly prohibited as an attempt to smuggle cocaine when cocaine smuggling is prohibited; the Transfer Principle applies, and the wide sense of trying supports that result. So far so good.

But the wide sense of “try” does not capture what we are after, as we can see from considering other kinds of cases, particularly those involving mistakes. In the wide sense, a person who tries to take a suitcase that he reasonably but falsely believes to be his own has attempted theft. Imagine, for instance, that the suitcase looks exactly like his and happens to have his luggage tag, with his name on it, attached to it. He reaches for the suitcase, acting on an intention to take it, and is stopped by the suitcase’s true owner, who explains that the luggage tags were switched by mistake by the airline employees. It’s a simple mistake. Is it an attempted theft in the sense of relevance to criminal law? Of course not. But it counts as such under the wide sense of “try,” since the following is true: were this person to have succeeded in taking the suitcase, he would have taken something that was not his own.

The problem here is not only conflict with intuition about what should be criminal. We can see this in part by noticing that, had the person in the example just given been charged with the crime of attempted theft, he would have been able to cite the fact that he did not know the suitcase was not his in his defense, and such a defense would have succeeded. Such a person would never, for this reason, be convicted of attempted theft. The problem is that under the wide sense of trying, such a person would be speaking falsely were he to say, in his defense, “I wasn’t trying to take someone else’s property.” A person can respond to an accusation either by showing that he did not do what he is accused of having done, or by showing that, although he did it, he is not rightly punished for it (because it is justified, for instance, or because he was insane when he did it). In the example just described, the first sort of response is appropriate, but it is denied to the defendant under the wide sense of trying; under the wide sense of trying, the defendant was trying to take someone else’s property. The problem is that such a person ought not to be understood as needing to account for his behavior in light of the fact that completed theft is a crime; he did not attempt theft in the sense in which the criminality of completed theft transfers, and so he can admit that those who attempt theft need to account for their behavior and simply deny that he is among them. Theft is criminal for reasons that do not transfer to all wide attempts, and so the wide sense of attempt is not the one that informs the Transfer Principle, or the ordinary practice that shows the Transfer Principle to be implicit in everyday thought.

Under an alternative “narrow” conception of trying to act, also found in ordinary language, what a person is trying to do is determined entirely by the set of conditions that he is committed, by his intention, to promoting. The narrow conception has its appeal. Under it, the person who thinks the bag he tries to take is his is not attempting theft, since he is in no sense committed by his intention to making it more likely that the bag that he takes is not his; in fact, he would have held back from trying to take the bag had he known it belonged to someone else. But the narrow conception implies that the smuggler of white powder who believes the stuff is cocaine, but does not care, is not attempting to bring cocaine into the United States. After all, he is not committed to making it more likely that cocaine should be smuggled. Were someone to convince him that the stuff is talcum, he would still carry it because he is being paid to carry it, no matter what it is. Something has gone wrong. The narrow conception of trying to act cannot be the sense that informs criminal law; it provides too stringent a standard for attempt. Put in the language of the Transfer Principle, the criminality of the completed acts in cases of this sort transfers to the corresponding attempts; the attempts were criminalized implicitly when we criminalized completion. But under the narrow conception of trying, these are not attempts of the relevant completed crime. Still, there is some ordinary sense of trying under which criminality does transfer in these cases. What follows is that the narrow conception is not the sense of “trying” under which attempt is criminalized.

C. Taking Stock

We are seeking an account of the kind of trying that is of relevance to the criminal law of attempts. Our first step was to discover the Transfer Principle, which arises naturally from the observation that in ordinary life the prohibition of completion brings with it, intrinsically, the prohibition of some well-defined, although not explicitly defined, class of failed efforts to engage in the prohibited act. The same is true in the criminal law. Criminal attempts are all and only those failed efforts that are implicitly criminalized when we criminalize completion. We then started to look for the necessary and sufficient conditions for membership in that class. We found that neither the “wide” nor the “narrow” sense of trying provided us with an adequate account of those conditions. This leaves us, then, with a question: if it is neither the “wide” nor the “narrow” sense, what sense of trying is of relevance to the criminal law? The answer will be offered in Part III of this Article.

Before we provide the answer, we need to take a detour into the study of the nature of intention. To see the need for it, notice that both the “wide” and the “narrow” senses of trying take intention to be crucial to trying. Part of what it is to try is to intend. The two senses of trying grant different roles to intention in determining what a person is trying to do, but, still, intention is in both cases crucial to trying. In fact, this is no accident, for in any ordinary sense of trying, intention is essential. To identify a third ordinary sense of trying, we need to identify a third possible role for intention to play in establishing what a person is trying to do. But to do that, we need to know more about what an intention is. So, progress on the problem of specifying the kind of trying that inherits its criminality from success—the kind of trying that matters to criminal law—starts with reflection on the nature of intention.

II. intention

A. The Rationally Constituted Nature of Intention

Intention is a distinctive state of mind. Intending an event is different from believing it will occur, for instance. Most who believe that the sun will rise tomorrow do not intend it to rise. Intending an event also differs from wanting it to occur. Someone who wants to eat the chocolate cake for dessert may nonetheless intend to have no dessert at all. Intending differs also from wishing, hoping, and anticipating. This is not to say that intending bears no relationship to these other attitudes. Typically, for instance, those who intend particular events also believe that those events might occur. There is at least this close connection between intention and belief, and there will be similar connections of many different sorts between intention and desire, intention and wish, and so on. But, still, intentions are different from these other mental states.

It is a project in philosophy of mind to determine how intention differs from these other mental states. The project is to specify the necessary and sufficient conditions that must be met for a person to intend a particular event or condition, including but not limited to the event of performing a particular action.23 It is a closely related project in the philosophy of law to identify the necessary and sufficient conditions that must be met for a person to intend a particular event or condition in the sense of relevance to intentional torts, or in the sense of relevance to formation in contract, or in the sense of relevance to intentional discrimination, or in the sense of relevance to criminal responsibility. However, it is a hypothesis worth exploring that all of the various senses of the legal term “intention” circle around a core notion that has its natural home in ordinary discourse.24 The law is better when it uses ordinary terms in ordinary ways and employs concepts that bear a close resemblance to those used in everyday life by citizens asked to conform their behavior to law. It is the core notion of intention that has been investigated by philosophers of mind and action. If indeed such a core notion informs the law, then it is important for those interested in criminal responsibility to understand what philosophers of mind have discovered about the nature of intention.

As a first step, consider the following example. A shopper goes to the store, equipped with a list of things to buy. He has a limited budget, so while he needs everything on his list, he also cannot buy more than is on his list. He walks around filling his cart. Meanwhile, a spy follows him and writes down everything that the shopper puts in the cart. At the end of the trip, both the shopper and the spy have a list that matches the world: both of their lists correspond to the contents of the cart. These two lists, however, had very different functions. The shopper’s list functioned to make the world match it; the spy’s list functioned to match the world as it came to be. Were a third party to remove eggs from the cart, the fact that the shopper’s list includes eggs ought to lead him to fix the situation by putting more eggs in his cart; he should change the world to match his list. By contrast, the fact that the spy’s list includes eggs (he wrote that down before the third party removed them) ought to lead him to cross it off his list; he ought to change the list to match the world. The shopper’s list is like an intention, the spy’s like a belief.25 Intrinsically, both consist in nothing but a representation of the world. But they have different functions.

What this example suggests is that the right way to inquire about how intention differs from other mental states is to reflect on the distinctive functional role of intention. What are intentions for? What do they help us to do that other mental states, such as beliefs, desires, hopes, or wishes, do not help us to do? We should expect that together with an account of the functional role of intention will come an account of principles of rationality governing those who have intentions. If we know what intentions are for, we will also have some idea of what kinds of things a rational agent who has an intention will do, or what kinds of other attitudes, including beliefs and other intentions, he will have. This matters for our purposes here, recall, because we are seeking an account of the notion of trying that is implicitly criminalized when we criminalize completed conduct like murder or rape. Since trying necessarily involves intending, the task of understanding attempt law and describing how it ought to be structured necessarily requires an account of what it is to intend. Therefore, we need to know what distinctive roles intentions play in guiding reasoning behavior in contrast to other mental states that are not essential ingredients of trying.

Michael Bratman’s important work on intention, which he began publishing in the 1980s, provides a great deal of insight about the functional role of intention and the associated norms of rationality that govern those who intend.26 Under what Bratman calls “the planning theory of intention,” intention’s function is to make the world as intended and to make that happen in a way that allows agents to efficiently achieve long-term goals.27 So, for instance, it is part of an intention’s role to provide for coordination between one’s self at one time and one’s self at another. The person who intends in the morning to cook spaghetti for dinner will not succeed in doing as he intends unless his midday self helps by stopping at the store for spaghetti. His intention prompts such help by leading the midday self to do just that.

In this example, the intention plays its role by prompting the formation of intentions to undertake necessary means, but often an intention plays its role by instead preventing the agent from acting in a particular way. Deliberation, for instance, is a costly activity. When deciding this morning what to have for dinner, the agent has to focus his energies on thinking that through instead of doing a variety of other things, and he might have to collect information, such as information about who will be joining him for dinner or about what time he will be free to start cooking. This expensive deliberative process culminates in the formation of the intention to have spaghetti. The intention, in turn, functions to cut off later deliberation about what to have for dinner in the absence of new information. It is because the intention settles the question of what to have for dinner that at midday the agent heads straight for the spaghetti aisle rather than rethinking the question of what to buy at the store.

Plans play various roles in making possible and effective organized behavior that takes place over extended periods of time. Reflection on these roles leads to the articulation of several norms of rationality that govern those with intentions. If an agent is rational, then nothing about him will defeat his intentions from functioning as they ought or performing their distinctive roles. A fully rational agent who intends to do something, for instance, will not intend to do acts incompatible with completing necessary means to doing as intended. A fully rational agent who intends to do something will not at the same time believe with certainty that he will fail to do as intended. We can argue over the details of how to formulate the relevant norms of rationality governing those who intend. But our purposes here will not require settling such arguments. Intrinsically, an intention is just like any other mental state that depicts a future state of the world, such as a future state of one’s body. What distinguishes intentions from other such representations are the norms of rationality under which a person labors thanks to what the intention is for. An intention places an agent under rational pressures; he must conform to certain requirements of coherence among his intentions, other mental states, and conduct on pain of irrationality.

The intention’s primary function is to make the world as it depicts the world, so as to further the agent’s goals. If the intention is to play that role successfully, then the agent must not defeat it by constructing conflicting plans or by failing to undertake means to its fulfillment. There are many ways to fail to live up to one’s intentions. The basic insight, however, is that many such failures amount to falling short of certain norms of rationality—norms that apply to the agent because he has an intention to act in a particular way.

B. Intention-Based Commitments and Responsibility

There are two interrelated and important points about the insight that intention is characterized and distinguished from other mental states by the norms of rationality that govern those who have intentions. First, the position sits comfortably with a particular sense in which to intend something is to be committed to it. In what is, perhaps, the paradigm case, to be committed to something is not just to have a reason to promote it; that reason must be special to oneself and, in many cases, will derive from one’s own will. We all have a reason to promote world peace, but only some of us are committed to it. Those who are have a reason to promote it that derives from the fact that they have chosen to promote it or have devoted themselves to its promotion. Having such a reason involves, undoubtedly among other things, rational pressure to deliberate in certain distinctive ways. A rational person, committed to world peace, is under pressure at least to consider the fact that a particular company provides support to an ongoing war when deciding whether to buy that company’s products. Furthermore, such a person is under pressure at least to grant some weight to that consideration when deliberating about what to do.

In fact, these are two forms of rational pressure that we are under when we intend to act. A cashier who intends to steal from his employer faces distinctive rational pressure with respect to both the facts he considers and the weight to give those facts when he deliberates about what to do. In considering whether to under-report the day’s sales, he ought, rationally, to consider the fact that by doing so he increases his chances of stealing from his employer. He may decide not to do it—maybe he has another, better plan for stealing—but, still, he is not fully rational if he does not grant that consideration some weight in his deliberations (provided that it occurs to him that it will help him to commit the intended theft, and provided that he does not give up his intention). In short, to intend is to structure one’s own rationality. It is to generate reasons for oneself—reasons that do not apply to others. And this is the key to understanding the sense in which agents are committed to that which they intend. The cashier’s intention commits him to stealing from his employer; what that means is that he has special self-generated reasons to structure his practical reasoning around stealing. The view of intention under discussion here, that is, has provided more precision to the idea that our intentions commit us to what they depict.

The role of intentions in constituting commitments explains, also, why intention is of such paramount importance to culpability and criminal responsibility. What a person intends tells us a great deal about what kinds of considerations he recognizes as giving him reason, and about how he weighs those considerations in his deliberations about what to do. In fact, it is in part constitutive of those facts. Someone who intends to steal from his employer takes the fact that the contents of the cash drawer are not his as either providing him with no reason not to take those contents, or as providing a reason of insufficient significance to outweigh considerations in favor of stealing. These facts about the person’s modes of recognition and response to reasons are of crucial importance to assessing his responsibility. It is partly because of those facts that he deserves censure for the act of taking what is not his; they sit at the root of his culpability. He deserves censure not merely because his employer suffers at his hand—although that is, of course, significant—but also because he has misused, misdirected, his capacities for the recognition and response to reasons, capacities that are distinctive of moral agents.28

Intention, norms of rationality, commitment, and modes of recognition and response to reasons constitute a family of intertwined notions bearing deep and important relations to culpability and criminal responsibility. These notions are at the heart of what is distinctive about human agency. They are at the heart of what we respond to in others when we judge them to be blameworthy for wrongdoing, both in and out of criminal law contexts. Our resentment and outrage when confronted with another’s wrongdoing, not to mention our guilt and remorse when confronted with our own, is a response to corruption in the way the actor recognized and responded to reasons and thereby guided his conduct. Part of what we are outraged about is that the actor does not care that the property he took was not his, or cared insufficiently about the fact that another would be harmed by his act, or cared too much about lining his own pockets. The actor’s intentions and commitments are of particular importance because they are inextricably connected with modes of recognition and response to reasons, but also because there is a meaningful sense in which modes of recognition and response to reasons that have their source in intention and commitment are self-inflicted; they have their source in the agent’s will.

C. Broadening Our Perspective on Intention-Based Commitments

The theory of the nature of criminal attempt being developed in this Article began with the appealing idea that attempts are those failures that are implicitly criminalized when we criminalize completion; they are those that support the Transfer Principle. The observation that all forms of trying include intention launched an account of the nature of intention. This account, derived from Michael Bratman’s work, was shown to have an important, albeit abstract, implication for our understanding of criminal responsibility: intentions constitute commitments to the conditions they depict by generating special reasons for the intending agent to structure his practical reasoning around those conditions. Because the intending killer’s intention depicts another’s death, he is under rational pressure to ignore options incompatible with the other’s death, and to form intentions to take means, among other rational pressures. This tells us something of great significance to the assessment of his criminal responsibility, for it tells us how he employs and directs his distinctive human capacity for self-consciously recognizing and responding to reasons.

But this is not all that needs to be said about the nature of intention if we are to develop an adequate account of what it is to try to act—an account that ought to, and sometimes does, inform the law governing criminal attempt. We cannot just take what philosophers of action have said and apply it; we need to go beyond what has been discerned about the nature of intention by Bratman; we need to make further philosophical progress. In particular, more needs to be said about the range of commitments that are constituted by our intentions. Reflection on the issue demonstrates that there are at least three different kinds of commitment that one can have to an event, such as an element of a crime, thanks to the fact that one has a particular intention. As we will see, all three kinds of commitment turn out to be important to the proper adjudication of criminal attempts.

In the typical case—so typical, in fact, as to blind us to the existence of atypical cases—someone who intends a particular condition is thereby committed to increasing the likelihood that the world should be in that condition. The effort to live up to this commitment often manifests itself in “tracking” behavior. To engage in “tracking” behavior is to respond to obstacles to the realization of an intended condition by taking steps that sidestep or weaken the obstacles’ effect. So, for instance, a person who intends to prevent another from leaving a room will not just lock the door, but will also respond, if he can, when the prisoner picks the lock; perhaps he will then throw the deadbolt, or push the prisoner back into the room, or call in reinforcements, or take some other act that will correct for the facts that are defeating or threatening to defeat the world from matching his intention. In fact, he is under rational pressure to “track” the condition in this way: to fail to do so, absent some further reason, would be irrational. Such rational pressures constitute the commitment to the condition. Since the pressures in question are pressures to take steps to improve the chances of the condition’s coming to be, call this a commitment of promotion.

It is very tempting to say that all intention-based commitments to conditions are commitments of promotion. In fact, virtually everyone who has written about intention has assumed this, usually implicitly.29 Bratman, for instance, insists that among the distinctive norms governing those who intend are norms of “means-end coherence” that place intending agents under normative pressure to intend acts that they believe to be necessary to fulfill their intentions.30 And, to be sure, there are such pressures wherever there are intention-based commitments of promotion; rationality requires us to intend necessary means to an event’s occurrence whenever we are committed to increasing the likelihood of that event’s coming to pass. In assuming that norms of means-end coherence apply to intending agents no matter what they intend, Bratman is assuming that every time a person intends an event he thereby incurs a commitment of promotion with respect to that event.

But the assumption is false. Say that I intend to go running at 9:00 AM. As I’m about to start running, I notice a clock that says it’s 8:00 AM; I had forgotten that the time changed as a result of the end of daylight saving time. I go running anyway, and I’m done by 8:30. Did I do what I intended to do? Well, I went running, as intended. But I didn’t go running at 9:00 AM, which was part of what I intended. On the one hand, then, it does seem that I did all that I was committed by my intention to promoting; there’s no real sense in which I fell short of my goals. On the other hand, however, there is still a meaningful sense in which the world did not match my intention; to fully match my intention, I would have had to run at 9:00 AM, not 8:00. The result: included in the content of my intention was the condition that I run at 9:00 AM. But despite its inclusion in the content of my intention, that condition is not one that I incurred any commitment to promote.

Who cares? As will be demonstrated below, we should care; that is, we who care about criminal responsibility and the mental states required for it should care about this. But a first step to seeing why this matters is to see that our intentions commit us, in a sense now to be described, to intended conditions even when we lack commitments to promote those conditions.

Say that when I am getting ready to run, I see that it is, indeed, 9:00 AM. But I change my mind and decide not to run. When asked why I changed my mind, I say, “Well, it’s 9:00 AM.” Something has gone wrong. I can reconsider my intention for many good reasons: I remember a pressing 9:15 appointment, the hills are steeper than expected, it starts to rain, there are too many dogs around, etc. If I did not intend to run at 9:00 AM, I could even reconsider in light of the fact that it is 9:00 AM—maybe I think it is too late in the morning to run. What I cannot do in full rationality is reconsider for that reason given that that was part of what I intended in the first place. The fact that the condition is included in the content of my intention, then, places me under a very particular form of rational pressure: it places me under pressure not to reconsider the intention in light of the fact that the condition is met. Call this a commitment of non-reconsideration. As the example illustrates, it is possible to have an intention-based commitment of non-reconsideration with respect to a particular condition without having a commitment to promote the condition.

There is yet another kind of intention-based commitment that one can have to a condition in the absence of a commitment of promotion with respect to the condition. Imagine that, acting on my intention to go running at 9:00 AM, I go running and so I fail to call a friend whom I promised to call at 9:00 AM. The friend complains. I respond, “You can’t blame me that it was 9:00 AM when I was running. I would have been happy to run at 8:00.” In making this remark, my goal is to show my friend that the condition thanks to which I failed her—namely that I was running at 9:00 rather than 8:00—was not something that I was committed to promoting. The claim is that since I was not committed to promoting that fact, it is not something in light of which I can be held responsible. But, given that I intended to run at 9:00 AM, my friend will see my remarks, quite rightly, as providing no reason to temper her censure. Given that I intended to run at 9:00 AM, I can’t shield myself from responsibility by pointing out that I was not committed to promoting that fact. At least, I cannot do so rationally. The condition that accounts for my harming my friend is also one that matches the content of my intention, and so it is not an unintended condition for which I bear no responsibility. Thanks to my intention, that is, I have incurred a particular commitment to the condition that it is 9:00 AM when I run. Call this a commitment of non-complaint. A person cannot rationally complain that the world turns out the way he intended it to be. More precisely, a very particular kind of complaint is silenced, namely the complaint that might be expressed by saying, “That’s not what I intended.” Even if it were not something that he was committed to promoting, it would still be something that he intended and so something that he would be committed to not complaining about when it came to pass. Commitments of non-complaint are commitments to acquiesce in the world’s turning out a certain way—a form of acquiescence that precludes us from pointing to the absence of a commitment to promote that condition in justifying our behavior.

D. Taking Stock

This Part began with Bratman’s two-part discovery about the nature of intention: intentions serve distinctive needs in coordinating our behavior over time, and as a result, those who have intentions are under distinctive rational pressures with respect to their behavior and their other mental states. To fail to conform to those pressures is to interfere with the proper functioning of intention. It is these special roles and accompanying rational pressures that distinguish intentions from other mental states. This is also the sense in which intentions constitute commitments to that which they depict: when a condition is depicted by one’s intention, then rationality requires various things of a person in light of the presence or absence of that condition. That’s what it is to be committed by one’s intention to the condition.

These observations then prompted a further and more probing examination of varieties of commitment constituted by an intention, varying in what, exactly, rationality requires of you in light of the presence or absence of a condition depicted by your intention. What we found was that in addition to the familiar commitments of promotion—in many cases, a depicted condition is one that the agent is rationally committed to taking steps to promote—there are commitments of non-reconsideration and non-complaint where rationality requires other things of intending agents, things that fall short of promoting the depicted condition.

This raises the question asked already: why should we care? Why should we think that subtleties about the differences between the kinds of commitments engendered by our intentions matter to criminal responsibility? The answer, as we will see in Part III, is that recognizing these subtle differences allows us to formulate, with a certain degree of rigor, an account of the kind of trying that is implicitly prohibited when we prohibit success. It allows us, that is, to develop a principled account of what a criminal attempt is. And once we are armed with such an account, we will be able to go on, in Parts IV, V, and VI, to make concrete recommendations for how the courts should resolve difficult problems arising in the adjudication of attempts.

III. the legally relevant sense of “try”

A. The Guiding Commitment View

At the end of Part I, I argued that neither of two ordinary senses of “trying” aligns precisely with the sense of trying that we prohibit in the law of criminal attempts. The “wide” sense, under which anything that would be true of your act were you to do as you intend contributes to what you are trying to do, is too wide; too much is criminalized as attempt under that conception. The “narrow” sense, under which only that which you are committed by your intention to promoting contributes to what you are trying to do, is too narrow; much that ought to be criminalized as attempt under that conception is not so criminalized. I then promised that further reflection on the nature of intention would allow us to articulate the necessary and sufficient conditions of trying in the sense that is consistent with, and supports, the Transfer Principle. Now that we have a view of the nature of intention in hand, I am in a position to fulfill that promise.

Despite its problems, the narrow sense of trying gets something right: what a person tries to do, in the sense of relevance to criminal law, is a function of what he is committed to by the intention on which he is acting. But to limit the range of relevant conditions to those that we have an intention-based commitment to promoting is to overlook the other ways, discussed in Part II, in which our intentions can commit us to particular conditions. This is the key to understanding the sense of trying that is relevant to the criminal law. To try to act, in the sense of relevance to the criminal law, is to have an intention that commits one (in one of the three senses of intention-based commitment) to each of the conditions involved in completion, and for one’s behavior to be guided by that intention. I call this the “Guiding Commitment View” of attempt.

Under the Guiding Commitment View, to use an example from Part I, a defendant who attempts to take a bag that he falsely but reasonably believes to be his own has not attempted theft under the Guiding Commitment View. His intention does not commit him, in any sense, to the bag’s belonging to someone else. This condition is not depicted by his intention, and so he incurs no intention-based commitment to it. As we will see in Part V, the Guiding Commitment View, in contrast to the narrow view of trying, also implies that the person who believes the white powder is cocaine, but would transport it even if it were heroin or talcum, and tries to transport it across the border, has attempted to smuggle cocaine. So, in contrast to the wide and the narrow senses of trying, the Guiding Commitment View provides us with a way of conceptualizing these examples that is consistent with the Transfer Principle. As we will see in Parts IV, V, and VI, the Transfer Principle and the Guiding Commitment View also provide us with defensible and independently plausible resolutions of a variety of difficult problems of adjudication.

B. Guidance and “Mere Preparation”

It is important to emphasize that there are two parts to trying under the Guiding Commitment View: intention-based commitment and behavior guided by intention. Much has been said already here about intention-based commitment, but a bit more needs to be said about guidance. To be guided by an intention is to be moved or motivated by it to do that which is intended. To be motivated by an intention is for it to be the case that the causal sequence initiated by the intention would culminate in the world coming to match the intention, were obstacles removed and were the agent not to change his mind.

There is more in this brief account of guidance than meets the eye. The account involves a particular view of how motivation differs from other causal influences of intention. A person intends to climb the stairs. This intention causes two things: it causes him to announce “I will ascend the staircase!” and it causes him to take the first step. The intention motivates the taking of the first step but does not motivate the pronouncement. In making the pronouncement, he is instead motivated by an intention to tell the world of his plans. If you ask him why he took the first step, a sufficient answer is, “Because I intended to climb the stairs.” But if you ask him why he made his pronouncement, that answer will not suffice. Making the pronouncement in no way contributes to ascending the stairs, nor does he think it will. It is not rational or worth doing thanks to its contribution to that endeavor, while taking the first step is. But how do we distinguish the two causal influences of the intention to climb the staircase? The intention to climb the staircase does cause the pronouncement; he wouldn’t make the pronouncement if he did not have that intention. So why is the intention’s influence on the one form of behavior (the first step) motivational, but not on the other (the pronouncement)? The best we can do in answer is to note that were the causal sequence leading to the first step to continue without obstacles and without change of mind, the agent would climb the stairs. The same is not true of the causal sequence leading to the pronouncement. That causal sequence will culminate instead in his informing the world of his plans. Therefore, the right way to determine whether the influence of a person’s intention on his behavior is motivational, and so is an instance of guidance, is to ask the following question: had he the ability and opportunity to act, and did not change his mind, would the causal sequence in question have culminated in the intended action?

Efforts on the parts of courts and legislators to concoct “tests” for the act element of attempt can be construed as efforts to describe acts that provide sufficient evidence of motivational influence by intention. They are efforts to identify conduct from which it is safe to infer that the defendant’s intention was having a causal influence on his behavior of a sort that would culminate in commission of the crime in the absence of obstacles. The driving issue, that is, is whether the defendant’s acts provide sufficient evidence of guidance by the intention. Reflection on the notion of “proximity” animating Oliver Wendell Holmes’s influential “dangerous proximity test” leads to this conclusion.31 According to Holmes, the defendant’s act suffices for the act element of attempt if and only if it brought him into dangerous proximity to completion of the crime. But the only coherent and normatively defensible conception of “proximity” in this domain is counterfactual: the defendant’s act is in “dangerous proximity” to completion if and only if all that needs to be added to get completion is ability, opportunity, and absence of change of mind. What Holmes really wanted to know was whether things had gone so far as to convince us that failure could only be attributed to the fortuitous occurrence of obstacles that defeated ability or opportunity, or prompted the defendant to change his mind. The dangerous proximity test, that is, is implicitly motivated by something very close to the Guiding Commitment View itself.

Demonstrating that all the various tests for the act element of attempt are either indefensible, or amount to efforts to capture this idea, requires looking at the details of all the tests that have been offered, and not just one. For our purposes here, however, it suffices to note that the frustrating imprecision that is necessarily involved in this approach—under what conditions, exactly, is the relevant counterfactual true?—cannot be overcome for principled reasons. Since to try, in the sense of relevance to criminal law, is to be guided by an intention that commits one to success, the best characterization of the act element of an attempt will be an account of what it is to be guided by such an intention. It is highly unlikely that we can do better in distinguishing guidance from other influences of intention on behavior than to ask what would happen in the absence of obstacles or change of mind. Like it or not, that’s the question we have to answer to determine if the defendant’s conduct suffices for attempt.

C. Neither “Subjectivism” nor “Objectivism”

The Guiding Commitment View preserves exactly the idea we find in ordinary moral thought—an idea expressed in the Transfer Principle. We justify criminalizing an attempt to commit a crime by citing the features of the very crime attempted. We do not justify criminalizing an attempt to do one thing by citing the features of something other than what was attempted in explanation. So, we need an account of trying to act that preserves this tight connection. We need an account that allows us to identify the attempt in the way that exhibits congruity between it and completion. The Guiding Commitment View serves the turn. It identifies the sense of trying that is relevant to criminal law.

The truth about attempts has eluded theorists of criminal law precisely because they start in the wrong places. They do not start with the question of what kind of failure, what kind of trying, we have implicitly prohibited in our prohibitions of success. As indicated earlier, subjectivists start with the idea that attempts are thought crimes involving some special species of thoughts that it is acceptable to criminalize (such as resolute intentions). Objectivists start with the idea that attempts are crimes involving actions that approximate, or bear some close relation to, harmful conduct of the sort that we typically criminalize (such as conduct that is “proximate” to completion, or that imposes a serious risk of completion).

What subjectivist and objectivist approaches share is the belief that the first and most natural description of the conduct involved in a criminal attempt is a description under which it is not in any sense wrongful. “All he did,” we say, “was light a match. What could be criminal about that?” The subjectivist embraces the claim that the act is not in itself wrongful and concludes that the criminality of the attempt must derive from the accompanying mental state; all he did was light a match, but he lit it with the intention of burning down another’s house. The objectivist, by contrast, seeks an alternative description of the act under which it is wrongful in explanation of the attempt’s criminality; he did not merely light a match, he imposed a serious risk that another’s house would burn down. The problem with both approaches is that the first and most natural description of the act is a description under which it is wrongful. The act is the act of trying to burn down another’s house, or trying to have sex with a minor, or trying to bring drugs across the border. These are wrongful, prohibited behaviors; they were prohibited when we prohibited burning down another’s house, having sex with a minor, and bringing drugs across the border. Trying was prohibited when success was prohibited. The place to start in thinking about attempts is with an effort to identify this kind of trying. And, as we have seen, this leads to the Guiding Commitment View.

In fact, something in the nature of the conceptualization of criminal behavior endemic to criminal law and its practice gives rise to the error shared by subjectivists and objectivists alike. It is because criminal law practitioners and theorists insist on dividing crimes into actus reus and mens rea components that they have overlooked the right way to start thinking about attempts.

The reason is that implicit in the divide is the thought that the actus reus and mens rea components of crimes make independent contributions to the criminality of the conduct. This independence idea prompts the thought that we should be able to explain the contribution to criminality of the actus reus without making any reference to the mens rea. Often this is indeed the case. Burning another’s property is morally salient; it is something one needs to answer for. Intending to burn another’s property is also morally salient. When we put these two things together, there is crime. In this case, the act has features that make it morally salient independently of the mental states that gave rise to it: someone’s property, after all, was burned; someone’s legally protected interests were invaded. And so in the case of completed arson, it is both useful and clarifying to specify the actus reus and the mens rea as distinct and conceptually separate conditions. But if we assume that this is true of all crimes, then we quickly find ourselves on the road to either subjectivism or objectivism, for the truth is that many an attempted crime is not wrongful under any description that applies to it independently of the mental states that gave rise to it. The conduct in question is wrongful alright, but only under descriptions that are not mens rea-independent, namely descriptions that specifically identify the fact that the conduct is a trying. To describe the act as trying to burn another’s property is to describe it in a way that implicates and refers to the mental state that gave rise to it, namely an intention that committed the actor to each of the features of completed arson. Given that students of criminal law are taught to characterize actus reus and mens rea independently, it is no wonder that theorists take the central puzzle about attempt to derive from the thought that it involves conduct that is not on its face wrongful; they have been taught to attach a description to the act that leaves out exactly that in virtue of which it is wrongful, namely that it is an instance of trying to do something wrongful. The tidy conceptualization of crimes as consisting of distinct actus reus and mens rea elements, useful and illuminating as it is for understanding the structure of most, if not all, completed crimes, misleads when it comes to attempt.

When we avoid the error of assuming that actus reus and mens rea make independent contributions to the criminality of the behavior they constitute, we find ourselves on the middle way. After all, if a defendant fires a gun and misses, and if the criminality of his action emerges from the fact that the actus reus and mens rea together constitute the defendant’s trying to kill, then we need to know in what sense of “trying” this is true. The answer is that it is true in the sense of trying that is implicitly criminalized when we criminalize completed homicide or any other completed crime. From here we find ourselves with the Guiding Commitment View, for this view identifies all and only the failures that are implicitly prohibited when we prohibit success. Further down this road, as we will see in the next three sections, lie solutions to many difficult problems encountered by those tasked with adjudicating attempted crimes.

D. Taking Stock

Acting on the hunch that adjudicatory problems about attempt can be solved by a defensible theory of what an attempt is, we were led to an account of the criminalization of attempt, namely the Transfer Principle. Acceptance of the Transfer Principle allowed us to formulate the question of what an attempt is in a way that made it tractable. That became a question about what sorts of tryings are implicitly prohibited when we prohibit success. Reflection on that question first led us to an account of intention as a source of a range of commitments to those conditions that are depicted by the intention. And this account led us, in turn, to an account of trying, namely the Guiding Commitment View, according to which to try to commit a particular crime is to have an intention that commits one to all of those conditions involved in completion of the crime, and to be guided by that intention. It is in that sense of trying that trying is implicitly criminalized when we criminalize completion.

In difficult attempt cases, what the defendant did—or even what mental state he did it with—is not in dispute. What is in dispute is whether what he did constituted an attempt of the sort he is accused of having committed. The Guiding Commitment View provides just what the inquiry needs: an account of the necessary and sufficient conditions that are met when a person has indeed committed an attempt, in the sense of “attempt” that matters to criminal law. This should make us optimistic that the Guiding Commitment View can help us with our adjudicatory problems. It is to those that we now turn.

IV. trying by asking: solicitation as attempt

Return to Ronald Decker, who, thinking he was dealing with a hitman, paid an undercover detective to kill his sister, and who, fighting for something less than a life sentence, argued that while his solicitation was admittedly a crime, it did not amount to an attempt. In 2007, the Supreme Court of California announced that Decker was, in fact, attempting to murder his sister.32 In making this announcement, the court overturned a case from the 1970s, People v. Adami,33 another case in which the defendant paid an undercover police officer to kill someone but in which the court ruled that a solicitation did not amount to an attempt. In fact, there is no standard view across jurisdictions on the matter. Pay someone to kill in Idaho, Nevada, or South Dakota, for instance, and you’ve committed only the lesser crime of solicitation;34 pay someone to kill in Georgia, Iowa, New York, or Ohio, and you’ve attempted the murder.35 We find even greater variation when we consider crimes other than murder. If you ask a child for oral sex, have you attempted sexual battery?36 If you ask a minister to marry you to your niece, have you tried to marry someone incestuously?37 If you ask someone to bribe a witness, have you attempted to bribe a witness?38 If you ask someone to burn down a barn, have you attempted arson?39 Courts regularly have to answer questions like this. They answer them differently in different places and in different times, and they give different answers when different crimes are involved.40 In fairness to the courts, it is not easy. The issue stands at the intersection of three independently thorny parts of the criminal law: attempt, solicitation, and complicity, and if we aren’t careful, we find ourselves embroiled in all but intractable questions about causation as well.41

The Guiding Commitment View can help us to identify the crucial questions that need to be answered in order to provide a principled solution to the problems faced by courts. The solution proposed here is as follows: in the most difficult cases, if the defendant has asked another person to bring about an event that is a result element of the completed crime, then the defendant may have thereby attempted the crime; if, however, the event that the defendant has asked another person to bring about figures into the definition of an act element of the completed crime, then the defendant has not attempted through his solicitation. Thus, a distinction that seems to be merely formal—the distinction between act elements and result elements of crimes—turns out to be of substantive importance.

A. The Insufficiency of Accomplice Liability

The Decker case, and many others like it, defies resolution through appeal to doctrines concerned with complicity, and this is part of the reason why courts like the Adami court have concluded that such cases do not involve attempt. Had Holston attempted the murder that Decker asked him to commit, then we could have conceptualized Decker as an accomplice to Holston’s attempt.42 But Holston didn’t go through with the murder.43 If Holston had performed an act that would have sufficed for an attempt had he, Holston, in fact intended to commit the crime, then perhaps we could have employed a legal fiction according to which Holston’s act is treated for legal purposes as though it were Decker’s. This is what we do in cases in which, for instance, the defendant asks a child to put poison in another’s coffee, and the child complies.44 But Holston took no steps at all that could be construed as steps toward committing murder.45

Intuitively, it seems that some cases that complicity doctrines are incapable of resolving are attempts, and some are not.46 For instance, say that the defendant is holding a quantity of heroin and asks another person to take it—maybe the defendant wants to kick the habit but just can’t bring himself to destroy the stuff; the solicited party refuses. In such a case, the defendant has solicited possession of heroin. But he obviously hasn’t attempted possession of heroin by so doing. After all, what the defendant is trying to do is to dispossess himself of the heroin. Imagine, to give another example, that the defendant asks another to try heroin and is thereby acting on an intention that the other should use; the solicited party refuses. Did the defendant attempt to use by soliciting the other? Of course not. Imagine that the defendant asks another person to drive himself home knowing full well that the solicited party is far too drunk to drive legally; the other refuses. The defendant has clearly not attempted to drive drunk in such a case, even if his conduct is as wrongful, or even more wrongful, than such an attempt would have been.

By contrast, imagine that the defendant asks another to carry a quantity of heroin across the border into the United States; the other refuses. Has the defendant attempted to import drugs into the United States? It seems so. Imagine that the defendant asks another to deface a piece of public property; the other refuses. It seems that the defendant, in such a case, has attempted to damage public property. If examples fitting facts of these sorts are constructed with sufficient care, then all will be irresolvable through accomplice liability principles. Intuitively, then, some such hard cases are, and some are not, attempts. What facts are being tracked by our intuitions? What principles ought a court use to decide in such cases?

B. Act Crimes, Result Crimes: How Courts Should Decide

The solution to our problem can be reached by recognizing that in cases like Decker, the defendant’s intention differs from that of the paradigmatic attempter. As we will see, despite this difference, sometimes the defendant’s intention commits him to all of the components of the completed crime, and sometimes it does not. Therefore, sometimes the defendant’s intention suffices for the attempt, and sometimes it does not. In other words, the distinction crucial to the resolution of hard cases is a distinction in mental state; in some such cases, the defendant has the mens rea for attempt, and in others he does not. The Guiding Commitment View tells us what intention the attempter has: an intention that commits him to each of the components of the completed crime. So it seems likely that we will be able to draw a distinction between those solicitations that rise to the level of attempt and those that do not by distinguishing between those solicitations that are motivated by an intention that commits the defendant to all the elements of the solicited crime, and those that do not.

In hard solicitation-as-attempt cases like Decker, the defendant’s relevant intention is not an intention to perform any act that would serve as the act element of the completed crime. Decker did not intend to point a gun at his sister and pull the trigger, for instance. But in all such cases, the defendant does intend that the solicited party act in some particular objectionable way: Decker intends that Holston kill his sister. A crucial question, then, for determining if the defendant’s solicitation amounts to an attempt is as follows: under what conditions is an intention with that content sufficient for the intention required for the attempt?

The answer to this question turns on the distinction between “act” and “result” elements of crimes. There is a clear distinction between what we merely cause and what we do. To do is to cause, but to cause is not necessarily to do; sometimes we cause events that we are not active with respect to in the way that is required for action. A woman calls her dog from across the street. The dog runs to her and, much to the woman’s horror, is crushed by a passing car. The woman caused the dog to be crushed, but she did not crush the dog. A particular event, such as a dog being crushed, is an act element of a crime if, and only if, three things need to be shown by the prosecution to establish that the defendant committed the offense: that the event occurred, that the defendant caused the event, and that the defendant was active with respect to it. It must be shown, that is, that the event occurred and was caused by the defendant in the way that qualifies it as his doing. By contrast, if the event is a result element, it needs only to be shown that the event occurred and that the defendant caused it; the defendant need not be shown to have been active with respect to it. Circumstantial elements of crimes are those conditions that need to be shown to have been present, but do not need to be shown either to have been caused by the defendant or to have been something with respect to which the defendant was active.47

The distinction between acts and results is both highly intuitive and difficult to specify with precision. When you walk into a room, you cause the temperature in the room to rise, but that is not something you do. Part of the difference between results and doings derives from facts about mental state. The things that you cause unwittingly, without any kind of prior awareness of even the possibility that you will cause them, are not things with respect to which you are active. However, intention is not required for activity; there are things that we do, but unintentionally. Further, prior awareness falling short of intention, such as awareness of significant risk, plus causation does not suffice for activity. Even if the woman, in the example just given, who calls her dog was aware of a risk that her dog would be crushed, it is not the case that she crushed her dog. In that case, what is missing is a kind of personal involvement in the crucial event—dirty hands—that is present when the event is one with respect to which the agent is active. The project of laying bare the conditions that underlie our intuitive judgments about the line between what we do and what we cause is one of the central projects of the philosophy of action. Ultimately, a proposed specification of the conditions that constitute this distinction is beholden to our intuitive judgments. To test such a proposal, we need to compare it to the verdicts that we make in ordinary contexts. We need to decide whether the events that an agent is active with respect to under the account are those that we would judge, in sober moments, to possess that extra something that we find in action and for which causation by itself is insufficient. Luckily, the law’s sorting task—the task of sorting those who have committed crimes from those who have not—can be carried out to a large extent, even if not always, by appeal to the intuitive judgments to which a precise account would be beholden. We can move forward trusting that when the prosecution needs to show that a defendant was not just the cause of an event but was also active with respect to that event, juries will know activity when they see it, and will know also when it has not been shown.

Therefore, under the Guiding Commitment View, the precise intention needed for an attempt is a function of the classification of the elements of the completed crime as act, result, or circumstance. The Guiding Commitment View tells us that what a person has attempted is a function of what his intention committed him to. Only if the defendant’s intention commits him to all that is involved in the completed crime has he attempted. But what is involved in the completed crime is in part a function of the classification of its various elements as act, result, or circumstance. If the completed crime includes the act of defacing property, then for attempt the defendant’s intention must commit him to property being defaced, to causing property to be defaced, and to being active with respect to the defacement of the property. By contrast, if the completed crime includes the result of property being defaced, then the defendant’s intention must commit him only to property being defaced, and to being the cause of that; it need not commit him to being active with respect to the relevant event.

While crimes have distinct act elements and result elements, the acts that the criminal law cares about are virtually always described by reference to their results. The California murder statute of relevance to Decker’s crime, a statute that is perfectly typical of jurisdictions that have not adopted the Model Penal Code, reads, “Murder is the unlawful killing of a human being . . . with malice aforethought.”48 Should we characterize this as a crime with the act element of killing or a crime with the result element of death? It is far from clear what the answer is to this question. (More on this below.) Still, it is natural to characterize it in the first way, given the appearance in the statute of the term “killing.” So characterized, among the things that must be proven in a murder trial in California is not just that the defendant caused a death, but also that the defendant was active with respect to that death. By contrast, the Model Penal Code defines one form of “criminal mischief” like so: “A person is guilty of criminal mischief if he . . . purposely or recklessly causes another to suffer pecuniary loss by deception or threat.”49

Here it seems that the act element is deception or threat—a guilty defendant deceived or threatened another—while pecuniary loss is a result element.50 A guilty defendant caused pecuniary loss, but it need not be the case that he was active with respect to the other’s loss of money. For a defendant to be guilty of criminal mischief under the Model Penal Code, he must have caused another to be deceived or threatened while active with respect to that condition, and must have caused another to lose something of pecuniary value. Buthis activity with respect to the latter condition is irrelevant; it will be present in some guilty defendants, absent in others. The broker who, through deception, takes control of another’s money and loses it is guilty of criminal mischief; the defendant is active with respect to both the deception and the loss in such a case. But the person who deceives another and thereby leads him to make an investment whereby the victim loses his own money has also committed the crime; such a person has committed criminal mischief even though the defendant is not active with respect to the loss of money.

The divide between act elements and result elements is not merely formal; it is of normative significance. It is something about which legislators ought to debate when defining a crime by statute. To criminalize the causation of a fetus’s death, for instance, would be to criminalize a much larger number of acts than would be criminalized were the crime to require, instead, the killing of a fetus. If a defendant is at fault in a minor car accident with a pregnant woman that results in the death of the fetus, then he has certainly caused a fetus’s death, but it is far less clear that he has killed a fetus. We can imagine cases in which activity with respect to an event is absent despite the fact that there is causation. Whether we want to criminalize such behavior under a particular statute will depend on our normatively significant goals in writing the statute and will influence what penalty we take to be appropriate for the crime we are defining. One, although not the only, relevant factor is this: when an event, like a fetus’s death, figures into the description of the act required for the crime—when what is criminalized is “the killing of a fetus”—themens rea requirement for the act and the result cannot be different. A person who recklessly kills a fetus is also reckless with respect to the death of the fetus. By contrast, when an act element is separately defined from the result element—when what is criminalized is, for instance, any act that “causes the death of a fetus”—then there is room for the mens rea with respect to the act and result to diverge; it is possible to intentionally run a red light and only negligently thereby cause a fetus’s death. So the act-result distinction makes a difference to what is criminalized. Although it is a formal distinction, the act-result divide is a formal distinction of normative importance.

Some intentions commit one to causing a particular result and commit one to being active with respect to it. Others commit one only to causing it. Although it is an imperfect guide, we often register this difference by describing what the person intends either with the word “to” or, instead, with the word “that.” An intention to buy milk constitutes a commitment to both milk being bought and to the buying of it being something one does, or is active with respect to. An intention that milk is bought constitutes a commitment to causing that, but no commitment to doing it oneself. Someone with the intention to buy milk might be prompted to take a trip to the store. Someone who intends that milk be bought might be prompted to put it on the list of things that her friend, and not she, is to buy later.

Under the Guiding Commitment View, a person has attempted a crime only if his intention commits him to each of the elements of the completed crime. So, if the completed crime includes a particular result, such as “pecuniary loss,” it suffices that the attempter intend that there should be pecuniary loss. If the completed crime includes a particular act, such as “deception or threat,” then it is necessary that the attempter intend to deceive or threaten; an intention that a victim should be deceived or threatened will not suffice.

This distinction is the key to understanding the conditions under which a solicitation is an attempt. Recall that in cases like Decker, the defendant does not intend that he act himself, but only that the solicited party perform an act that would serve as the act element of the completed crime. Under what conditions does this show that the defendant lacks the mens rea of attempt? The defendant’s intention is sufficient for the mens rea of attempt when the event that the defendant intends the solicited party to cause is a result element of the completed crime, but his intention is insufficient for the mens rea of attempt if that event figures in the specification of an act element of the completed crime. This conclusion follows immediately from the account of the intention needed with respect to acts and results just offered.

So does Decker have the intention needed for an attempt to murder his sister? Since California’s homicide law apparently provides that one must have killed another in order to have committed murder—thereby incorporating the other’s death into the definition of an act element of the crime—the answer appears to be “no.” Under such a statute, an attempt to murder requires an intention to kill, something that Decker lacks; his intention that his sister die is not enough for attempted murder under California law because it does not commit him to being active with respect to that result, and so does not commit him to killing anyone. However, consider the Model Penal Code’s alternative definition of homicide: “A person is guilty of criminal homicide if he purposely, knowingly, recklessly or negligently causes the death of another human being.”51 Here the act element is left undescribed; any act at all could serve for the act element of the completed crime, even an act that was not a killing. The death of a human being is a result element of the crime. Under the Model Penal Code, all purposive or knowing—as well as some reckless—criminal homicides are classified as murders provided that they are not committed under extreme emotional distress. Therefore, a person should be able to try to commit murder, as defined by the Model Penal Code, by intending to perform some act and intending that another dies. Decker has these two intentions, for he intends to solicit Holston and intends that his sister die. Therefore, Decker has attempted a murder under the Model Penal Code’s definition of that crime.52

A court faced with a statute that specifies “causing A” as an element must consider whether there is an appropriate verb that would have allowed “A-ing” instead of “causing A.” If so, then that is a powerful reason to take A to be a result element. Consider, for instance, a statute prohibiting “causing the release of classified information.” That the legislature did not instead prohibit “releasing classified information” is evidence that the information’s release is a result rather than an act element of the crime. However, if there is no such verb, then the statute is ambiguous; agency can be present even when there is no relevant verb. Consider, for instance, a statute prohibiting “causing a wildfire.” Since there is no verb “to wildfire,” it is unclear whether the occurrence of the wildfire is an act or result element of the crime. In such a case, the court may have no choice but to ask whether the normatively relevant category of conduct is the one in which the agent is active with respect to the event, or the one in which he is merely the cause of it.

Conversely, when a court is faced with a statute that specifies “A-ing” as an element of the crime, but the legislature could just as easily have criminalized “causing A,” that is one piece of evidence that the element in question is an act element. In the absence of a special reason to think the element is a result element, it should be classified as an act element. However, the fact that the legislature could have written the statute either way is a highly defeasible piece of evidence because legislatures are not as attuned to the distinction between acts and results as they should be and so may have intended the element in question to be a result while using an active verb to refer to it. In such cases, an argument showing that causing A is just as bad as—or just as legitimately criminalized as—A-ing will support the characterization of the element in question as a result element, despite the statutory language.

Although I will not defend the claim here, it seems to me that the Model Penal Code’s definition of homicide (with the death of another as a result element rather than killing another as an act element) is superior, on normative grounds, to definitions, like that in California, under which a killing is apparently required—or, rather, in which such a requirement fits most naturally with the statute’s language.53 But the important point for our purposes is that the question of whether a solicitor has the intention needed for an attempt has to be settled by a normative argument over how to define the completed crime. One can imagine the Decker court’s having reached the verdict it sought, even given the language of California’s murder statute, by claiming that, despite appearances to the contrary, the death of another is a result element of the crime of murder in California. To do this, the court would have had to abandon the most natural interpretation of the explicit language of the statute, but there may be an argument for doing so. After all, the court might have reasoned, common law homicide definitions were formulated using language that was developed without the act-result distinction firmly in mind, and so it would be no surprise if some statutes were drafted with language misleadingly suggesting that a result element figures, instead, into the description of an act element. The fact that, on normative grounds, it is arguably better for death to be considered a result strengthens the case for this form of argument.

C. Taking Stock

In attempt and solicitation, not to mention conspiracy, we recognize a sense in which the criminality of the act is diluted in comparison to that from which its criminality derives, namely the completed crime attempted or solicited. If completed crimes are blood red, then attempts and solicitations seem to be shades of pink. This metaphor tempts one to imagine that the more we stack attempts and solicitations on top of one another, and the more attenuated the relation to the completed crime becomes, the more diluted is the criminal liability. However, this is only a metaphor and one by which we should not be misled. In attempted crimes we do what we can to extend our agency into the world in objectionable ways. In solicitations, we do what we can to extend another’s agency into the world in objectionable ways. What has been shown in this Part is that one way to extend one’s own agency into the world is to lead another to extend his. This is so when all that is required to extend one’s own agency into the world in an objectionable way is to cause a result. In order to try to cause a result, it is enough to ask another to bring it about. Others cannot act for us, but they can alter the world on our behalf, and sometimes to ask someone to do so is enough to try to do so oneself.

It is important to see how the idea that solicitation rises to the level of attempt in result crimes, and not in act crimes, derives from the theoretical framework developed in Parts I, II, and III. Under the Transfer Principle, the prohibition of an action is also an implicit prohibition of an attempt to engage in that action; a prohibition of causing a result is an implicit prohibition of an attempt to cause that result.54 Either way, the prohibited attempt is of the sort that would count as an attempt under the Guiding Commitment View. This type of prohibited attempt, in turn, requires an intention that commits one to that which is involved in completion. If completion involves an act, then the attempt requires an intention that commits one to three things: an event’s occurrence, being the cause of it, and being active with respect to it. An intention that a solicited party act in the prohibited way does not commit the defendant to being active with respect to that which the solicited party is asked to cause. Therefore, the defendant is not, in such a case, attempting to engage in the prohibited conduct in the sense of relevance to criminal law; he lacks an intention-based commitment as required for the attempt. By contrast, if what is prohibited is the causing of an event, then the defendant has attempted only if he has an intention that commits him to the event’s occurrence and commits him to causing it. He can be so committed by an intention that a solicited party cause the event. He therefore may have an intention that suffices for the attempt. The conception of attempt on offer in this Article, that is, provides us with principled grounds on which to resolve hard solicitation-as-attempt cases of a sort that have baffled the courts.

V. circumstances and “impossibility”

A. Clearing Ground and Setting Aside Legal and Factual Impossibility

There is perhaps no area of criminal law in which there is more confusion, on the parts of both courts and commentators, than in the adjudication of so-called “impossible” attempts. Consider an appealing line of thought that leads to a disastrous set of doctrines, namely the doctrines employing the distinction between “legal” and “factual” impossibility. Start with cases sometimes dubbed as “pure legal impossibility” cases: a married man tries to commit adultery and fails. The object of his affections is not interested. Ashamed and falsely believing that adultery is illegal, he runs to his local police station to turn himself in for attempted adultery. The police laugh in his face and send him home. Why? Because: it is not a crime to attempt to do something that would not be criminal were you to have done as intended. Had the attempted adulterer succeeded, he would have committed no crime, and so his attempt was no crime. So far so good. Or so it seems.

But now let’s extend that principle. Consider a case that has occupied the imagination of many a criminal law theorist, namely People v. Jaffe.55 The defendant, who was suspected of regularly aiding thieves by selling the goods they took, was charged with an attempt to receive stolen property after he purchased some fabric that he believed to be stolen56 but that was in fact falsely represented to him as stolen as part of a sting operation.57 At least part of what Jaffe intended was to receive property. But in fulfilling that intention, Jaffe did not commit a crime, because the property he intended to receive was not stolen. Apply the principle appealed to in the case of the attempted adulterer, and it appears to follow that Jaffe has not attempted receipt of stolen property. In both cases, that is, the defendant has an intention such that, were it fulfilled (and in Jaffe’s case it actually was fulfilled), there would be no crime. Now, if Jaffe did not attempt receipt of stolen property, then attempt liability radically shrinks, as do the tools available to law enforcement. In order to avoid this outcome, judges sought a distinction between a case like that of the attempted adulterer and a case like Jaffe. It is not implausible to think that what distinguishes the cases is that the attempted adulterer made a mistake of law, while Jaffe made a mistake of fact. Hence a bit of doctrine is born: if the obstacle to completion is factual, the attempt is criminal (the case is one of “factual impossibility”), while if the obstacle is legal, the attempt is not criminal (the case is one of “legal impossibility”).

But the doctrine is a disaster, and, thankfully, has almost disappeared in the United States. The problems are many and need not be rehearsed here.58 What is of importance to us is a subtle error that leads to the doctrine, an error that we must not make in thinking about cases like Jaffe. The reason that the attempted adulterer has committed no crime is that he is not committed by his intention to all of the components of any crime. He is committed, instead, to the components of a non-criminal form of conduct, namely adultery. The mistake is in thinking that the right question to ask is, “Would the attempted adulterer have committed a crime had he done as intended?” This is the wrong question because it invites an answer in which we appeal to facts about the world to which the defendant is not committed by his intention—namely any facts that would be in place were he to do as intended. Recall the “wide” sense of trying, discussed in Part I, under which any facts about what the person would do were he to do as he intends—including facts about whether his conduct would be criminal—can be appealed to in a description of what he is trying to do. To explain why the attempted adulterer has committed no crime by appealing to the fact that he would not have committed a crime had he done as intended is to assume that the sense of “try” that informs the criminal law is the “wide” sense, rather than the sense defined under the Guiding Commitment View. Accepting such an explanation of the non-criminality of the attempted adulterer’s conduct would be to invite mistake in the treatment of a case like Jaffe. The right question to ask about Jaffe is not “Would he have committed a crime had he done as intended?”—a question to which the answer is “no.” The right question is, “Was he committed by his intention to all of the components of the crime of receipt of stolen property?”—a question to which the answer may be “yes” (for reasons to be explained below). We need to attend not to what would be the case were the defendant to do as intended, but to that to which he is committed by his intention.

This change in orientation radically alters the problem. The important issue not only has nothing to do with the fact-law distinction, but also has nothing to do with impossibility. Sometimes we are committed by our intentions to things that cannot come to pass, sometimes to things that can. What is crucial is what we are committed to, not whether it can come to pass. The next step is to remember that, as emphasized in Part II, a person can be committed by his intention to a condition without being committed to promoting that condition. We can also be committed to not reconsidering our intention in light of the presence of the condition, and thanks to an intention, we can be committed to not complaining that the world is in a certain condition. When we have these forms of commitment, we are under no pressure to adopt necessary means to realizing the condition to which we are committed. Under the Guiding Commitment View, when a person is not committed to promoting a condition, but is committed to it in one of these other two ways, then the condition can be appealed to in a characterization of what he is trying to do. So if it were the case that Jaffe was committed in one of these ways by his intention to the property being stolen, then he attempted receipt of stolen property in the sense that is criminalized under the Transfer Principle. Was he?

B. Belief and the Rebuttable Presumption

It is obvious that in a case like Jaffe a crucial fact about the defendant that informs our belief that he attempted a crime is that he believed that the property he received was stolen. Similarly, recall that Crow believed that the person he tried to sexually exploit, “StephieFL,” was a minor. And there are many other cases with the same structure. Not all such cases have led to conviction for attempt, even though that is by far the most common result.59 In United States v. Berrigan,60 for instance, the defendant thought that he was smuggling letters out of the prison in which he was housed without the warden’s knowledge.61 In fact, the warden knew all about it and was allowing the letters to leave so as to help the police track the activity outside the prison that Berrigan was directing through the letters.62 The court acquitted Berrigan of the attempt to smuggle letters out without the warden’s knowledge on the ground that the presence of the circumstantial element of the completed crime (namely that the warden did not know) was required for the attempt.63 If this seems like the wrong result in this case, it is at least in part because Berrigan believed the warden did not know of the letters. The defendant’s belief seems to matter, but why?

One might think that any view, such as the Guiding Commitment View, that requires for attempt an intention-based commitment to each of the conditions involved in the completed crime must say that belief is insufficient in cases like Jaffe, Crow, and Berrigan. After all, it is common to believe a condition to be in place without it being something to which one is committed by one’s intention. The runner believes that he lives in Los Angeles, but we cannot infer from this fact, together with the fact that he intends to go running, the further claim that he intends to go running in Los Angeles. That is something that he takes for granted, not something that he directs his will toward in the form of an intention. Perhaps it is like that with Jaffe, or Crow, or Berrigan. Perhaps they believe that the relevant conditions are in place, without those conditions being things to which they are committed by their intentions.

And indeed this is possible. It is just extremely unlikely. There are quite complicated principles that govern the flow of information from one mental state to another; sometimes content bleeds from one mental state to another, and sometimes it does not. Often, people who believe that Obama is Hawaiian and believe that Obama is President also believe that a Hawaiian is President. But a particular person might not have this further belief. He might not “put two and two together.” If asked if a Hawaiian has ever been President, he might have to think about it, or he might even answer “no” before realizing that this conflicts with an implication of other things that he believes. Failures to put two and two together have two related sources of importance. First, we sometimes fail to put two and two together because one belief or the other is buried; it is not “before the mind” in the way that it would need to be to serve as a premise in reasoning through which its content would bleed into the other. This might be the case for someone who knows that Obama is Hawaiian but has not given that any thought in some time. Second, and alternatively, a person can fail to put two and two together because one of the facts is extremely unimportant to him; it is not salient. Someone who doesn’t care about such things will not have the thought that the President is a dog owner, even if he has the thought that Obama is a dog owner, and the thought that Obama is President.

These examples involve pairs of beliefs, but we find the same limitations on content moving from a belief to an intention. Someone who intends to vote for Obama might not intend to vote for a dog owner, even though he knows that Obama is a dog owner. That fact just does not matter to him. Or even if it does matter to him, it might not be before his mind at the time that he intends to vote for Obama. But notice how rare it is in criminal cases like Jaffe, Crow,and Berrigan for either the relevant beliefs to be buried or for the conditions not to be salient to the defendant. The relevant conditions are labeled as parts of crimes, which itself makes them salient in the minds of many people. Furthermore, in many cases they are labeled as such because they are morally important. And even many of those with radically different moral views from those enshrined in criminal law recognize that the prospect of punishment is linked to the presence or absence of the relevant conditions, and that, for most, makes the conditions salient. This is not to say that there are no defendants in whom the beliefs are buried or the conditions not salient; there undoubtedly are, but they are rare. They are rare enough for the fact of belief in the relevant condition to provide a rebuttable presumption that the condition is included in the content of the defendant’s intention. Given that he believes that the property is stolen and intends to receive property, we can be virtually certain that Jaffe intends to receive stolen property; given that he believes that StephieFL is thirteen and that he intends to sexually exploit her, we can be virtually certain that Crow intends to sexually exploit a minor; given that he believes the warden is ignorant of his activities and that he intends to send letters, we can be virtually certain that Berrigan intends to send letters without the warden’s knowledge. Can we be certain that the relevant conditions bleed from the defendants’ beliefs to their intentions? It is not a conceptual truth, but in the absence of evidence that defeats the claim, there is every reason to think the intention includes a representation of the believed condition.

So if we ask, “What is the appropriate mens rea standard in attempt with respect to the circumstantial elements of the completed crime?” the answer is “intent.” That is an implication of the Guiding Commitment View that we should embrace. But mental states “lower” on the mens rea hierarchy can provide extremely good evidence, sufficient evidence in the absence of rebuttal, for the needed intention. If there are remaining intuitions that intent is too high a standard in this domain, then they arise from a failure to appreciate the fact that one can intend a condition, and thereby be committed to it, without having any commitment to promote it. To say that Crow intends to sexually exploit a minor is not to say that he would have dropped his pursuit of her photographs had he discovered she was not a minor. Maybe he would have been happy to receive sexually explicit pictures of either a minor or an adult. Who knows? But still, given that he intended to sexually exploit a minor, he was committed to her being a minor in both the senses of non-reconsideration and non-complaint. He could not rationally have changed his mind on encountering further evidence of her minor status; nor could he complain that the world came out differently from the way he intended if, in the end, he sexually exploited a minor. These are commitments that he incurs compatible with the absence of any commitment on his part to see to, or promote, the object of his sexual exploitation’s being a minor. And it is thanks to the fact that he has these intention-based commitments that he is properly said to have attempted to sexually exploit a minor in the sense of relevance to criminal law.

C. Recklessness and the Relevance of the Facts

Cases in which the defendant believes the relevant circumstantial element of the completed crime to be present are not, by any means, the only sort. A defendant who tries to take something that he lacks permission to take might fall short of belief that he lacks permission; perhaps he is merely reckless in this respect—his wife was supposed to ask permission for him to take his neighbor’s car, but he’s not sure if she remembered. His belief is aptly characterized as probabilistic, rather than “flat-out.” He does not flat-out believe that he lacks permission; he believes, instead, that there’s a decent chance he lacks permission. Did such a defendant attempt theft? A defendant who tries to buy a stereo off the back of a truck falls short of belief that the stereo is stolen—maybe it’s just a roaming garage sale—but he knows there’s a good chance that it is. Did such a defendant attempt to receive stolen property?

Notice that these recklessness cases are importantly different from those involving defendants who are negligent or even blamelessly oblivious to the presence of certain conditions. In negligent attempts, there is no mental state the content of which can bleed into the content of the intention. Hence there is no intention-based commitment to the relevant condition. This is why it is just false to say that someone who tries to have sex with someone underage, but whom he unreasonably believes to be an adult, has attempted rape. This is false even if negligence, or less, will suffice for the completed crime. When it comes to the attempt, we need intention-based commitment. Furthermore, when there is no mental representation of the relevant condition, there is no such commitment.

Under the Guiding Commitment View, the question in recklessness cases, by contrast to negligence cases, is parallel to the question involved in the flat-out belief cases: is the defendant committed to the condition (that he lacks permission, that the property is stolen) by his intention? If so, then we can appeal to the condition in describing what he is trying to do (steal rather than merely take; receive stolen property); and if not, then we cannot. To a point, although only to a point, the solution is the same as well: the content of the probabilistic belief can be presumed to bleed into the content of the intention in the absence of a failure on the defendant’s part to put two and two together. However, there is an important difference. When the content of the probabilistic belief bleeds into the content of the intention, the defendant does not intend to receive stolen property, but instead intends to receive possibly stolen property. So we have a new question: does an intention-based commitment to the property’s possibly being stolen suffice for the kind of commitment to its being stolen that is needed for attempt to receive stolen property?

The answer is, for reasons to be explained, that it suffices only if the property is actually stolen. In short, whether the intention to receive possibly stolen property constitutes a commitment to the property’s being stolen depends on whether it is, in fact, stolen. To see this, start by noticing that an intention to receive possibly stolen property does not generate a commitment of non-reconsideration with respect to the property’s being stolen. A person with that intention whose doubt is removed—the guy on the back of the truck, for instance, says, “By the way, this stuff is hot”—could rationally reconsider his intention in light of the newly formed belief that the property is stolen. A rational agent might be willing to go through with the purchase when he’s not certain and unwilling to go through with it when he is. So, if the intention to receive possibly stolen property commits one to its being stolen, it cannot be because it generates a commitment of non-reconsideration.

What about a commitment to not complaining? Here there are conditions under which the intention in question generates such a commitment. Say, for instance, that the stereo is purchased and the purchaser is arrested, and it is demonstrated that the stereo is stolen. Can the purchaser rationally deflect criticism from himself by noting that at least the world is not in a condition he intended it to be in? No, for the intention depicts the property as stolen, with some degree of likelihood, and unstolen with some other degree of likelihood. Given that the property is in fact stolen, the intention’s depiction of the property’s status is accurate; the intention matches the world. If you think the world might be a certain way, and it is that way, then you were right. If you think the world might be one of two ways, and it turns out to be one of them, then you were right. Similarly, if an intention depicts the world as possibly a certain way, and the world is that way, then the intention matches the world.

Since it would be irrational for the defendant to insist that the stereo’s being stolen is something he did not intend, he has a commitment of non-complaint with respect to the property’s being stolen. But notice: he only has that commitment if the property is actually stolen. If the property turns out not to be stolen, then that is something that he is committed to not complaining about. What he is committed to not complaining about is a function of how things are. What follows is that when the defendant is merely reckless with respect to the relevant condition, he has attempted the crime only if the condition is actually in place. So, whether the prosecution bears a burden to show that the condition is in place depends on what the defendant’s mental state is. If the defendant believes the condition to be in place, then the prosecution bears no such burden. But if the defendant is merely reckless with respect to the condition, then the prosecution must establish that the condition is in place in order to establish that the defendant attempted the crime. Put conversely: when the prosecution is able to prove only that the defendant was reckless with respect to a particular circumstance, then the prosecution must also prove that the circumstance was in place in order to show that the defendant attempted the crime.

One of the things that is uncovered here is the root of a particular source of confusion in the conceptualization of attempt. We recognize, and without difficulty, that when it comes to a completed crime like receipt of stolen property, there are two different and separable questions to ask. First: was the property stolen? Second: did the defendant believe that the property was stolen? There are two questions because there are two different morally salient facts: one about the property’s status and another about the mental state of the defendant with respect to that fact. But when theorists have tackled cases of “impossible” attempt—cases like Jaffe, Crow, or Berrigan, or equivalent cases involving recklessness rather than knowledge—the question of whether the circumstances need to be in place has seemed to be inextricably intertwined with the question of what the defendant’s mental attitude needs to be toward the circumstance. The result is that the opposite sides of the bar talk past each other. One side sees the import of the circumstantial facts, the other of the mental state—and it can seem as though adjudicating the dispute requires deciding which of the two matters, to the neglect of the other.

But there is a reason why the facts and the defendant’s mental state with respect to them seem entwined in attempt: they are entwined in attempt. When the defendant’s intent is, for instance, to sexually exploit someone who might be a minor, he is attempting to sexually exploit a minor only if the victim is in fact a minor. The facts are relevant to what his mental state commits him to. What his mental state commits him to is relevant to what he is trying to do in the sense of relevance to the criminal law. And what he is trying to do in that sense is relevant to whether he has committed a criminal attempt, because we criminalize attempts under the Transfer Principle. The abstract bit of progress made here in philosophy of mind and action—noting the range of commitments beyond commitments of promotion—is not merely that; instead, it is identifying something about which those who care about crime and culpability ought to care.

D. Taking Stock

In our resolution of the problem of solicitation-as-attempt, we needed to appeal to the Guiding Commitment View and to a plausible theory of the distinction between result and act elements of crimes. Only one implication of the Guiding Commitment View mattered to the analysis: where we find the kind of trying that matters to criminal responsibility, we also find an intention that commits the agent to each of the conditions involved in completion. The problem of solicitation-as-attempt did not require appeal to the distinction, developed in Part II, among the three different types of intention-based commitment, namely commitments of promotion, non-reconsideration, and non-complaint.

However, our resolution of the problem of “impossibility” in this Part has indeed required appeal to that tripartite distinction. A failure to appreciate that there are ways to be committed to a condition that do not amount to commitments of promotion can lead one to the mistaken conclusion that intent is too high a mens rea standard with respect to circumstantial elements in attempt. This failure can lead one to the conclusion, that is, that attempt does not require intention-based commitments to each of the elements of the crime attempted; the circumstantial elements, at least, are exempted. But the recognition that there are indeed forms of intention-based commitment that fall short of commitments of promotion allows us to see something new: the Guiding Commitment View’s implication that attempt requires intent (even with respect to circumstantial elements of completion) is not just unproblematic, but in fact allows for an insight. It allows for the recognition that sometimes what your intentions commit you to is partly a matter of what is in fact the case.

In short, it is because of the tripartite division among types of intention-based commitment that it is defensible to view intent as the appropriate mens rea standard with respect to circumstances in attempt. That tripartite division also leads to the conclusion that a rebuttable presumption of mens rea is generated by both (a) belief that the circumstance is in place or (b) recklessness with respect to the circumstance when, in fact, the circumstance is in place.

VI. abandonment and change of mind

Recall how we got here. We started with the idea of the Transfer Principle: attempts are crimes because they are implicitly prohibited in our prohibitions of completed crimes. This insight provided us with a constraint on an account of what, exactly, attempts are in the legally relevant sense: they must be such as to inherit their criminality from completion. This constraint, in turn, led to the formulation of the Guiding Commitment View, an account of attempt that is both intuitively appealing and supports the result that attempts are implicitly prohibited whenever completions are. The Guiding Commitment View, informed as it is by recent work on intention, allows principled resolution of the problems of solicitation-as-attempt and (so-called) “impossibility.” But as we will see in this Part, the Transfer Principle allows for resolution of at least one thorny problem about attempt, namely the problem of abandonment, even without appeal to the Guiding Commitment View and its associated theory of the nature of the commitments constituted by our intentions. The Transfer Principle, that is, is independently fruitful. Just by recognizing that the reasons for criminalizing attempt all derive from the reasons for criminalizing completion—an implication of the Transfer Principle—we are able to see under what conditions abandonment is relevant to criminal responsibility for attempt, and how.

A. The Problem

Consider Arin Ahmed, a Palestinian who tried to kill a number of Jews in a public place through a suicide bombing but changed her mind before the plan was complete.64 Here is how Ahmed describes how it came to pass that she stopped herself before anyone was hurt:

I got out of the car. . . . I saw a lot of people, mothers with children, teenage boys and girls. I remembered an Israeli girl my age whom I used to be in touch with. I suddenly understood what I was about to do and I said to myself: How can I do such a thing? I changed my mind.65

There is no doubting the prevalence and strength of the intuition that someone like Ahmed is importantly different, and ought to be treated differently under the law, from someone who got out of the car intent on a suicide bombing and was wrestled to the ground by police before any damage could be done. The fact of change of mind seems intuitively to be of tremendous ethical importance to responsibility for attempt. In fact, the Model Penal Code gives expression to this idea in its affirmative defense of abandonment. Under Model Penal Code § 5.01(4), a person who abandons an attempt (under certain conditions that appear to be met in Ahmed’s case) has an affirmative defense from the charge.66 Twenty-six jurisdictions in the United States follow the Model Penal Code approach in allowing an affirmative defense of abandonment,67 and some commentators who oppose it favor, instead, granting mitigation rather than a complete defense to those who, like Ahmed, abandon their attempts for laudatory reasons.68 It is very hard these days to find anyone defending the common law position that abandonment is of no relevance to guilt for attempt. Even judges in jurisdictions that have never granted a defense or mitigation on the basis of abandonment tend to avoid asserting that none is to be granted as a matter of law in their jurisdictions. For instance, while there is officially no abandonment defense in federal law, federal judges faced with the question tend to do their best to avoid the issue by claiming that even if abandonment were a defense or provided mitigation, the defendant before them provided insufficient evidence of abandonment.69 In fact, we see this even in cases in which the evidence of abandonment would likely suffice for defense or mitigation in a jurisdiction that explicitly recognizes defense or mitigation in cases of abandonment.

Moral intuition (not to mention legal codes, like Model Penal Code § 5.01(4)) supports the further idea that the defendant’s motive for abandoning his attempt is of crucial importance to the question of whether to punish the defendant or to reduce his punishment. To give the most obvious example, a defendant who abandons his attempt because he discovers that if he goes through with the crime he will be caught—perhaps a police car happens to be passing by unexpectedly—does not deserve a defense or mitigation. After all, he is in no better a moral position than the defendant who is stopped through frank physical force by the police; the fact that he stops himself is irrelevant. There are other examples, too. A defendant who stalks a victim intending a robbery, but decides to wait until later to complete it when he comes to believe that if he waits he can commit the robbery after the victim has visited an ATM, does not deserve a defense or mitigation. Broadly speaking, we recognize various impure motives for abandoning—motives importantly different from Ahmed’s motives—that undermine whatever mitigating force the abandonment has. But why is this? Whatever rationale we give for granting a defense or mitigation on the basis of abandonment must support the idea that the motive for abandonment matters; a bad motive for abandoning can undermine abandonment’s mitigating force.

In addition to accounting for the fact that the motive for abandoning matters, a principled account of how the courts should respond to abandoned attempts must be consistent with the fact that abandonment of the kind that is puzzling happens after the defendant has attempted a crime. To be sure, there are cases in which the defendant abandons before he’s even tried to commit the crime; imagine the person who intends to rob a bank, has a hearty breakfast in preparation, and then changes his mind. But the explanation for why criminal penalties are inappropriate in cases like that does not appeal to the fact of change of mind. Instead, it points to the fact that the defendant has not done enough, or has not done the right kind of thing, to count as having attempted the crime at all.70 The cases where special conceptual resources are needed are those in which the defendant has committed an attempted crime but changed his mind before completing it. In this way, abandonment sits in the same temporal relation to the charged crime as remorse: it happens after that which we take to be criminal. But abandonment of attempt seems to many people to be relevant to responsibility in some way that is entirely different from the way in which remorse is relevant. (No jurisdiction offers an affirmative defense of remorse, not even those that offer an affirmative defense of abandonment.) This must be explained by an adequate account of the relevance of abandonment to responsibility for attempt.

One initially appealing account of the relevance of abandonment is incapable of accounting for some of its central features. Consider the claim that we give those who abandon a punishment discount—100% off under the Model Penal Code, which allows an affirmative defense—in order to incentivize abandonment.71 Since the police can’t stop everyone, we want to give people incentives to stop themselves. So we want someone who has attempted a crime to see that he has not thereby earned himself the punishment for completion; he can still avoid that punishment by abandoning. There are two problems with this line of thought: first, in a regime in which failed attempts are punished less heavily than completions, as is the case for the vast majority of crimes in the vast majority of jurisdictions in the United States, defendants already have incentives to abandon; by abandoning, they can avoid the enhanced penalty for completion. This implies that it is at least an open empirical question whether any more attempts are abandoned in a regime that allows mitigation or an affirmative defense for abandonment than are abandoned thanks to the punishment discount for falling short of completion. Second, and more importantly, this rationale is not available to those who think, as they should, that abandonment mitigates only when it is motivated in the right way. Someone who is motivated to abandon by the prospect of a punishment discount is not properly motivated.72 Such a person is really no different from one who abandons because he thinks if he follows through he will be caught. So this intuitively appealing aspect of legal doctrine cannot be explained in its entirety by appeal to the hope to incentivize abandonment.

How is it best explained? As we will see, the best explanation implies that abandonment ought to mitigate a defendant’s sentence but ought not provide an affirmative defense.

B. Mitigating Factors and the Ceiling on Attempt’s Sanction

The right way to think of the relevance of abandonment arises from appreciating the grounds for the following undeniable fact: it is illegitimate to punish a failed attempt more heavily than it would have been punished had it been completed. This is not to say that there are no attempted murders, for instance, that are rightly sanctioned more heavily than some completed murders. Rather, it is to say that no attempted murder is rightly sanctioned more heavily than that very murder would have been rightly sanctioned had it been completed. The best rationale for this is that every reason to sanction attempt is a reason to sanction completion, even if there are additional reasons to sanction completion that are not reasons to sanction attempt. Only if there were reasons to sanction attempt that were not reasons to sanction completion would it be acceptable to give a greater penalty to an attempt than to completion of that attempt. It is an implication of the fact that attempts are criminalized under the Transfer Principle that there can be no reasons to punish an attempt that are not reasons to punish completion of that attempt. So the appealing idea that the punishment for completion sets the ceiling for the punishment of attempt is an implication of the Transfer Principle.

The fact that reasons for sanctioning attempt are all reasons for sanctioning completion of the attempt also implies that many mitigating factors for the sentence for completion are also mitigating factors for the sentence for attempt. Seeing this requires recognizing that often a factor mitigates thanks to the fact that it silences a reason for sanctioning. If a reason for sanctioning completion is silenced by the presence of a mitigating factor, and the silenced reason is also in general a reason for sanctioning attempt, then the mitigating factor also mitigates sentence for attempt. I explain.

Mitigating and aggravating factors are never decisive. When such a factor is present, there is a reason for issuing a lower or higher sentence, or one fewer reason not to; the reasons in question rarely settle the question of what sanction to give. Hence a mitigating factor could be present and yet it might make sense to give the defendant a greater than typical sentence because there are several aggravating factors present also. Similarly, a mitigating factor might be present, but there may be a reason for sanctioning that is in no way lessened by the presence of the mitigating factor and which supports issuing a sanction of typical size. In such cases, a mitigating factor is present but makes no difference to the final sentence. In this way, mitigating and aggravating factors are importantly different from either justifications or excuses. When a defendant has a justification—he injured another in the execution of a public duty, for instance—that settles the matter against punishment of the defendant.73 We do not weigh the justification against, for instance, the suffering of the victim, or the dangerousness of the defendant; a justification deflects criminal liability entirely. Similarly with excuse. Mitigating and aggravating factors, however, are different. They merely figure into the calculus of reasons for sanction and are therefore subject to all the vicissitudes of reasons; they can be present without changing the outcome, or absent even when the outcome is to be changed.

Affirmative defenses, note, are all-or-nothing. When they are present, they shield a defendant completely from criminal liability. This is why they are a useful legislative tool for providing justified and excused defendants with what they deserve. If a defendant acts in self-defense, for instance, the best way to give him what he deserves is to provide him with a way of avoiding criminal liability altogether. But an affirmative defense is the wrong tool to use when the facts in question mitigate; to grant an affirmative defense when a mitigating factor is present is to provide defendants more than they deserve, for mitigating factors, by their nature, can be outweighed, at least potentially, by other, aggravating factors. The conception of a factor as having some degree of weight, and thus possibly being outweighed, is not available with an affirmative defense.

Mitigating factors (and perhaps aggravating factors, too) can potentially function in at least two importantly different ways. First, they can provide positive support for giving a lower sanction. Considerations of mercy function in this way. If giving a lower sanction would be merciful, and if there are reasons to be merciful, then there is a positive reason to give a lower sanction than would otherwise be appropriate. Second, and more important for our purposes here, mitigating factors can cancel the force of a reason to give a particular sanction (say, a typical sanction) rather than a lower one. For instance, if the fact that a crime was the defendant’s first offense mitigates at all, it mitigates in this way.74 Sometimes a reason in support of issuing a particular sanction rather than a lower one is the fact that the behavior punished is part of a pattern of criminal conduct by the defendant. This reason—if it is one—for issuing a particular sanction rather than a lower one is muted by the fact that the defendant’s crime was not part of a pattern at all. It will be argued here that abandonment mitigates in this second way.

C. Why Abandonment Mitigates

Judges should and do pursue multiple goals in sentencing. They ought to seek to deter the defendant from future crime, whether in or out of prison. They ought to have their eye on the deterrence of other people besides the defendant. Furthermore, they ought to consider whether the sanction is sufficient to express solidarity with the victim in his loss. They ought to consider what the defendant deserves and his prospects for, and need for, fundamental rehabilitative change. There is more besides. But there is one particular goal that, it will be argued, is of relevance to our discussion here: judges ought to pick the minimum sanction that would provide, by the defendant’s own standards at the time of the crime, a sufficient reason for him to have refrained. That is, one of the things a judge should aim to do in sentencing is to give a sanction harsh enough that, had the defendant anticipated it at the time of the crime, he would have recognized sufficient reason to refrain. (More on this in a moment.) In general, however, judges have reason to give the lowest sanction that is supported by this and other reasons for sanctioning. That is, if there is just as good a reason to issue one sanction as another, then there is an additional reason to issue the lower of the two. Putting these two things together, we reach the result that, if knowing that he would suffer a year in prison would have prompted the defendant to recognize sufficient reason to refrain, then that is a reason to give a year rather than a year and a day. Conversely, if anticipation of a year would not have been sufficient to prompt awareness of sufficient reason to refrain, then there is reason to give a higher sanction.

When a defendant has abandoned an attempt for good reasons, he has already recognized sufficient reason not to complete the crime without anticipation of any sanction. Therefore, when the defendant has changed his mind, one reason to give a sanction of some particular size is nullified. When a defendant has abandoned an attempt, a judge cannot say, in favor of his decision to give the defendant a year in prison, that anticipation of any lower sanction would not have prompted the defendant to recognize sufficient reason to refrain. Since the defendant recognized sufficient reason to refrain without anticipation of sanction, such a claim would be untrue. Other reasons to sanction remain, however—for example, the defendant is deserving of sanction for having wrongfully tried to commit a crime, deterrence goals will be furthered by sanctioning, the judge needs to express solidarity with those targeted by the defendant—and so the fact of abandonment mitigates the sentence by silencing a reason in favor of any given sentence without thereby silencing all such reasons.

Why should the sentence be such as to provide sufficient reason to have refrained from the crime? One rationale for this is rooted in deterrence. The mechanism of deterrence is engagement with the defendant’s mechanisms for directing his own conduct; conditions are arranged so the defendant will direct himself away from crime. On the not unreasonable assumption, common in economics, that the mechanisms through which people direct their behavior are guided by their conceptions of their reasons for action, deterrence then requires sanctions harsh enough to provide people with sufficient reason, by their own lights, to refrain. But this is not the only possible rationale. Those who hold that the root of criminal culpability is misuse of the distinctive human capacities for recognizing and responding to reasons75 will see a different reason to avoid a sanction so low that the defendant would have recognized sufficient reason to commit the crime even if he had anticipated the sanction. Such a sanction would fail to place the defendant in anything like the place of the law-abiding citizen, who would recognize sufficient reason to refrain from crime even without anticipation of sanction. So there are multiple perspectives from which we might justify issuing a sanction large enough that anticipation of it would have prompted the defendant to recognize sufficient reason to refrain from the crime. There are therefore multiple perspectives under which there is good reason to mitigate a sentence because one abandoned an attempt.

Notice that the argument just offered for treating abandonment as a mitigating factor in the sentencing of attempt appeals, implicitly, to the fact that the reasons for sanctioning an attempt are all of them reasons for sanctioning completion (a claim that follows from the Transfer Principle). After all, while remorse after the commission of a crime might be a reason to give a lesser sentence, it is not a mitigating factor thanks to the fact that the defendant would have prospectively recognized sufficient reason to refrain even had he anticipated the lower sanction. For all we know, the remorseful defendant was prospectively aware of the sanction and did not take it to provide him with sufficient reason to refrain; not until after completion did he realize that he shouldn’t have committed the crime. Why not say the same thing about the abandoned attempt? After all, abandonment happens after the defendant’s crime, namely the attempt, has already taken place. The answer is that abandonment tells us something about the appropriate sanction for the completed crime the defendant attempted. The completed crime would have deserved mitigation since the defendant recognized sufficient reason to refrain from it without considering the typical sanction for the crime. Furthermore, since the reasons to punish attempt derive from the reasons for punishing completion, abandonment mitigates the attempt. It does so for distinctly different reasons from those that would support mitigating in light of remorse.

This point can be put in another way by responding to an objection. It was acknowledged earlier that there can be reasons for sanctioning completion that are not reasons for sanctioning attempt. For instance, we might sanction completion as a way of expressing solidarity with an injured victim. This would not be a reason to sanction attempts of that crime that fall short of injuring anyone. Here we have our objection: how do we know that the reason for sanctioning completion, which is cancelled by the fact of abandonment, is not of this kind? How do we know, that is, that the silenced reason is also a reason to sanction attempt? The answer ultimately comes from the Transfer Principle. To see the answer, first consider the relevance to attempt of the fact that there are victims of the completed crime. If we ask why we criminalize murder, for instance, the harm to murder victims is a central part of the answer. But as I indicated in the discussion of the Transfer Principle in Part I, if we ask why we criminalize attempted murder, the harm to murder victims, a harm not realized in the attempt, is still a central part of the answer. That observation, in fact, is central to appreciating the role of the Transfer Principle in the law of attempt. So while it is true that the expression of solidarity with victims is not a reason to sanction attempts that have no victims, it is also the case that the fact that the completed crime has victims does provide reason to sanction attempt; the reason just is not the expression of solidarity with the victims, but something else. Similarly, part of the reason to criminalize completed crimes is that they spring in part from a failure to recognize sufficient reason to refrain from the crime. And part of the reason to criminalize attempts is that completions involve that failure. Therefore, if this reason to sanction completion is silenced, as I’ve suggested it is when there is abandonment, then that reason for sanctioning attempt is similarly silenced.

The further and final point worth emphasizing in this section is that we also now have at hand an argument against granting an affirmative defense of abandonment. As indicated earlier, to grant an affirmative defense in the face of a particular condition’s being met—such as that the defendant abandoned—is appropriate only if that condition undermines the rationale for assigning criminal liability. An affirmative defense might also be appropriate if the relevant condition cancels all positive reasons to sanction. But an affirmative defense is the wrong tool if the condition in question provides one reason against sanction, but potentially leaves in place many others. Such is the case with abandonment. It cancels a reason for sanctioning—that is, failure to recognize sufficient reason to refrain—but does not cancel all such reasons. Hence it should be treated as a mitigating factor only; there should be no affirmative defense of abandonment.

D. The Motive for Abandoning

The rationale for mitigation on the basis of abandonment supplied here explains why the motive for abandonment matters. In brief, the reason for this is that when the defendant abandons from some objectionable motive, then the argument for abandonment-based mitigation fails to go through. To see this, we need to consider each of the various types of objectionable motive for abandonment in turn.

The Model Penal Code offers an account of the various motives for abandoning that undermine abandonment’s mitigating force. Under the Model Penal Code, the abandonment—what the code chooses to call the “renunciation of criminal purpose”—must be both “voluntary” and “complete” to permit mitigation. But the Code defines these terms in such a way as to identify certain motives for abandonment that undermine abandonment’s mitigating force. The relevant section of the Code reads:

[R]enunciation of criminal purpose is not voluntary if it is motivated, in whole or in part, by circumstances, not present or apparent at the inception of the actor’s course of conduct, that increase the probability of detection or apprehension or that make more difficult the accomplishment of the criminal purpose. Renunciation is not complete if it is motivated by a decision to postpone the criminal conduct until a more advantageous time or to transfer the criminal effort to another but similar objective or victim.76

So the Code identifies roughly four kinds of motives for abandonment that undermine mitigation. There is no mitigation if the defendant abandons because (a) there is a better chance he will get caught than he expected; (b) completion is harder than he expected; (c) there will be a better time or place for the crime; or (d) there’s a better victim for the crime.

Consider, first, defendants who abandon for the reasons described under (a). The bank robbery, for instance, is abandoned after the defendant walks into the bank, armed and ready to hold up the teller, but notices that a security camera has been installed since he cased the bank. In cases of this sort, the abandonment is predicated upon completion being punished in a way that is typical. The reason that the new security camera provides the defendant with a reason to walk away is that the defendant expects that completion would result in a certain punishment that he has a strong desire to avoid. For all we know, were he to anticipate a smaller than typical punishment, then the added risk of being caught, given the unanticipated camera, would not outweigh the possibility of filling his sack with money. Thus, by abandoning, he has not shown, as the argument for abandonment-based mitigation requires, that he would recognize sufficient reason not to complete the crime even given a lower-than-typical sanction for completion. For all we know, a lower-than-typical sanction would not, given his motives for abandoning, provide him with sufficient reason to refrain and so the argument for abandonment-based mitigation in his case loses its force.

Now consider cases in which the motive is of the sort described under (b), in which completion is more difficult than expected. The bank robbery, for instance, is abandoned at the same point as in our previous example (after the defendant has walked into the bank, armed and intent on robbing it). But in this case, the defendant abandons because he notices that the vault doors are closed, where they were open when he cased the bank, and so the robbery would require him to persuade a bank employee to open the doors for him. There is no reason to believe that the defendant would have recognized sufficient reason to refrain from completing the robbery had he anticipated a smaller than typical sanction. The reason is that anytime anyone undertakes any complicated task, such a task is thought to be worthwhile only if the rewards outweigh the costs that must be borne in order to earn them. For all we know, this defendant included the typical sanction (corrected for the probability that he would have to suffer it) in his calculations before he walked into the bank. The added work of having to open the vault doors tipped the scales against the robbery. But were a cost of the robbery lowered—were a lower-than-typical sanction for completion promised—then it is quite possible that the added work to open the vault doors would be worth undertaking. As in the previous hypothetical, for all we know, a lower than typical sanction would persuade the defendant that he did not, after all, have sufficient reason to abandon. Like before, then, the argument on offer in this Part for abandonment-based mitigation fails when a defendant abandons for the reasons described under (b).

Similar considerations apply when the motives to abandon are like those in (c), in which the defendant realizes that there will be a better time or place to complete the crime, and (d), in which the defendant realizes that there will later be a better victim. Consider someone who abandons the bank robbery when he realizes that a large daily deposit has been delayed and that it will therefore be worth a lot of money to him to rob the bank later. Such a person decides that the risk in this instance is not worth the reward, but that the risk will be worth the reward at a later time. But if robbing the bank now promised a smaller than typical sanction, then for all we know, the risk involved in robbing the bank now would be worth the reward to be expected from robbing it now; after all, less would be risked if the expected sanction for robbing the bank now were lower than typical, and so less would be gained by waiting to rob the bank later. Again, the promise of a lower than typical sanction, given the motive for abandonment, might actually prompt the defendant not to recognize sufficient reason to abandon. Similarly, the defendant who abandons his attempt to rob Bank X because he decides he will be better off robbing Bank Y instead (a case involving a motive of type (d)) sees attractions in robbing Y that outweigh the costs associated with the robbery, including the anticipated sanction (corrected for the probability of being apprehended). But if the robbery of X were to promise a lower than typical sanction, then there might not be sufficient reason to switch targets.

In all these hypothetical cases, the motive for abandonment shows a sensitivity on the defendant’s part to the typical sanction. The defendant is weighing pros and cons and deciding to abandon given an assumption on his part that there’s a chance he’ll suffer the typical sanction should he complete the crime. Thus, when we have these motives for abandonment, it is also true that had the defendant known before he committed the crime that he would suffer a certain smaller than typical sanction for committing it, then he might have found sufficient reason not to abandon his attempt. We therefore have no reason to think that anticipation of a smaller than typical sanction would have been sufficient for awareness of sufficient reason to refrain. As a result, when the defendant has abandoned because of the motives the Model Penal Code identifies, the reason for giving a particular sanction rather than a lower one still applies; it is not silenced by the fact of abandonment. So the argument for abandonment-based mitigation does not go through.

Notice that Arin Ahmed’s reasons for abandoning are not of the sort that are sensitive to the expected sanction for the completed crime. That sanction—if it is even possible to imagine what kind of sanction there could be for such behavior, perhaps a penalty issued to one’s surviving family members—plays no role in her reasoning; it is not something to be outweighed by other considerations but is simply trumped by the reasons that she had for abandoning. Hence, even if the sanction were smaller, she still would have abandoned her attempt. What follows is that in her case, the argument for abandonment-based mitigation goes through, and so her abandonment is a legitimate mitigating factor.

Could there be other motives for abandonment that undermine the case for abandonment-based mitigation, motives that do not fall into any of the four categories theModel Penal Code identifies? Yes, there could. Consider again the case of People v. Taylor,77 mentioned at the opening of this Article, in which the defendant’s intended rape victim talked him into believing that she would have consensual sex with him, and thus convinced him to stop trying to force her to have sex with him, before locking him out of her apartment. In Taylor, the court rejected the possibility of an abandonment defense under a statute derived from the Model Penal Code. The court did this on the ground that the defendant’s renunciation was not “voluntary and complete.” However, as noted at the opening of this Article, it is far from clear why it was not, under the Model Penal Code’s official definitions of those terms. The best case for thinking that it was not comes from construing the defendant’s motives as of type (c)—to construe him, that is, as having merely decided to postpone the rape until later, perhaps after the two returned from the liquor store. But it is hard to see why he would have stopped his attack if this were his motive. After all, if he were intent on rape, he could have completed it despite the victim’s efforts to persuade him that he could later have consensual sex with her. Rather, it seems that Taylor forced his way into the apartment with a preference for consensual over non-consensual sex, but intent on having sex with the victim regardless. When he became convinced that consensual sex was a possibility, he pursued it, perhaps anticipating that, should it stop seeming possible, he would return to his pursuit of non-consensual sex. However, even construed in this way, it is not unreasonable to think of Taylor as having changed his mind. His intention to have non-consensual sex was predicated on an assumption: consensual sex was not possible. When he came to believe this assumption was false, he changed his mind and abandoned his plan to have non-consensual sex with the victim. This isn’t to say that Taylor ought to enjoy abandonment-based mitigation. He ought not to. But why not? The Model Penal Code does not provide us with a tool for answering; only by distorting the facts are we able to construe Taylor’s motive for abandonment as one of the four sorts that undermine abandonment’s mitigating force under the Code.

The reason that Taylor’s sentence for attempted rape should not be mitigated is because, for all we know, Taylor’s preference for consensual over non-consensual sex was at least in part derived from the fact that non-consensual sex brings with it a certain punishment, the typical punishment for rape. So, if we ask whether someone who abandoned rape for Taylor’s reasons would have done so even if the penalty for rape were lower than typical, we find that the answer might very well be “no.” After all, had the penalty been lower than typical, then Taylor may have thought that the prospect of consensual sex, bringing with it the risk that he would not have sex with the victim at all, was not worth pursuing. He would abandon only if he thinks that consensual sex outweighs non-consensual sex, a calculation quite possibly made by considering, among other things, the typical penalty for non-consensual sex.

In short, the argument offered here helps us to identify the crucial question that must be answered in deciding whether to grant mitigation on the basis of abandonment: does the fact that the defendant abandoned his attempt show that the defendant would have been dissuaded from completing the crime even if a lower than typical sanction were given for completion? The answer is no when the defendant abandoned for any of the four types of reasons identified by the Model Penal Code. But the answer is no also when the defendant abandoned for reasons like Taylor’s, even though those reasons do not fall into any of the Model Penal Code’s four categories. Legitimate abandonment-based mitigation is hard to come by, even harder than under the Model Penal Code, since there are motives for abandonment that undermine its mitigating force but that are not identified as such by the Code.

E. Taking Stock

This Part began by identifying an implication of the fact that attempts are properly criminalized thanks to the fact that completions are: the reasons for sanctioning an attempt derive from reasons for sanctioning the corresponding completed crime. This allowed us to identify a class of factors that mitigate sentence for attempt—those that silence a reason for issuing a particular sanction, rather than a lower one, to the corresponding completed crime. By silencing a reason for issuing a high sanction for completion, we thereby necessarily silence a reason for issuing a high sanction for attempt, given the Transfer Principle. In general, the fact that a defendant prospectively recognized sufficient reason to refrain from the crime, without considering the typical sanction for that crime, silences a reason to issue a typical sanction rather than a lower one. This reasoning supports mitigating sentence for the attempt, even though abandonment occurs after the attempt, thanks to the special relationship between the reasons for sanctioning attempt and the reasons for sanctioning completion.

Conclusion

This Article offers a theoretical framework for thinking about criminal attempts and draws doctrinal implications from that framework. The theoretical framework consists, in essence, of three components: (1) an account of the grounds for the criminalization of attempts, labeled the Transfer Principle, according to which attempts are implicitly criminalized when we criminalize completed crimes; (2) an account of a wider range of intention-based commitments than has heretofore been recognized; and (3) an account (drawing on (2)), labeled the Guiding Commitment View, of the necessary and sufficient conditions for being a criminal attempt of the sort that is implicitly criminalized when we criminalize completed crimes. To attempt a crime is to be committed by one’s intention to each of the components of success and to be guided by that intention.

As we have seen, this theoretical framework allows principled resolution of three seemingly intractable problems that arise in the adjudication of attempt. Three particularly important doctrinal recommendations arise from this resolution: (1) a solicitation rises to the level of a criminal attempt only if the defendant asked another to realize a result element of the completed crime, and not if he asked another to realize an act element. (2) If the defendant believed or intended that the circumstantial elements of the completed crime were in place, then they do not need to be shown to have been in place for conviction for attempt; if, however, the defendant was aware only of the possibility that the circumstantial elements were in place, then the prosecution must show them to have been present to establish the prima facie case for the attempt. Finally, (3) abandonment of an attempt for reasons that show the defendant to have recognized sufficient reason to refrain from the crime without consideration of the prospect of sanction ought to mitigate sentence for attempt and ought not to constitute an affirmative defense. If adopted, all three of these recommendations would bring our practices of adjudicating criminal attempts in line with the principles that animate their criminalization in the first place.

The philosopher Peter Strawson, writing about free will, insisted that in thinking about that topic, we should remind ourselves of some “commonplaces”:

We should think of the many different kinds of relationship which we can have with other people—as sharers of a common interest; as members of the same family; as colleagues; as friends; as lovers; as chance parties to an enormous range of transactions and encounters. Then we should think, in each of these connections in turn, and in others, of the kind of importance we attach to the attitudes and intentions towards us of those who stand in these relationships to us, and of the kinds of reactive attitudes and feelings to which we ourselves are prone . . . .

The object of these commonplaces is to try to keep before our minds something it is easy to forget when we are engaged in philosophy, especially in our cool, contemporary style, viz. what it is actually like to be involved in ordinary inter-personal relationships, ranging from the most intimate to the most casual.78

As Strawson saw it, the necessary conditions of moral responsibility are those that we find implicit in indispensable social practices of interaction and engagement with others. He urged us, in thinking about what human beings must be like such that we can be held morally responsible for our behavior and hold others responsible, to focus on the varied and complex facts about others and ourselves to which we respond as part of social life, and to which certain responses and not others are particularly appropriate. It is because of its role there that, for instance, intention is so important to responsibility. It is one of the things to which we respond with, among other things, resentment and gratitude; it is one of the things to which some responses, including but not limited to moral emotional responses, and not others, are appropriate.

It is not just this that “our cool, contemporary style” can lead us to forget. It can also lead us to forget the degree to which the assignment of criminal responsibility ought to be, even if it is too rarely, itself an entrenched and integral part of the web of social interactions to which we are necessarily subject thanks to living together in a state. We need an account of what is and should be involved in the assignment of criminal responsibility that demonstrates and exhibits this entrenchment. We must recover the sense in which the assignment of criminal responsibility is an essential part of participation in the many and varied relationships we have with each other as citizens.

The study of criminal responsibility for attempt is not special in this regard. But it is one place in which the theorist is likely to despair, as Jerome Hall did,79 of the possibility of providing an account of what we are, or should be doing, that bears deep links to the simple interactions of everyday life. We are prone to conceive of the activity as nothing more, for instance, than the post hocrationalization of the zealous desire to intervene earlier and earlier in the lives of those who, for independent reasons if for any at all, we have classified as inhabiting the wrong side of the law. But this is just pessimism. The law of attempt is the simple interaction between father and daughter writ large. It bears more than an analogical relation to the prohibition of that form of trying that we prohibit implicitly when we prohibit conduct. In fact, it is that prohibition. Or, at least, so it has been argued here.