1. In some replies to critics since publication of the book, I complicate the story here. In particular, I add a further conjunct: the act type must implicate, by its nature, legally protected interests. I claim that if C implicates legally protected interests, then so does any act that is intentional under the description “an attempt to C,” hence the argument for the transfer principle goes through with this more complicated property. (This idea is important in my reply to Moore, below.) The new, more complex argument for the transfer principle has the virtue that it does not support the criminalization of other, watered-down versions of the completed crime that are not themselves attempts, such as those we find when a defendant who causes no harm had the mental states needed for the completed crime but lacked an intention that commits her to all of the completed crime's components. For detailed discussion, see Yaffe, Gideon, The Legal Importance of Trying: Reply to Enoch, Dahan-Katz and Berman, 6 Jerusalem Rev. Legal Stud. 1–28 (2013), doi:10.1093/jrls/jlt012Google Scholar; and Yaffe, Gideon, More Attempts: A Reply to Duff, Husak, Mele and Walen, 6 Crim. L. & Phil. 429–44 (2012)Google Scholar. But this difference in the argument that I now endorse in comparison to the one that appears in the book is not important to my reply to Bratman.

2. Bratman notes that in the book I explicitly mention only the right-to-left direction of this biconditional, but the argument I offer also requires the left-to-right direction, as he says.

3. I myself have doubts that the clauses directing attention to the circumstances “as [the defendant] believes them to be” are, as Alexander asserts, intended to be only inculpating and never exculpating. The results that follow under that reading are not just absurd to those who oppose negligence and strict liability; they are just absurd period. For instance: say that rape requires negligence with respect to nonconsent, as in many jurisdictions. V is consenting. D is consenting. D truly believes V is consenting, and vice versa. But a reasonable person in D's position would have believed that V was nonconsenting. D and V have consensual sex in the mutual true belief that it is consensual. If §5.01(1) is to be read as Alexander suggests, D has attempted rape. That just cannot be what the framers of the Model Penal Code had in mind. Rather, D's belief that V is consenting exculpates. Notice that the problem does not go away if the mens rea standard is recklessness. Say that it is and say that D both believes that V is consenting and at the same time recognizes a substantial and unjustifiable risk that she is not. Say, for instance, that he believes anything that he takes to be more than 95 percent likely, but it is unjustifiable to move ahead while aware of a 4.5 percent chance that V is nonconsenting. Still, the case is one of mutually consensual sex between partners each of whom truly (and in this case rationally) believes the other is consenting. How can that be attempted rape?

4. Of course, this might not be true, for the legislature may, in its wisdom, have criminalized the attempt without criminalizing completion. Say, for instance, that it is a crime to attempt to walk from the rim of the Grand Canyon to the river and back in one day. The fines for commission of this attempted crime are used to recoup the costs of rescue operations. But it might not be a crime to succeed in walking from the rim to the river and back in a single day. After all, those who manage it never need to be rescued, and so no money is spent on their rescue. An attempt of this kind is not criminalized under the transfer principle. It is not criminalized implicitly when we criminalize completion but instead through another route. I use the same example in Yaffe, Legal Importance, supra note 1.

5. By “the probability” of an event, Alexander means the conditional probability of the event, given all facts prior to its occurrence. If this set of facts includes the laws of nature, and if determinism is true, then it is true that “the probability” of the event is 1 or 0. But it is far from obvious that the laws of nature can be understood in this way. If, for instance, the laws of nature supervene on all that is true, past and future, in a possible world, then holding fixed the laws of nature means holding fixed the future, in which case it is tautological and unilluminating to note that “the probability” of every event is 1 or 0. And it is far from clear that determinism is true. And more importantly, what justifies the claim that this conditional probability is “the probability” of the event? What notion of probability is the relevant one depends a great deal on the purposes of inquiring about probability. Further, even if this is the right notion of probability, it does not follow that events of zero probability in this sense are not possible. The laws of nature are not necessary truths, nor are the initial conditions that under determinism ineluctably lead to the conditions prior to the event. There is too much literature to cite about these issues to list it all here, but for a start, see Lewis, David, Are We Free to Break the Laws?, 47 Theoria 112–121 (1981)Google Scholar.

6. A belief that p and an intention to p, in contrast to an intention that p, may have a different content, an observation that is consistent with what I say in the main text.

7. Here I adopt a convention also used in my book of placing square brackets around the content of the intention. As will become clear, it is best to construe what appears between the square brackets as the words in the language of the thought involved in the relevant mental state.

8. This is as good a place as any to respond to Moore's objection to my way of distinguishing between act, result, and circumstance elements of crimes. I say that which category an element falls into is determined not by the nature of the element but by facts about prosecutorial burdens. It is not as though, in nature, absence of consent, for instance, wears a little tag that says “circumstance” on it. Rather, it counts as a circumstance element because the prosecution bears no burden to show that the defendant caused it. Moore thinks that ultimately I classify the elements on the basis of an independent fact about them rather than on the basis of a fact about the way in which the legislature has imposed burdens on prosecutors. A circumstance is a circumstance, he suggests, for instance, because the guilty defendant did not cause it. But that is not right, for it is a circumstance still even if a particular guilty defendant did cause it. What makes it a circumstance has nothing to do with what did or did not cause it. What makes it a circumstance is what need or does not need to be shown about it for guilt. However, I would be happy to put my point as not about prosecutorial burdens but instead about the set of all conditions that are part of the crime. What determines whether an element is act, result, or circumstance, that is, is not something about it but instead something about the full set of elements of which it is one. Say E is one of the conditions involved in a crime. E is a circumstance iff the set of conditions do not also include E's being caused by the defendant. E is a result iff the conditions do include that (and do not require that D be the agent of E). And E is an act iff the set of conditions includes D causing E in the special way that is required for her to be the agent of it. Since the set of conditions that are included in the crime is equivalent to the set of burdens the prosecution bears, this formulation is equivalent to the one I offer in the book. But it may help to head off the mistake that Moore makes in objecting to my position.

9. Moore makes it sound as though he has this preference because he thinks normative explanations of a thing's nature are unilluminating. But that cannot be true in general. Some things have essential normative properties; others do not. And someone with Moore's realist commitments about the normative should not find that troubling.

10. The canonical statement and defense of the language of thought hypothesis is, of course, Jerry Fodor, The Language of Thought (1975). See also Jerry Fodor, LOT 2: The Language of Thought Revisited (2008).

11. See Yaffe, Gideon, Intending to Aid, Law & Phil. (2012), doi:10.1007/s10982-012-9168-8Google Scholar.

12. It is worth noting, as I think Moore recognizes but I mention because it is easy to miss, that my claim that there is no censure-luck—the claim that two people who do not differ except in respect of facts that are under the control of neither do not deserve different degrees of censure—does not commit me to the claim that what you cause makes no difference to what you deserve. The reason is that much that we cause is under our control in all the senses that matter to desert of censure.

13. I try to remedy this in two replies to commentaries written since the publication of the book. See Yaffe, Legal Importance, supra note 1; and Yaffe, More Attempts, supra note 1.