^1. Hart, H.LA., Acts of Will and Responsibility, The Jubilee Lectures of the Faculty of Law, University of Sheffield (Marshall, O.R. ed., Stevens & Sons 1960)Google Scholar, reprinted in Hart, Punishment and Responsibility (Oxford University Press 1968). Citations are to the reprinted edition.

^2. Hart, supra note 1, at 90.

^3. In legal theory, see Michael Moore, Act and Crime: The Philosophy of Action and Its Implications for Criminal Law (1993); in legal doctrine, see Model Penal Code §2.01.

^4. See Douglas Husak, The Philosophy of Criminal Law ch. 4 (1987); Husak, Does Criminal Liability Require an Act?, in Philosophy and the Criminal Law 60–100 (R.A. Duff ed., 1998); and Husak, Rethinking the Act Requirement, 28 Cardozo L. Rev. 2437 (2007). See also Antony Duff, Action, the Act Requirement and Criminal Liability, in Agency and Action (J. Hyman & H.C. Steward eds., 2004); P.F. Fitzgerald, Voluntary and Involuntary Acts, in Oxford Essays in Jurisprudence (A.G. Guest ed., 1961); and Simester, A.P., On the So-called Requirement for Voluntary Action, 1 Buff. Crim. L. Rev. 403 (1998)Google Scholar.

^5. See, John Austin, Lectures on Jurisprudence lecture 18, at 429 (R. Campbell ed., London, John Murray 1879). I do not mean to say that existing criminal codes embody, explicitly or implicitly, a specific philosophical account of action. For instance, the Model Penal Code declines to define what is meant by the “voluntary act or omission” required for liability, and proceeds by listing conditions–convulsion, unconsciousness, etc.–which, if proven, negate voluntary action. See Model Penal Code §§2.01(1) and (2). However, the challenge posed by this doctrinal structure is (1) to say more explicitly what unifies the various kinds of defeating conditions, and (2) to explain why the law should require voluntary action (however defined) in the first place. To the degree Anglo-American criminal theorists have taken up this task, they have traditionally done so by appeal to an account of voluntary action as willed bodily movement.

^6. Austin, supra note 5, at 424–425. There is extreme difficulty with this picture in actually getting the phenomenon of agency into view. Holmes is an instructive instance, e.g., when he writes that there is a “mysterious accuracy with which the adult, who is master of himself, foresees the outward adjustment which will follow his inward effort.” Oliver Wendell Holmes, The Common Law 54 (S.M. Novick ed., Dover, 1991).

^7. See Holmes, supra note 6, lecture 2.

^8. The tendency to run together the legislative and judicial functions of the act requirement is evident, for instance, in James Fitzjames Stephen's claim that “the general definition of crimes . . . is, that they are actions punished by law. Certain qualities are or are supposed to be common to all actions which the law punishes, and the existence of those qualities in the particular case is a necessary condition of criminality.” James Fitzjames Stephen, General View of the Criminal Law of England 75 (London, Macmillan 1863). It is, in my view, the failure to distinguish these two roles that explains much of the confusion still surrounding the act requirement.

^9. It is possible–though this is sheer speculation on my part–that the generalization would have seemed more plausible to theorists of Austin's era because of the rather more limited scope of criminalization at that period, focusing on crimes such as murder, rape, arson, and so forth, each of which not only typically requires but is essentially concerned with willed bodily movement. In contrast, contemporary substantive doctrine includes crimes of possession, regulatory offenses, and wide-ranging group and conspiracy crimes, where even if an act is “included” as an element of the crime, it is far from obvious that the act is the object of punishment. These developments in the criminal law may have contributed to making the act requirement seem much more questionable now than it perhaps once was. (I owe this suggestion to Bill Stuntz.)

^10. Holmes, supra note 6, at 75. My thanks to an anonymous referee for pushing me to clarify this point.

^11. Jeffrie Murphy, writing nearly forty years ago, describes Austin's theory and comments that:

Murphy, Involuntary Acts and Criminal Liability, 81 Ethics 332, 334 (1971). (To be fair, Murphy goes on to defend what he considers “a form of the classical theory” against Hart's criticisms.)

^12. See also Saunders, Kevin, Voluntary Acts and the Criminal Law: Justifying Culpability Based on the Existence of Volition, 49 U. Pitt. L. Rev. 443 (1988)Google Scholar.

^13. Moore, supra note 3, at 132.

^14. Hart, supra note 1, at 102.

^15. Id. at 101.

^16. Moore, supra note 3, at 161–162.

^17. As Meir Dan-Cohen notes, “the practical significance” of considering voluntariness to be an investigation of actus reus rather than mens rea is that involuntary acts will escape even strict liability offenses, where “no mens rea needs to be proved.” Dan-Cohen, Actus Reus, in Encyclopedia of Crime and Justice 18 (S. Kadish ed., 1983).

^18. An alternative would be to define volitions as: those things that most proximately cause bodily movements intentionally described from the agent's point of view. While not incoherent, such a view is about as informative as the view that opium causes sleepiness because it has a dormitive power; volitions cause actions because they are the things that proximately cause whatever it is people take themselves to be doing. Saunders comes, I think, perilously close to saying just this. (See Saunders, supra note 12, at 443, 456.)

^19. Austin, supra note 5, at 424.

^20. Perhaps it could be said that Moore only needs to show that every action is a volitionally caused bodily movement under some description; he does not need the further assumption that the legally relevant description of an action is that under which it is willed. I am agnostic as to whether every action has some description under which it is a willed bodily movement, but I do think that something like the further assumption is critical to Moore's endeavor. For if it cannot be shown that the legally relevant description of an agent's conduct–the description that provides the basis for punishment–somehow rests on a description of that same conduct as willed bodily movement, then what would be the legal interest in having such a description?

^21. Duff, supra note 4, at 69. Duff is here describing the traditional view, not endorsing it.

^22. I do not mean to suggest that generalizations can never justify; obviously, they do so in many cases (“April is usually a rainy month in these parts; it is April; therefore, it is likely to be rainy”). The point is rather that as a bedrock normative principle regulating the distribution of punishment, the act requirement should furnish a moral reason that justifies the pattern of liability it produces rather than simply being a shorthand description of that pattern of liability.

^23. See Rawls, John, Two Conceptions of Rules, 64 Phil. Rev. sec. 3 at 6 (1955)Google Scholar.

^24. See Simester, supra note 4, at 404.

^25. See Hart, H.LA., Legal Responsibility and Excuses, Determinism and Freedom (Hook, S. ed., New York University Press 1958)Google Scholar; reprinted in Punishment and Responsibility 39 (Oxford University Press 1968); and Rawls, supra note 23.

^26. The act requirement has not uniformly been understood as a principle rather than a generalization. See, e.g., Glanville Williams, Criminal Law 2, 13–15 (1961). But even when implicitly described as a generalization, it has nevertheless been claimed to have the role of a principle, in particular, as describing a necessary condition for responsibility. This vacillation is explained, in my view, by insufficient attention to the distinction between the role of the act requirement in its legislative and its judicial guises.

^27. This is particularly the case in U.S. federal criminal law. For constitutional reasons, the overt action in federal crimes is usually crossing a state line, use of the mails, etc., though what the person who is convicted under such a statute is being punished for is obviously not the act of crossing a state line or putting a letter in the mail. See, e.g., Schmuck v. United States, 489 U.S. 705 (1988) (mailing element of mail fraud satisfied by a third party's mailing of used car title application forms to the state).

^28. The literature on attempts is extensive. For a start, see Whybrow, (1951) 35 Crim. App. 141 (per Lord Goddard, C.J.); Duff, supra note 4, at 72; Husak, Philosophy, supra note 4, at 95–96; Alexander, Larry and Kessler, Kimberly, Mens Rea and Inchoate Crimes, 87 J. Crim. L. & Criminology1138 (1997)Google Scholar; and Wechsler, Herbert, Jones, William Kenneth & Korn, Harold L., The Treatment of Inchoate Crimes in the Model Penal Code of the American Law Institute: Attempt, Solicitation and Conspiracy, 61 Colum. L. Rev. 571 (1961)Google Scholar.

^29. This distinction, it should be noted, does not accord with the Model Penal Code. Model Penal Code §2.01(1) conditions liability on “a voluntary act or the omission to perform an act of which he is physically capable,” and §1.13(4) defines an omission as a “failure to act.”

^30. A similar problem arises in the context of possession. The Model Penal Code simply defines the problem away in its general statute on voluntary action: “possession is an act, within the meaning of this Section, if the possessor knowingly procured or received the thing possessed or was aware of his control thereof for a sufficient period to have been able to terminate his possession.” Model Penal Code §2.01(4). The “act” of possession, defined by reference to mens rea standards, is no longer “indifferent” in Holmes's sense. The strategy is, of course, legally sufficient but wildly unconvincing on the merits, since knowingly receiving X and/or having the ability to divest oneself of X is in no way plausibly construed as a “bodily movement whether voluntary or involuntary,” as the Model Penal Code defines an act. See Model Penal Code §1.13(2). The comment to §2.01 suggests that possession be dealt with as an application of prior-acts analysis, discussed below; see Model Penal Code §2.01 cmt. at 224. For further discussion, see Dubber, Markus Dirk, Policing Possession: The War on Crime and the End of Criminal Law, 91 J. Crim. L. & Criminology829 (2001)Google Scholar; and George Fletcher, Rethinking Criminal Law (Little, Brown 1978) 197–204.

^31. See Model Penal Code §2.01 cmt. at 217.

^32. Sometimes it is suggested that the prior act must be a culpable act. Glanville Williams writes, “when an act seems at first sight to be lacking it is sometimes possible for the court to look critically back through the accused's past until a culpable act is discovered.” Williams, supra note 26, at 13. Williams' example is of a robber whose gun goes off accidentally, thereby killing his victim; the culpable act “is not the act of pressing the trigger (which, in this instance, is absent) but the act of presenting the gun, which brings the case within the law of implied or constructive malice in murder.” Id. (Williams rightly points out that the availability of such a maneuver means that the act requirement is of “comparative unimportance.”) Williams's account faces difficulties in explaining why the robber should be punished for the shooting at all, if the only legitimate object of criminal responsibility is the prior act of presenting the gun; or why every armed robbery is not equally “within the law of implied or constructive malice in murder.” The Model Penal Code takes an arguably broader line than Williams in requiring not a prior culpable act but simply “conduct which includes a voluntary act or the omission to perform an act of which he is physically capable.” Model Penal Code §2.01(1) (emphasis added). Since not every criminally culpable nonact is premised on a prior culpable act (failure to pay taxes, etc.), the Model Penal Code version is of broader appeal for our purposes.

^33. Australia is the most notable example in the Anglo-American legal tradition. See Commonwealth Electoral Act of 1924, esp. §12. It is not entirely unknown in the United States: Georgia's 1777 Constitution, art. 12, penalized failure to vote with a monetary fine.

^34. Military Selective Service Act §462(a). To my knowledge, however, there have been no prosecutions since the 1980s for failure to register.

^35. For further and very useful discussion of the distinction between punishing “for” an act and punishing for conduct that merely “includes” an act (and some similarly skeptical remarks), see Husak, Rethinking, supra note 4.

^36. My thanks to Joshua Dressler and an anonymous referee for pushing me to clarify my thoughts in the preceding four paragraphs.

^37. But see Fletcher, George, On the Moral Irrelevance of Bodily Movements, 143 U. Pa. L. Rev. 1443 (1994)Google Scholar for important qualifications on these standard arguments against omission, thought, and status.

^38. 31 Ala. App. 334, 17 So. 2d 427 (1944). See also the infamous British case, Larsonneur, (1934) 24 Cr. App. R.74, in which a French national was charged with being found in Britain illegally although the only reason she was “found” there was because she had been carried there by the police.

^39. It is thus somewhat misleading to say, as an account of the act requirement, that “a person is morally responsible for an outcome unless the occurrence of that outcome is involuntary vis-à-vis that person.” Simester, supra note 4, at 406. For the sort of involuntariness in consideration is a more basic failure of agency altogether than the involuntariness of a person who acts in the shadow of a threat.

^40. Several theorists have suggested that what is really at issue is not the act itself but the voluntariness of that act. P.F. Fitzgerald, writing contemporaneously with Hart, suggests that the underlying rationale behind the act requirement is that we should not punish involuntary acts and that the reason for this is that, “without inquiring too deeply into the moral justification for punishment, we may recognize the existence of a moral principle that we should not blame or punish one who could not help doing what he did.” Fitzgerald, supra note 4, at 18. Simester sounds a similar note when he writes that the “foundation of moral responsibility is not action but voluntariness; or, strictly, the absence of involuntariness.” Simester, supra note 4, at 411. Douglas Husak, in contrast, explicates the intuition not in terms of voluntariness per se but in terms of a person's control over a state of affairs. Husak has long been explicit in his dissatisfaction with the act requirement and defends the claim that an agent's control over a resulting state of affairs better tracks our considered moral judgments regarding when it is appropriate to hold someone criminally responsible. In Husak's view, “a person lacks responsibility for those states of affairs he or she is unable to prevent from taking place or obtaining.” Therefore, Husak contends, “control is more plausibly regarded as a condition of both moral and criminal responsibility.” Husak, Does Criminal Liability Require an Act?, supra note 4, at 77. Finally, Michael Corrado defends the view that “there is indeed a voluntary act requirement, and . . . the essence of it is that the actor must have been able to avoid choosing to break the law. She must have been able to control her choice.” Corrado, Is There an Act Requirement in the Criminal Law?, 142 U. Pa. L. Rev. 1529, 1557 (1994).

^41. Is practical reasoning broadly speaking “cognitive”? Are the conclusions of practical reasoning themselves motivating or do they require supplementation by desire or some other kind of “pro-attitude”? Or are the conclusions of practical reasoning the actions themselves? While these issues (and others) are central to the contemporary philosophical debate about practical rationality, I am cautiously optimistic that the practical-agency condition does not presuppose any particular theory of practical rationality, at least so long as that theory allows us to describe agents as considering and selecting among alternatives and acting on the basis of those decisions.

^42. I do not mean to suggest that one can necessarily draw a sharp distinction between judgment and agency; indeed, I find such a suggestion rather implausible. But even if certain kinds of judgment are intrinsically action-motivating (and thus do not need supplementation by a further pro-attitude), the distinction I am drawing can nonetheless still be made out; for it is the distinction between a defect in the person's grasp of reasons leading to his forming bad action-motivating judgments, and a person not being in a position to form such judgments in the first place.

^43. See Model Penal Code §§2.01(b) and (c). See also People v. Newton, 8 Cal. App. 3d 359 (1970) (appealing conviction on basis of automatism as a reflex response); Fulcher v. State, 633 P. 2d 142 (Wyo. 1981) (appealing conviction on basis of unconsciousness); People v. Higgins, 5 N.Y. 2d 607 (1959) (appealing conviction on basis of seizure-induced conduct); and Corrado, Michael, Automatism and the Theory of Action, 39 Emory L.J. 1191 (1990)Google Scholar (developing incompatibilist theory of voluntariness for the criminal law).

^44. “The agent may be asleep, yet the language, not just of action, but of purposive action, be appropriate beyond dispute.” Williams, Bernard, Voluntary Acts and Responsible Agents, 10 Oxford J. Legal Stud. 1 (1990)Google Scholar. See also Williams, , The Actus Reus of Dr. Caligari, 142 U. Pa. L. Rev. 1661, 1663 (1994)Google Scholar, in response to the “metaphorical” theory developed in Moore, supra note 3.

^45. “Where there is evidence of the existence of that state of the mind wherein the individual's conscious mind has ceased to operate and his actions are controlled by the subconscious or subjective mind it would be error to refuse instructions as to the legal effect of such unconsciousness.” People v. Sameniego, 118 Cal. App. 165, 173 (1931).

^46. Perhaps a start–but only a start–would be to unpack executive control in terms of the motivations for a person's actions. They must be, one is tempted to say, one's own reasons and/or desires; i.e., they must be motivating because one has judged them to be worth pursuing. For one development very roughly along these lines, see Richard Moran, Authority and Estrangement (2001). See also Frankfurt, Harry, Freedom of the Will and the Concept of a Person, 68 J. Phil.5 (1971)Google Scholar. It is also not, at least in the realm of the law, a purely philosophical inquiry; there are empirical questions about how well we understand shock, automatism, somnambulism, and the like, and what sorts of rational self-guiding capacities agents so afflicted are likely to have.

^47. Decina, 2 N.Y. 2d 133 (1956). See also Fulcher, 633 P. 2d 142 (Wyo. 1981). It is cases like Decina that motivate the prior-acts analysis discussed in Sec. II.

^48. Suppose that Decina gets a peculiar sensation in his temples exactly five minutes before he has a seizure, and that this sensation is a completely accurate predictor of his seizure. Now he is driving along; he notices the sensation; he fails to move his foot off the accelerator; as a consequence, he kills several children. I see no reason to suppose that the result should be any different in this case from in the actual case.

^49. The practical-agency condition also has the doctrinal advantage of dealing more consistently with possession crimes than the Model Penal Code's somewhat ad hoc definition–required by their reliance on voluntary acts.

^50. Husak, Does Criminal Liability Require an Act?, supra note 4, at 77.

^51. The debate took off with Frankfurt, Harry, Alternate Possibilities and Moral Responsibility, 66 J. Phil.829 (1969)Google Scholar. Since then, an extremely productive cottage industry has sprung up to debate these issues. For discussion and further references, see Fischer, John Martin, Recent Work on Moral Responsibility, 110 Ethics (1999), 93–139Google Scholar; Levy, Ken, Why It Is Sometimes Fair to Blame Agents for Unavoidable Actions and Omissions, 42 Am. Phil. Q. 93–104 (2005)Google Scholar; and the essays in Moral Responsibility and Alternative Possibilities (David Widerker & Michael McKenna eds., 2003).

^52. See Fischer, supra note 51, at 109–125 for a much more detailed discussion of the inner workings of Frankfurt-type examples. I note that the status of these examples remains hotly contested.

^53. Ken Levy, in personal discussion, has suggested that it may be possible to develop an account of control that does not presuppose the principle of alternate possibilities. If so, that would put the control requirement on a similar footing to the practical-agency condition.

^54. Corrado may adopt a view along these lines when he defends the necessity of “voluntary choice” rather than “control” as an interpretation of the act requirement. See Corrado, supra note 41.

^55. See, e.g., Breunig v. American Family Insurance Co., 45 Wis. 2d 536, 173 N.W. 2d 619 (Wis. 1970); and Restatement (Second) of Torts §283B and §895J (1979).