The VAR is a legal representation of a legitimate moral idea. There is something morally objectionable about violating it, as it would have been violated had the trial court’s conviction of Martin for appearing drunk in public not been overturned. The question is what, exactly, the underlying moral idea is. As we will see, many of the valid moral principles that one might take the VAR to legally enshrine fail to capture important aspects of it. Further, it will be argued that, perhaps surprisingly, the moral idea that in fact justifies the VAR is the familiar one that what qualifies wrongful conduct for criminal punishment is the way in which that conduct manifests the defendant’s objectionable mental states, such as bad intentions.

Section 1 explains some of the features of the VAR that need to be accounted for by a moral rationale for it. Section 2 discusses several attractive rationales and argues that each falls short of providing a fully satisfactory account. Section 3 offers an alternative, drawing on an appealing picture of the morally relevant relationship between criminal conduct and the mental states of the criminal, a relationship which, it is argued, is at risk of being absent when the VAR is violated.

The Legal Doctrine

Although it is rarely made explicit, the law uses a definition of the term “voluntary act” that departs in significant respects from the ordinary concept. For legal purposes, a voluntary act is a willed bodily movement. To will, in the sense of relevance to the VAR, is not merely to mentally represent the bodily movement and for that bodily movement to be guided by that representation. If the mental state that guides the bodily movement is not “conscious”—a term explicitly used in the law—then that mental state is not a willing, or a volition, and the bodily movements it guides are therefore not voluntary actions. So the law provides two significant limitations in its definition of a voluntary act: only bodily movements count, and, of those, only bodily movements guided by conscious mental representations count. There are thus (at least) two kinds of voluntary action, in the ordinary sense, that do not count as voluntary actions in the law’s sense: (1) voluntary thoughts (e.g., voluntarily thinking about a problem); and (2) bodily movements guided by unconscious mental representations of them.

The second departure from the ordinary notion of a voluntary act—the exclusion from the category of bodily movements guided by nonconscious mental representations of them—is more puzzling and in need of independent explanation. To see that this second category includes behaviors that we would ordinarily characterize as voluntary, consider the case of People v. Newton (87 Cal. Rptr. 394 (1970)). After a traffic stop and an altercation with police, Newton was shot in the gut. Immediately following, he shot and killed one of the police officers and fled the scene, arriving shortly afterwards at a hospital. Later he claimed not to remember the shooting, or traveling to the hospital, and claimed to have been unconscious for the crucial period of time beginning from the moment at which he was shot. Medical testimony supported the contention: a doctor testified that people suffering from traumatic injuries often engage in complex, motivated bodily movements in the absence of consciousness. Newton was convicted at trial, but his conviction was overturned on the grounds that the mental states that guided his bodily movements—the finger movements on the trigger, for instance—were not conscious and so were not willings; but if they were not willings, then the relevant bodily movements were not voluntary, and so it would be a violation of the VAR to punish Newton for the officer’s death. Newton’s finger movements on the trigger were not likely to be purely reflexive; they were clearly goal-directed. Newton seems to have been aiming the gun at the officer, and so must have been mentally representing a particular goal, namely to shoot the officer, a mental representation that was involved in guiding his bodily movements. Many people would take such considerations to show that the relevant bodily movements were voluntary actions in the ordinary sense, but they were not voluntary in the law’s sense.

Morally justifying the VAR requires explaining what justifies the exclusion of behaviors that we would ordinarily classify as voluntary acts. In addition, however, there are events that we would probably not classify as voluntary actions—in fact, they may not even meet the law’s definition—but that the law treats as though they were. This is true, in particular, of omissions—failures to act—and of habitual actions. If a defendant is shown to have omitted a bodily movement (under certain conditions), or shown to have engaged in a habitual bodily movement, then that’s good enough to comply with the restriction imposed by the VAR.

This is not as peculiar as it might seem. In many places in the law we find that X, which is different from Y, is treated for legal purposes as though it were Y. Under, for instance, the doctrine of transferred intent in homicide, a person who intends to kill one person and, acting on that intention, kills another instead, is treated as though he intended to kill the person he killed. The law does not make the mistake of holding that the defendant in such a case really intends to kill the person he kills; everyone recognizes that the death in such a case was, in a sense, an accident. Rather, it is that the defendant is treated as though he had that intention since, it is thought, his actual intention was just as bad (or worse) than the one the law actually takes to be an element of an intentional homicide (namely, an intention to kill the person who was killed). Similarly, the law treats both omissions and habitual, often reflexive, acts as though they were voluntary acts for the purposes of the VAR, even though they may not be in fact.

A word about habitual actions is also in order. Consider a defendant who has been trained by the military to spin around and fire immediately, and without thinking, on a threat behind him. This behavior has become, thanks to his training, habitual. Is he to be held guilty of a crime when, at the local firing range, he spins and fires on a person behind him who yells something threatening? Or would such a verdict violate the VAR? The bodily movements in cases such as this are routinely taken to provide an acceptable basis for criminal liability (cf. Model Penal Code § 2.01(2)(d)). Noting that the defendant spun and fired only thanks to his military training will not serve to undermine the case against him through appeal to the VAR, although it might support a claim that he lacked some further mental state, such as an intention to kill, relevant to his degree of criminal liability. As suggested above, the law does not take this stance because it is believed that habitual actions are, in fact, willed. Nobody tasked with making legally binding decisions knows whether such actions are willed or not, or even whether some types are and some are not. It is an empirical question that would require careful investigation by psychologists and neuroscientists. Normally, our ignorance about a feature of pertinence to criminal liability is enough to supply reasonable doubt, and thus enough to support an acquittal. But not in habitual action cases. If there is reasonable doubt about whether the defendant’s bodily movement was voluntary deriving from the fact that it is habitual, that reasonable doubt fails to undermine the case for guilt. We treat habitual bodily movements, that is, as though they were voluntary acts in the legal sense, even though we have no idea whether they are in fact. This fact, also, about the VAR should be explained, ideally, by a moral justification of it.

Insufficient Rationales for the VAR

Broadly speaking, there are two different kinds of moral rationale that can be given for a criminal law doctrine that draws distinctions among defendants and then predicates differences in criminal punishment on the distinctions that it draws. Under the first kind of rationale, the doctrine is shown to draw a distinction on the basis of a property that makes a difference to desert of punishment. For instance, many offenses are graded by the mental states of the defendant. We punish more heavily someone who intentionally burns a neighbor’s barn out of spite than someone who negligently burns it by forgetting to put out a campfire. This doctrine is explained by the fact that intentional wrongdoing is morally worse than negligent wrongdoing.

The second kind of rationale appeals to a moral norm governing the behavior of those who are to inflict punishment. Ordinarily, for instance, we think it wrong to punish someone whom we lack good reason to think deserving of punishment even if he is deserving of punishment. To punish someone deserving without knowing for sure that he is deserving is not to give someone something other than what he deserves—if he’s deserving, he’s deserving—but it is to nonetheless do something wrong; it is to violate a norm governing the conduct of punishers. This moral norm—do not punish unless you know the punished is deserving—explains, for instance, some criminal law doctrines pertaining to evidence: we often punish only one of two equally deserving defendants when one, but not the other, has been proven to be guilty by the admissible evidence.

Two of the possible rationales for the VAR discussed in this section—called here the “Evidentialist” and “Actual Invasion” rationales—are of the second sort: they identify moral norms governing punishers and claim that those norms would be violated were the state to fail to follow the dictates of the VAR. The other two—what will be called the “PAP” and “Reasons–Responsiveness” rationales—are of the first sort: they identify features on the basis of which the VAR distinguishes among defendants that speak to a difference in desert of punishment.

The Evidentialist Rationale

The Evidentialist Rationale provides a satisfactory explanation for our allowance of criminal liability for omissions, even if they fall short of voluntary actions. What a person does not do sometimes tells us all we need to know about his desert-relevant mental states. It should be no surprise that the absence of sufficient concern for others, for instance, could be properly evidenced by the absence of action, at least in some circumstances; there are things that people who are sufficiently concerned would do, and so the failure to do them can evidence the absence of sufficient concern.

However, precisely what makes it possible to explain the law’s treatment of omissions under the Evidentialist Rationale undermines the case for thinking that that rationale captures the relevant moral principle enshrined by the VAR. Omissions can provide satisfactory evidence of desert-relevant mental states even if omissions are not voluntary acts. In fact, there are many sources of adequate evidence of such mental states other than voluntary acts. Involuntary actions tell us a great deal about mental state—think of blushing. In fact, because it is much harder to prevent involuntary manifestations of mental state than voluntary manifestations, involuntary actions sometimes provide significantly better