Free & Equal – Elise Sugarman’s “Supposed Corpses and Correspondence” with a précis from Gabriel Mendlow

25 Replies to “Free & Equal – Elise Sugarman’s “Supposed Corpses and Correspondence” with a précis from Gabriel Mendlow”

1. Hearty thanks to Gabe for his wonderful, probing précis, and to PEA Soup’s Kartik and Susanne for all the work they did to make this discussion possible. Thanks too in advance to anyone willing to wade into the dark world of supposed corpse cases with me!

   In my paper, I saw supposed corpse cases as presenting a correspondence puzzle—a puzzle about how a t1 mens rea can be related to a t2 actus reus so as to ground justifiable imposition of liability (by way of grounding culpability). In his précis, Gabe suggests I sidestep the correspondence puzzle rather than face up to it. Gabe suggests a bifurcated sidestepping strategy. According to the strategy, some of the troublesome cases can be dealt with by taking what I’ll call a mens rea approach: an analysis of what constitutes mens rea such that the mens rea requirement for murder is satisfied at t2; and some of the troublesome cases can be dealt with by taking what I’ll call an actus reus approach: an analysis of what constitutes actus reus such that the actus reus requirement for murder is satisfied at t1. The pay-off here is great. If the troublesome cases can be dealt with in one or the other of these two ways, maybe (although Gabe doesn’t outright say this) correspondence can remain what Causal Theory tells us it is.

   In advancing his bifurcated sidestep strategy, Gabe makes many excellent points. In this first post, I’ll try to say something high-level enough to assist in orienting additional discussion. I’ll make a preliminary point about the modesty of my aspirations and then briefly address each of Gabe’s two approaches in turn.

   To start, I want to draw out a point about sufficiency that connects up with Gabe’s last paragraph. In this paper, I make only a claim about sufficient conditions for justifiable imposition of liability for intentional murder: about how such liability can be justifiably grounded in a t1 intention to kill and a t2 killing. As I mention in the first substantive section of the paper, there are likely mental states that should count as mens rea for murder that aren’t intentions to kill. In a similar vein, if I’m in the mood to be conciliatory, I could add: there might be instances of conduct that should count as actus rei for murder that aren’t analyzable as killings, as that act description is ordinarily deployed (but see the discussion infra). My explanation for supposed corpse liability doesn’t rely on these possibilities. I claim: a jurisdiction J that only recognizes as mens rea intentions to kill and as actus reus killings (as that act description is ordinarily deployed) can nonetheless justifiably impose liability for intentional murder given the existence of a daisy chained correspondence relation. I do not deny that justified imposition of liability for intentional murder could be constructed out of other ingredients.

   But when I put things this way, the paper’s proposal might not sound particularly interesting. If we were advising jurisdiction J about how to design their liability-distributing system to account for supposed corpses, maybe, what we would advise the jurisdiction to adopt wouldn’t be a fancy new expansion of what constitutes satisfaction of their correspondence requirement: what we would advise the jurisdiction to adopt would be an appropriately capacious enough understanding of what constitutes the right *relata* for their correspondence requirement. If the categories of things that count as satisfying mens rea and actus reus requirements were appropriately capacious, the thinking might go, the puzzling aspect of supposed corpse cases would disappear. But should J take a supposed corpse defendant’s t2 mental state to be sufficient for mens rea? And should J take a supposed corpse defendant’s t1 conduct to count as an actus reus? This leads me to say a bit more about each of the two approaches in Gabe’s sidestep strategy: the mens rea approach and the actus reus approach.

   The mens rea approach: Gabe is happy to understand Williams as offering a kind of mens rea substitution principle: a principle stating that when the defendant’s t2 mental state is of a certain kind, it should suffice as a stand-in for an intention to kill. In particular: when a defendant’s t2 mental state is such that (1) the defendant believes that: their victim is dead because they killed them; and (2) the defendant shows signs of being animated by “the same malevolence that motivated their attack at t1,” then the defendant should be eligible to count as having satisfied mens rea at t2, despite not intending to kill at t2.

   There’s a preliminary worry here that the principle of legality would be violated if someone were convicted in a manner that accords with this substitution principle, *if* the law weren’t set up so as to include the principle as law. That is, they’d be convicted despite not violating the law. The principle-of-legality worry would be circumvented, though, if J adopted the mens rea substitution principle as law, thus expanding their mens rea requirement for murder so that it was appropriately capacious. But should J adopt such a principle to account for supposed corpse liability?

   I have doubts. For starters, the malevolence someone exhibits by disposing of a supposed corpse may often be less egregious than the malevolence exhibited when a defendant aims to kill; disposing of a corpse is just not as bad of a thing to want to get up to (I think). And it’s unclear why the belief condition should make a difference to the substitution Mendlow has in mind. Why would a defendant who hallucinates killing at t1 and then, with malevolence, buries the supposed corpse that featured in their hallucination be any more convictable than a defendant who does not so hallucinate and, with malevolence, buries a supposed corpse? (Based on Gabe’s application of his principle to inculpate RB3, it looks like a defendant need not have engaged in any t1 conduct toward the victim in order to count as satisfying the substitution conditions. But I’m not sure about this; his second condition as quoted in the paragraph above makes it sound as if there must have been an actual attack).

   An additional wrinkle frustrating application of the principle is that at least on one highly influential approach to fixing mens rea—the Model Penal Code’s—intentions can count as mens rea states regardless of the quality of motive undergirding them. Indeed, some heralded the MPC’s fixing of mens rea through descriptive rather than normatively thick judgments (of e.g. malevolence, or “wickedness”) as one of the major innovations of the Code. But if intentions to kill can be defensibly treated as mens rea regardless of their relationship to malevolence, it would be odd if malevolence were appealed to (at least, in an MPC-inspired system) to determine what an intention can be substituted in *for.* (The reaction here might be so much the worse for the “innovation” of the MPC; that’s a longer conversation!)

   Here’s a concrete example of the difficulties involved in application of the substitution principle for this reason. Say that at t1, a defendant intends to kill in defense of others; in so attempting to kill, they exhibit laudable motives. In virtue of acting on their intention, the defendant thinks they have succeeded in killing. At t2, full of hatred for their victim for what it is that the victim made them do, they unceremoniously bury the supposed corpse. I would think that though this defendant should be acquitted because they should be recognized as having a defense, the defendant should nonetheless come out as satisfying mens rea and correspondence. That is: the result in this supposed corpse case should parallel the result that would obtain if the defendant had simply succeeded at killing in defense of others at t1. But the defendant was not at all malevolent at t1; moreover, the defendant was differently malevolent at t1 and t2. The substitution principle doesn’t look to get us to the right result.

   Now, that result need not be reachable by the mens rea approach. Perhaps the actus reus approach would be more apt for dealing with that case. I turn to that now.

   The actus reus approach: Gabe thinks setting the actus reus at t1 will enable us to account for the intuition that supposed corpse defendants like Reformed Burier are liable for intentional murder even though (and even with the assistance of Gabe’s substitution principle) they do not have mens rea at t2. On the actus reus approach, proximate causation is recruited as the real workhorse for explaining supposed corpse liability.

   Gabe may well be right to point out that my unpacking of reasonable foreseeability as a test for proximate causation in my paper is too persnickety for my own good. As I admitted in the paper, my attack on causal analysis is vulnerable to the introduction of a more mature theory of proximate causation. But to lay my cards on the table: I find compelling the idea that proximate causal claims should track ordinary causal claims. In particular: I find it compelling to think that the ordinary act description of “killing” should share a close affinity with the more technical act description “act proximately resulting in death.” And I find it compelling that Reformed Burier performed an action describable as a killing at t2 but not at t1. (Is it the family members or the terrorist that kill the plane passenger? I’m stressed about this, but I want to say that it’s the terrorist.)

   That’s hardly an argument. But here’re two beginnings of supporting arguments against the adequacy of the actus reus approach. One or the other might provide fodder for further discussion.

   First: supposed-corpse-like cases could come about for so-called conduct crimes: crimes requiring just that conduct fit a particular description (rather than proximately cause a particular result). An example from an F&E anonymous reviewer goes as follows: Sloppy Stealer and a friend of his split a storage unit where they keep property of theirs that is difficult to tell apart (never mind the wisdom of this arrangement). At t1, intending to take his friend’s property, Sloppy Stealer takes what he falsely believes is his friend’s property out of the storage unit. At t2, Sloppy Stealer infers on the basis of his belief in what he has done at t1 that the property remaining in the storage unit is his. He then intentionally removes the remaining property from the storage unit, believing it is his own. In actuality, the property he moves at t2 is his friend’s. I think Sloppy Stealer is justifiably held liable for larceny. But it seems yet more natural in such a case to insist on a t2 actus reus. While Sloppy Stealer merely causes the taking of another’s property by acting at t1, he does the thing that counts as taking another’s property at t2.

   Second (and this is something Gideon Yaffe has prompted me to think about): turning back to genuine supposed corpse cases: if Reformed Burier’s t1 conduct serves as the actus reus for both an attempted murder and a murder, then there might be double-jeopardy issues implicated in convicting Reformed Burier of murder if he is first acquitted of attempted murder. In contrast, if the actus reus for attempted murder and the actus reus for murder are constituted by different instances of conduct, there won’t be such an issue. That might accord better with intuitions about the circumstances under which Reformed Burier could be held liable for murder. I was worried about pursuing this line of argument in my paper, because I was worried about wading into a debate about how best to understand the scope of the protection against double jeopardy. But it is food for additional thought.

   To wrap up this post, which is already way too long, I’ll emphasize that Gabe’s bifurcated sidestep isn’t just a suggestion about a different route to explaining why X collection of supposed corpse defendants can justifiably be held liable for murder: it’s a suggestion that changes the members of X. That is, Gabe’s strategy results in different convictions being green-lit as justifiable. RB2 (who causes a condition of apparent death in his victim without later realizing it) ends up justifiably convictable on an actus reus approach even though he flunks my daisy chain conditions. And Gabe’s Dana ends up justifiably convictable on a mens rea approach, even though she, too, would flunk my daisy chain conditions. (Gabe also says that like Dana, RB3, who mistakenly thinks he has caused a condition of apparent death, ends up convictable on a mens rea approach. That depends on whether RB3 was, like Reformed Burier, “reformed.” While I meant for all my “RBs” to be reformed at t2 (see p. 61), I appreciate that could have been made more clear). Given weaknesses in intuitions regarding these highly rarified cases, I am not sure the extensional differences are particularly helpful in settling the score, but they are certainly worth noticing, and I’d be happy to receive reports from others regarding which way intuitions cut.

   I’m really looking forward to the discussion! Thanks again for being here.

2. Thanks, Elise, for your probing and carefully argued article, which was fun to read, and thanks to Gabe for a great precis. Like Gabriel, I find myself pulled strongly by the thought that properly spelled out, the act at time t2 in corpse cases is the place where we can find sufficient mens rea for murder. I started with this suspicion after reading your abstract – which takes for granted in a way that I found suspicious the framing that corpse cases put pressure specifically on the correspondence requirement – a framing that makes sense if you think of these cases as needing to extend the mens rea from t1 through to the subsequent act at t2 but not as a plausible a priori constraint if we are just looking for why these cases count as murder. I retained this suspicion as you went carefully and systematically through the various possibilities of extending the contemporaneous account either by getting the t1 act to count or by “stretching” the actus reus to include both t1 and t2, but omitted to consider how to get contemporaneity at t2 alone. And my suspicion was reinforced when you tried to subsume your ultimate answer to a broader conception of “tracing”, since I have a prior suspicion of tracing, which I think is often motivated by trying to collapse genuine culpability for a prior event into supposed culpability for a later event, and the cases motivating which are often better treated by taking seriously the degree of culpability for the prior event. So I found myself very sympathetic to something like Gabe’s proposal – perhaps with a little more skepticism about whether “remorse” cases require separate treatment, and a little more sympathy with the diagnosis that they really don’t belong with the others. So I think that’s where my initial sympathies lie, and roughly why.

   But I actually want to ask you what you think about some different possibilities – each of which retreats a little bit from the snugness of the idea that criminal liability needs to track culpability. Of course criminal liability ought to have something important to do with culpability. But each of these proposals relaxes the snugness of the fit just a little bit.

   The first proposal that I want to consider starts with the observation that corpse cases are all cases of unsuccessful attempts. We don’t, of course, punish attempted murders to the same degree as we punish successful attempts. But maybe looking more systematically at what makes unsuccessful attempts less serious crimes than successful ones will help us to see why that doesn’t apply in corpse cases. This, of course, is Gideon Yaffe’s project in his book on criminal attempts. If I remember correctly, Yaffe says that culpability depends only on the frame of mind of the actor and hence there is no actual difference in culpability based on whether an attempt succeeds, but because different amounts of commitment to success result in different probabilities of success, the fact that an attempt failed gives us evidence that they may not have been as committed to the crime. On this view, corpse cases could simply be cases in which this evidence is defeated – someone who disposes of what they believe to be a corpse provides us with evidence that they would have taken the necessary steps, and hence with the same level of evidence about their commitment and therefore culpability as a successful attempt. So Yaffe’s view seems to me to sustain the initial promise of looking to the difference between successes and attempts in order to cover corpse cases. On this view these cases don’t tell us anything about correspondence at all, but are merely expectable exceptions to the general principle that attempts are less serious than commissions.

   You can think of Yaffe’s view as a kind of relaxation on the idea that criminal liability must fit closely with culpability. On his view, the point of criminal liability is to track culpability, but we have to do this through our evidence, and sometimes things that don’t themselves constitute differences in liability can nevertheless constitute evidence of such a difference. The next proposal that I want to consider relaxes the close fit between liability and culpability further. It starts with the thought that, contra Yaffe, attempts that are not successful genuinely involve less culpability than successful crimes – simply by virtue of failing to be successful. This assumption is a better home for the problem of corpse cases, because it sharpens the problem of getting liability for murder out of the events that come next, and not just liability for attempted murder. Earlier I suggested that I’m sympathetic to the view that the agent may incur sufficient liability for murder just in virtue of what they know and do at time t2. But if that doesn’t work, then maybe the sum of their culpability for attempted murder at time t1 and for negligent homicide at time t2 is close enough to the degree of culpability for murder as makes no difference. So you can think of this as another proposal – not to get culpability for murder directly at all, but just to find enough total culpability by casting around at all of the culpable behavior of the agent to justify our intuitive reaction that failing to convict for murder constitutes a grave omission. Again, this requires relaxing the snugness of the fit between criminal liability and what the agent is culpable for, but not in an egregious way – agents should still be held liable only to degrees to which they are culpable – this view just allows that the exact bookkeeping over what they may be liable for and culpable for is not always essential.

   I want to close with a final, further-out, suggestion. Suppose that, unconvinced by your tracing explanation of why it is natural and appropriate for corpse cases to count as well as by your takedown of alternative accounts of why they satisfy correspondence, we decide that corpse cases do not, after all, satisfy correspondence. This, I think, would be bad law quite apart from whether it better tracks culpability. It would create a moral hazard for every prospective murderer. By breaking their murder into two steps, one of which they could claim to have believed to have been successful even though it was not, they can impose an evidential burden on the prosecution to prove beyond a reasonable doubt that they knew at time t2 that their victim was still alive. In their defense, they would only need to prove that the victim was in fact still alive at that time and create sufficient reasonable doubt that they knew it. This, we might think, is bad law even if it better tracks true culpability, precisely because of this moral hazard that it creates. Again, endorsing this explanation as a justification of treating corpse cases in this way requires relaxing the snugness of the fit between proper liability and culpability, but arguably not so much that it is not consistent with appropriate principles about legal design that can be justified on more general grounds.

   Each of these alternative proposals takes us in a different direction than you ultimately pursue, Elise. None of them are going to get us started on the imaginative and ambitious project of reimagining what tracing involves in the theory of responsibility, so they are all in a way less exciting than the direction that you propose. But for that reason in particular it seems to me important to know why we can’t make sense of your puzzle in more conservative terms – terms that place the pressure, instead of on correspondence, on mens rea, attempt, total culpability, or on moral hazard. Thanks again for the opportunity to think about these things!

3. I do not feel the force of supposed corpse cases. Yet I think Sugarman’s Daisy Chain account is the best explanation of the correspondence relationship on offer. The Daisy Chain account is so good that it is wasted on the corpses.

   A. Supposed corpse cases leave me cold
   Sugarman (after Glanville Williams) frames the main argument in terms of criminal liability: “intuitively, supposed corpse defendants are criminally liable for murder.” I share this intuition. However, Sugarman clarifies the intuition further: the supposed corpse defendants are liable for what (in most U.S. jurisdictions) is called intentional or premeditated murder. I don’t share this intuition, perhaps because I am jaded.

   It’s worth clarifying a little about how murder liability works in U.S. jurisdictions. Murder is an offense, and there different theories that may be used to prove that offense. The paradigmatic theory of murder requires proof of the defendant’s premeditation or, equivalently, intention. This is the theory of murder that Sugarman aims to vindicate. In many jurisdictions, murder can also be proven on a felony murder theory: roughly, a defendant is liable for murder if they commit a type or token felony (the “predicate” felony) and the victim dies. In many (but fewer) U.S. jurisdictions, a defendant can be liable for murder on what is called a “natural and probable consequences” theory: the defendant is liable for murder if they intentionally act and the victim dies as a natural and probable consequence of the defendant’s intentional action. (There are a variety of other possible theories of murder. Some are specified in statutes, others are implicit or established via caselaw.) The tricky thing about learning criminal law in the U.S. is understanding that prosecutors can choose which theory of murder to charge in a particular case and can argue for multiple theories of murder in the same case. To convict the defendant of murder, the jury must unanimously agree that all the elements of a theory of murder have been proven, but they do not need to unanimously agree on which theory has been proven in any particular case.

   All this is to say that familiarity with how the criminal legal system in the United States works might undermine the intuition that supposed corpse cases involve intentional murder, or at least blunt the urgency of this question. Even if these cases don’t intuitively implicate intentional murder, they each implicate murder on other theories.

   For example, consider the Reformed Burier case. In a jurisdiction that has a felony murder theory of murder liability and allows attempted murder to count as a predicate felony, a replacement-level prosecutor could easily charge Reformed Burier with murder on a felony murder theory: victim died as a result of Reformed Burier’s attempted killing. A prosecutor of Mendlowian ingenuity could probably charge Reformed Burier with murder as a natural and probable consequence of an offense such as unlawful burial. So, practically speaking, little hangs on whether Reformed Burier is liable for intentional murder or some other theory of murder.

   These legal doctrines related to murder are, of course, hacks. They apply mostly (or, in the case of felony murder, exclusively) in the context of homicide and work to alleviate difficulties in exactly the kinds of cases that motivate Sugarman’s analysis. These hacks do not apply in all legal systems (and, I’d guess, none of the ones involving discussion of the Thabo Meli case). But to the extent they apply within the US legal system, they render supposed corpse cases unscary.

   B. The Daisy Chain account is nonetheless vital

   Sugarman uses the example of murder to motivate the Daisy Chain theory, but murder is the exception. Hacks like felony murder and “natural and probable consequences” allow supposed corpse cases to be charged as murder, but the availability of these hacks should not obscure the analytic deficiencies in the Contemporaneity and Causal theories of the correspondence relationship that Sugarman rightly notes.

   Regardless of whether the Daisy Chain theory is necessary to explain how murder liability might arise in single-person supposed corpse cases like Reformed Burial, there are a variety of criminal law contexts in which the Daisy Chain theory can help explain anomalies arising out of exactly the defects of the Contemporaneity and Causal theories that Sugaraman notes.

   For example, consider a criminal case that is used to illustrate correspondence in many US law schools (including mine). In State v. Hopple, 357 P.2d 656 (Id. 1960), the defendant locked away 15 yead of sheep belonging to Williams and was convicted of grand larceny (which, at the time, required proof that the defendant either took or deprived another of property with intent to deprive the other of the property permanently). The prosecution introduced evidence that the defendant thought he had a right to the sheep, which were tromping on his cattle field, but the trial court excluded evidence (over defendant’s objection) related to the timing of his intention to deprive Williams of the sheep permanently. On appeal, the Court overturned the conviction, finding that the evidence was relevant because if defendant had no felonious intent to steal the sheep at the time they came into his possession, he could not be guilty of grand larceny even if he “subsequently conceived the intent to appropriate them.”

   This case is consistent with what Sugarman calls the Causal View of correspondence: Hopple is not liable for larceny because his intention to deprive Williams of the sheep did not non-deviantly cause him to lock away the sheep. But what if we revise the Hopple case along the lines of Sugarman’s Reformed Burier case? Say that, at t1, Hopple was so blinded by his desire to see Williams impoverished him that he mistook the sheep being locked up for his own. Suppose also that, at t2, Hopple came to recognize that the sheep were actually Williams’s, but then he resolved not to return the sheep to Williams.

   In this version of the Hopple case, one might have the intuition that Hopple is criminally liable for theft. Yet the Causal View cannot explain why, since Hopple’s intention to deprive Williams at t1 was (arguably) only a deviant cause of his depriving Williams of the sheep. Yet something like the Daisy Chain view can explain how liability might arise in this alternative version of Hopple: Hopple’s intention to deprive Williams at t1 caused his ignorance regarding the provenance of the sheep, and so is appropriately connected to Hopple’s culpability for the theft.

   This alternative version of the Hopple case is not a direct extension of the Daisy Chain view. However, the notion that a mens rea intention at t1 might ground liability even if it is a deviant cause of behavior at t2 seems much more plausible as an explanation for how liability might work in larceny cases than a straightforward application of the Causal View.

   More broadly, a significant aspect of Sugarman’s contribution is to explain how one might preserve something like the Causal View in a world of deviant causation. Even if the Daisy Chain view isn’t for the corpses, it works well on the sheep.

4. Thank you so much for your post, Mark! Really appreciate your taking the time to engage with these ideas. There’s so much rich stuff in your post that I’m excited to start to unpack.

   On mens rea: I appreciate your noting that while my paper spends a fair amount of time on the strategy of placing what we treat as the actus reus back in time, I don’t spend time facing up to the analogous strategy of placing what we treat as the mens rea forward in time. I should have. My point in response to Gabe in the first post here (let me know what you think!) is that at least in a criminal legal system shaped by the MPC’s conception of what constitutes a mens rea state, there isn’t going to be a clean way to say of the mental state at t2 that it is the kind of thing we ought to substitute in. If we try to evaluate the wickedness of motive at t2, we shift into a mode of determining culpability that is in tension with the motive-agnostic mode of culpability determinations advanced by the MPC. While I am trying to advance a normatively defensible notion of liability, I’m trying to do so holding fixed some contingencies, so that what I’m advancing is familiar to actual Anglo-American systems. So I’m trying to see how correspondence might be understood in the context of a scheme, like the MPC’s, according to which mens rea states are descriptively characterizable rather than dependent on an evaluation of motive. I recognize that leaves open a deeper critique: that at bottom we really should be carrying about qualities of motive, and that direct appeal to qualities of motive might help us fix mens rea at t2. Even then, though, I think a problem remains: what a supposed corpse defendant takes themselves to be up to at t2 is just a different sort of thing as what the supposed corpse defendant takes themselves to be up to at t1—so much so that I have trouble appreciating the defendant could be similarly “malevolent” at t2.

   On especially serious attempts: I really like your proposal having to do with especially serious attempts. It’s something I’m going to have to think much harder about. I’ve been concerned with trying to explain why supposed corpse defendants are liable for killings, but your point is that perhaps they could just be liable for especially serious attempts that are meaningfully similar to killings, such that they’d be punished just as harshly. One potential sticky thing here has to do with supposed corpse defendants who have changes of heart. You write that “someone who disposes of what they believe to be a corpse provides us with evidence that they would have taken the necessary steps” to go about completing their attempt. But what if we have evidence that Reformed Burier is genuinely reformed after t1, and that he would have driven his victim to the hospital immediately, had he realized that his victim were still alive? Given the possibility of genuine changes of heart, what a defendant does at t2 might not be particularly good evidence of what his commitment was like at t1. Now, it sounds like you might be happy exculpating Reformed Burier. But my more general point is that evidence at t2 of what the defendant did while thinking their victim dead might not be particularly good evidence of commitment at t1, while thinking their victim wasn’t.

   Another thought about the proposal–say that at t2, the supposed corpse defendant thinks they dispose of the supposed corpse, but they don’t end up succeeding at doing that either, such that the victim remains alive. E.g.: once the defendant leaves (at t3), the victim crawls his way out the burial site, Kill-Bill style. On the especially serious attempts view you advance, I take it that this defendant would be just as culpable at t1 as a regular supposed corpse defendant, because at t2 they have provided us with just as much evidence of commitment to killing at t1 by acting at t2. But they wouldn’t be convictable for killing, because they never killed. Maybe one good way to test your proposal against mine would be to consider how intuitions about such a defendant compare against intuitions having to do with a regular supposed corpse defendant.

   On summing up: Your idea that culpability for an attempted murder at t1 and culpability for negligent murder at t2 might be summed up so as to get, roughly, an amount of culpability equivalent to the culpability for murder is exactly how Paul Robinson proposed we deal with supposed corpse cases in his seminal YLJ article, “Imputed Criminal Liability,” so you are in great company there. I have two qualms with the aggregative approach. The first is that to check and see that these two convictions would satisfy the itch to convict a supposed corpse defendant of murder, you need the help of an “aggregative” principle: a kind of principle that tells us that it would be a’okay to sum up the culpability for the t1 conduct with the culpability for the t2 conduct. I think such an aggregative principle is just as hard to come up with as a transaction principle—a principle that would tell us when you could block instances of conduct into one stretched actus reus. The second worry just takes us right back to your proposal about attempts. I don’t think supposed corpse defendants necessarily *are* liable for negligent homicide at t2. Negligence, at least on the MPC (which, again, is the model I’ve been working with) requires the failure to live up to an epistemic standard by failing to come to have a belief as to risk a reasonable person would have had. But supposed corpse defendants might be experts at belief management, such that they are never negligent. This means that in order for an “aggregative” approach to work quite generally, we’ll be back to having to say that the kind of attempted murder at t1 is sufficiently similar in culpability to completed killings to explain the relevant intuition.

   On moral hazard: Your last suggestion about moral hazard raises really interesting questions about just which considerations should shape how it is that criminal legal doctrine ought to be designed. Your point is that to avoid moral hazard, maybe we could recommend that a system should convict supposed corpse defendants, culpability notwithstanding.

   I’m totally on board with the fact that criminal liability need not, and should not, track culpability with any kind of ruthless precision. There might be all kinds of what Alex Sarch has called “institutional design considerations” that recommend we shape doctrine away from doctrines that would perfectly track to culpability. Values like sensitivity to perverse incentives, predictability in generating outcomes, useability of doctrine to factfinders, etc do and should shape criminal legal systems. To provide a concrete example: In most states, doctrine maintains that defendants cannot appeal to duress as a defense to murder. Is that because duress is somehow never exculpating in the context of murder? I don’t think so. Defendants may be under serious and genuine duress when they commit murder and yet not receive the benefit of a duress *defense*–which is to say, their duress might not be recognized by law. Nonetheless, it may well be good for jurisdictions to deny duress defenses for murder, precisely because of the pragmatic considerations involved. In a California Supreme Court case, People v. Anderson, the Court justified their carve-out rule by explicitly appealing to pragmatic concerns about deterring gang violence in the state. The Court might well have been on to something there.

   With that said, every time we design doctrines that impose liability despite no culpability (as in the case of a denial of a duress defense for murder), I take it we are incurring a kind of cost. Departures from culpability-tracking scheme need to be justified by appeal to some one or other institutional-design-type value that overrides the cost. To put it another way, that a defendant is not culpable is at least a *pro tanto* reason not to hold that defendant liable. But I think that by convicting supposed corpse defendants, we don’t even incur this cost, because we aren’t convicting defendants that a perfectly culpability-tracking system would acquit.

5. Thanks so much for your thoughtful post, Stephen. There’s a negative point and a positive point in your post, and I’ll take up each in turn.

   Your negative point is that I don’t need a daisy chain to support supposed corpse liability for murder because murder has a special place in the law. Given the special place murder occupies, there are all sorts of specific doctrinal mechanisms that one could appeal to (within at least some American jurisdictions) to explain supposed corpse liability for murder. Because of these mechanisms, it feels like overkill (pardon the pun) to go about positing a daisy chained correspondence relation.

   The worry with appeal to one or the other actual murder-specific doctrinal mechanism to explain supposed corpse liability is that appeal to such a mechanism may not ground the kind of normative claim I’m trying to make—that it would be justifiable to impose liability for intentional murder on a supposed corpse defendant. The mechanism would itself be normatively defensible for that to be the case. And some such mechanisms—just, for example, the doctrine of felony murder—might not be. They might sweep too much conduct in.

   A further point here is that, as you note, the doctrinal mechanisms specific to murder are wildly contingent. (In a really great piece in his Doing and Deserving, Joel Feinberg makes a similar point, denying Anscombe’s suggestion that there’s such a thing as being “morally speaking” a murderer). But one nice feature of my daisy chain is that it is the sort of thing that either exists or doesn’t, quite apart from which murder-specific doctrines are recognized by a given jurisdiction. My discussion abstracts away from the particulars of any one jurisdiction’s mechanisms for dealing with murder and examines actual features in the world—actual amalgams of defendants’ thought and conduct—that could ground culpability and so, make it such that an imposition of criminal liability could be justifiable.

   That’s not to deny that murder is, *in some ways,* special. I think there’s a reason supposed corpse cases have gripped criminal legal scholars’ imagination. While philosophers like all sorts of odd fact patterns, an interesting feature about this particular odd fact pattern is that the stakes can’t have been higher for real people convicted by real courts. But I’d be surprised if culpability for killings was really all that special, such that it played by special rules. I’m attracted to analyses that would enable us to demystify just when liability for murder is justified by appealing to general features grounding culpability that could exist in other contexts too, such as, just for example, that attempts non-deviantly cause conditions of apparent completion that then go on to explain mistaken performance.

   This segues into your positive point: that the daisy chain might be useful in other contexts, to explain liability in other kinds of cases of deviant causation. I am very happy with this diagnosis, as I hope my appeal to the Sloppy Stealer in response to Gabe’s precis makes clear. I want it to turn out to be the case that supposed corpse liability is but a species of a more general phenomenon.

   Even if (stick with me here) the daisy chain conditions I presented in my paper do explain supposed corpse liability for intentional murder, ultimately, an explanation of supposed corpse liability is an explanation that’s only so useful. The conditions I’ve posited as remain dissatisfyingly narrow along (at least?) three dimensions. First: they hard-wire in a particular mens rea state at t1: an intention to kill. But you can imagine—and there are real cases like this—a defendant who, e.g., acts recklessly so as to cause apparent death at t1 and then disposes of their victim’s supposed corpse at t2. Second: they hard-wire in a particular actus reus: of killing. But you could imagine supposed-corpse-like cases involving crimes that have different actus rei. See: the Sloppy Stealer Case, in response to Gabe’s precis. Third: they hard-wire in that the defendant’s t1 attempt is such that it is eventually supposed complete. But there will be some cases of justifiable imposition of liability that aren’t like this. One of my anonymous reviewers gave me a great example (see fn. 68): A defendant supposes they have poisoned, but not yet killed, their victim by acting at t1. At t2, they administer a substance the defendant thought would serve as an antidote, but because of the way the alleged antidote interacts with the not-poison, they kill their victim. My daisy chain conditions cannot capture this case, because there’s never a condition of apparent death that materializes. I am hoping to eventually get at daisy chain conditions explanatory of the genus rather than just the species of supposed corpse liability.

   With all that said, I’m not sure the most plausible generalization of the daisy chain conditions would support liability in Hopple. The issue with application to Hopple–at least, as you’ve recounted it–is that even if the defendant harbored an intention to take his neighbor’s sheep at t1 that caused him to do a poor job identifying his sheep shortly afterwards, the defendant did not act on his t1 intention to bring anything about non-deviantly. And I think that any plausible generalization of my daisy chain conditions, the daisy chain is a particular deviant causal chain that has some non-deviance built into it, as between a supposed corpse defendant’s intention to kill and a condition of apparent death. What do you think? Is this already too restrictive a notion of what features might ground a daisy chain?

   Thanks again for your engagement!

6. Thank you Elise Sugarman for this highly enjoyable and thought-provoking article, as well as Gabriel Mendlow for the précis, and everyone’s comments I have read through here. I have to confess I come from a different position, being most familiar with UK law and knowing very little about US federal or state laws, MPC etc – and being in some ways a ‘critical’ legal scholar.

   Stephen Galoob’s comment perhaps resonates with me the most, where it contrasts (part of) this analysis with courtroom practice. You write, Elise: ‘I’m attracted to analyses that would enable us to demystify just when liability for murder is justified’. Although that could be appealing, I would worry that there just is no consensus. As you probably know, the UK no longer has felony murder and we arguably work in law with Cooke’s mens rea, ‘malice aforethought’, but this not requiring either malice or premeditation, only intention. This is the intention to kill or to do grievous bodily harm – but then there is debate whether the latter should really come within murder. Outside of the law, ‘murder’ is arguably just the killing that we think of as most morally wrong.

   I found some examples in the article do not pump my intuition the defendant is guilty of murder as much as the article suggests it should (eg Patient Poisoner, possibly). There seems, too, to be some disagreement on the offence categorisations of the different examples shown within the comments on this page. Some some think are murder, others think are not. I think it is quite possible our law is, or remains, ‘casuistic’ and the outcomes are at best(?) the moral view of the judge or jury (better by contrast to them being entirely ad hoc and without foundation). This also applies in appeals, particularly in common law (non-codified) offences, with the elements of offences being reshaped for the outcome appeal judges consider just in the instant case (or to best shape the future law etc). An example of this is Fagan v Met Police Commissioner where the actus reus of assault is transformed to be a continuing act, whereby Fagan who has driven onto it accidentally, can be convicted for refusing to move his vehicle of an officers foot.

   Theft, which Gabriel opens with (also relating to Hoppel), quite well demonstrates these issues in English and Welsh law. AR and MR analyses do not well apply, the law has morphed to get convictions in wrongly charged cases (which did not fit within the application of the statute at the time charged) and the courts recently changed the meaning of dishonesty because they did not like the outcomes the leading case was producing (Ivy) etc.

   Theft, of course, also requires the ‘intention to permanently deprive’ the other of the property – this is evidentially problematic but we have additional rules where it can be found or inferred. We could, with a bit of work, adopt this in the definition of murder e.g. something along the lines of ‘intention to permanently deprive the other of life’, evidenced by acts consonant with that will. Properly formulated, this could capture many ‘supposed corpse’ cases, and loose others, but I think it would square with most of my own intuitions. I am not sure it would do so well with the intuitions of others, however, and perhaps this is a broader problem for analyses.

   I would be pleased if you would like to respond to any of these remarks and perhaps, in any case, I will be lured away from my skepticism by following comments and discussions!

7. Congratulations Elise for publishing a very fine example of rigorous and careful scholarship, and thanks to the PEA Soup editors for inviting me to remark on it.

   I want to stand up for the Single Transaction Analysis, since I think it’s dismissed too quickly. It also happens to find favour in Canada. In a case called Cooper, the Supreme Court of Canada decided that for there to be “concurrency” between actus reus and mens rea sufficient for a murder conviction, an intent to kill need only coincide with one wrongful act in a continuous series of such acts tied together as a single episode or transaction even if it does not coincide with every act in the series. Although Cooper was not a supposed corpse case, the Court pointed to Meli as an example of a single transaction where intent was present for only a part of the whole, amounting to sufficient concurrency.

   I’m not persuaded that Reformed Burier must be read as a case involving a broken transaction severing mens rea from actus reus. Elise says that, on a “plan-based transaction principle,” the Reformed Burier’s subjective change of plan brings an end to an initial episode of intentional killing via strike to the head and commences a new episode of respectful burial plus killing, such that the initial intent to kill cannot overlap with any part of a transaction that includes an act of killing. But it’s not clear to me why we should accept a plan-based transaction principle that yields this result. That’s arguably not what the Supreme Court of Canada endorsed in Cooper. In that case, the defendant began to strangle the victim with an intent to kill and blacked out from intoxication before the victim died of strangulation, at which point, being blacked out, the defendant could have no intent to kill. But regardless of what was going on in (or not going on in) the defendant’s mind, the Court decided that there was a continuous transaction encompassing the initial intent to kill via strangulation and the later death via strangulation. Concurrency obtained. The point is that in both Cooper and Reformed Burier, after intent was formed but before a killing occurred, intent disappeared due to some change in mental state, either a loss of lucidity or a change of plans.

   Let’s call the alternative to the plan-based view sketched above a “fact-based transaction principle,” to distinguish it as objective rather than subjective. On a fact-based transaction principle, just as in Cooper, the facts could objectively establish that the Reformed Burier’s initial strike and subsequent burial formed a single transaction regardless of what was going on in the agent’s head, i.e., the change of plans. If so, the initial intent to kill coincided with at least a part of that transaction, albeit not all of it, to a degree sufficient to ground liability, in accordance with intuition. If you’re not sure that Cooper helps generate a “salient similarity” between Meli and Reformed Burier, consider a variant on the facts of a British Columbia case called Bottineau that might seem closer to Reformed Burier. Defendants fail to provide the necessaries of life to a child, over a period of time coinciding with an intent to kill. The child becomes severely malnourished, and in horror at what they have done, defendants have a change of heart and try to rescue the child from starvation by offering food. But they fail since the malnourishment is too far gone, and the child dies. On a fact-based transaction principle, despite the change of heart, the initial intentional starvation and subsequent rescue attempt are part of the same transaction.

   Now, I surely haven’t said enough to flesh out this fact-based transaction principle, much less show its superiority over a plan-based one (other than enlisting the authority of the Supreme Court of Canada). But it seems to me that although the issue of whether there is a continuous transaction is a matter of fact, not a matter of the defendant’s plans, it is not a metaphysical or ontological matter of being able to individuate episodes or sets of acts into a series that can be said to exist in some robust realist sense out there in the world. Rather, whether two acts count as part of one transaction would seem to depend on what the British Columbia Court of Appeal calls “policy considerations” or what Glanville Williams calls “ordinary ideas of justice and common sense.” Whether there is a single transaction depends on whether we think it’s morally fair to ascribe the entirety of a series of acts to a defendant such that mens rea overlaps with a part of one transaction. The conceptual conclusion on concurrency is driven by a normative one in much the same way that conclusions about proximate causation concern fair ascription.

   On this approach, I suggest that the transaction in Reformed Burier counts as continuous despite the change in plans. For, having formed a culpable initial intent to kill and having struck the victim, morally speaking, it’s now just too late for him to abandon his plan and distance himself from his prior culpability by respectfully burying the victim; that won’t ever erase what he’s done. Regardless of his newfound laudable attitude, it’s fair to treat the burial (and killing) as part of the same transaction as the initial strike, made with intent to kill, and satisfying concurrency. Mutatis mutandis for the variant on Bottineau, the starvation case. I know, Elise, that you say that you haven’t given a knock-down argument against the Single Transaction Analysis. But I’m not yet sure that what you do say uncovers difficulties that make it high time to search for a different approach. Perhaps there’s more to say.

8. I thought this was a great paper, Elise! Have really enjoyed reading all this terrific discussion as well. I wanted to raise some methodological questions or worries about some of the strategies for dodging the problem Elise tackles head on with her daisy chain account — and then ask if maybe a deflationary (or biting the bullet response) in the end might be a good enough way to proceed. This follows on from the other interventions thus far, since all the comments by Gabe, Mark, Stephen and James seem in different ways to touch on strategies for dodging or dissolving the difficulties posed by supposed corpse cases. One benefit of this, I take it, is that it would seem to let us stick with a simpler picture of concurrence just as something like actuation of the lethal conduct by the mens rea (where concurrence just is a necessary condition for full-fledged culpability, not sufficient for it — not least since wrongful conduct in which concurrence IS present might still be excused).

   So, I think there are two groups of Dodging Strategies on the table (correct me if I’m wrong — just trying to map the landscape a little bit since there’s so much going on). Consider each in turn.

   I. Alternate Routes to the Desired Conclusion

   One type of Dodging Strategy is to explore alternative routes to justifying liability for murder in a supposed corpse case besides the daisy chain account. These alternatives include (I think):

   a. Constructive Liability: D is guilty of murder because D committed a wrongful act – either at t1 or t2 (or both) – and D’s liability for murder can be constructed out of this culpable wrong (perhaps because of something like a “thin ice” principle). I think this was suggested in Gabe’s discussion of the death-by-medical-negligence cases. (“However improbable, death-by-medical-malpractice is a thing that might happen if you attack someone with the intent to kill them. If it does happen, ‘it’s on you.’”) Something similar seems to be suggested by his discussion of the terrorist examples.

   b. Tweaking the Mens Rea for Murder: Suppose you think we should lower the mens rea for murder so it doesn’t require intent to kill, but rather something less — like the (controversial) UK rule that an intent merely to cause grievous bodily harm is enough for murder where death results. (I think Elise rules this out by stipulating we’re in a jurisdiction that requires an intent to kill for murder, though someone who likes approach b. might still say that the jurisdiction we’re concerned with has a law that’s normatively sub-optimal – and that is the source of the difficulties here, they might say.)

   c. Aggregation of Culpability: A third alternative is to secure a conviction for murder by aggregating the culpability displayed by D’s attempted murder at t1 and D’s mistaken killing at t2 (which D believes to be only something like a a corpse desecration). Perhaps these bits of culpability can be added up and thereby amount to the culpability on display in a full-fledged murder. (I think Elise raises good worries about whether these two acts are of a sufficiently similar type to genuinely manifest the same kind of insufficient regard toward the same interests or values, such that they could be legitimately added up. Nonetheless, maybe they can if they are expressions of the same underlying malevolence in the way Gabe suggests at the end of his precis?)

   These alternative explanations can be debated further, but one overarching question for me, Elise, comes from the fact that you – via your daisy chain account – are only trying to offer a sufficient condition for murder liability. (You say in the reply to Gabe: “In this paper, I make only a claim about sufficient conditions for justifiable imposition of liability for intentional murder: about how such liability can be justifiably grounded in a t1 intention to kill and a t2 killing.”)

   So does this leave open the possibility that all of the above strategies (a.-c.) could also, at least sometimes, provide a viable route to also supporting liability for murder? Could you even accept that in one and the same case, there are multiple routes to (grounds of) liability for murder for a supposed corpse defendant? If so, is the daisy chain account really a true competitor of – mutually exclusive of – these other accounts? Could we say there are multiple grounds of murder liability in at least some supposed corpse cases?

   [I guess I wonder why we couldn’t have multiple routes to murder liability in a supposed corpse case. Suppose you think constructive liability is defensible at least sometimes – i.e. where it does not result in disproportionate criminal liability, as plausibly would be the case in a supposed corpse case. Suppose you also think culpability aggregation can sometimes be defended (as I’ve maintained in other contexts) where the bits of culpability to be aggregated really are of the same type (manifest insufficient regard for the same rights or values – e.g. respect for human life and dignity in a supposed corpse case). If so, then maybe we’ve got a couple stories to appeal to about why the defendants in Thabo Meli and Jackson plausibly could be convictable for murder. I’m not sure, but I wonder why not.]

   Anyway, the question I’m struggling with is the extent to which your daisy chain account could co-exist with other accounts like the above. If you’re just defending a sufficient condition of murder liability in these cases, then couldn’t you agree that in the same cases (or at least some of them) there are multiple routes to the conclusion that murder liability is justified?

   On the flipside, though, if there are multiple routes to murder liability, then maybe we wouldn’t need to work so hard to develop an account of concurrence between mens rea and actus reus to secure a murder conviction in these cases? What do you think?

   (I guess there would still be motivation for solving the problem through an account of concurrence if you could argue that constructive liability and aggregation are not defensible in general, i.e. ever, and moreover that the mens rea for murder should not be anything lower than intention to kill – though those are going to be hard conclusions to establish as a general matter, I think.)

   II. Rejecting the Proposed Conclusion

   The second Dodging Strategy floated in the comments is to bite the bullet and just say that we needn’t secure a murder conviction in supposed corpse cases: instead, maybe we say it’s enough to convict the defendant of attempted murder based on the t1 conduct and some other lesser offense based on the t2 conduct (maybe negligent homicide, maybe just desecration of a corpse). I took Mark to be raising this possibility in his second suggestion (4th paragraph) in his comment.

   I also have some sympathy for this route – especially given its simplicity. Two additional points in favor of this strategy: For one, it’s worth noting that some jurisdictions do allow attempted murder to be punished just as harshly as full-fledged murders – and supposed corpse killers seem like plausible candidates for this equal treatment. Maybe these jurisdictions are onto something and that is the normatively best approach for the criminal law to take. Second, this strategy might seem to have the added benefit of providing a more accurate label of what the supposed corpse killers did than would be the case if they’re simply convicted of murder, full stop.

   Given this, I wonder if you can say more about why it’s so important to secure the result that supposed corpse killers be convicted exactly of murder, rather than of attempted murder at t1 plus another offense at t2 if there is one – though maybe there isn’t in some cases? (That is, maybe the t2 lethal conduct doesn’t actually amount to a wrong, or at least not a culpable one due to some excuse being present.)

   So why not stick with a simple actuation account of concurrence (at least as being good enough for legal purposes), and accept that supposed corpse cases do not satisfy concurrence, but instead D is guilty only of attempted murder plus maybe an additional t2 offense if there is one? This would entail that sometimes attempted murder of V is accompanied by one’s purely accidentally and unforeseeably causing V’s death (i.e. causing it non-culpably). But maybe that’s ok insofar as it’s a more accurate description of what D did than lumping the supposed corpse killer in with the paradigm cases of full-fledged murder?

   I know from earlier conversations that you’ll have more to say here but just wanted to offer an invitation to say more about these wrinkles! Thanks so much and excited to see how the discussion unfolds from here!

9. Elise,

   I thought the paper was fantastic and have really enjoyed the conversation so far! Like some of the others in the comments, I am both antecedently partial to some variant of the causal theory, and a little suspicious of tracing-invocations. So I’ve found the comments, and the replies by Elise, really illuminating.

   Rather than re-tread that ground, I’ll just jump in to the exchange concerning the ‘mens rea dodge’ and the question of whether the intent to dispose of the corpse really does (ever?; in the extensionally correct cases?) “manifest the same malevolence that motivated the attack at T1.”

   Elise replies, in her initial comments notes that she “ha[s] doubts. For starters, the malevolence someone exhibits by disposing of a supposed corpse may often be less egregious than the malevolence exhibited when a defendant aims to kill; disposing of a corpse is just not as bad of a thing to want to get up to.”

   This got me to wondering if some of the disagreement (and a potential resolution) might be the result of underlying disagreement about what it is to “manifest” mens rea, and what is meant by “the same malevolence.”

   First, on manifesting mens rea. There are, at least, two different conceptions of the manifestation relationship which might justy and explain the concept’s employment in criminal law. On one conception, the agent manifests (or expresses) their actual subjective reasons for acting. This view provides a nice normative justification of why criminal culpability matters (defendants who manifests more malevolence really are more malevolent). But, notoriously, and for some of the reasons you point out, it has difficulty explaining positive law (or at least post-MPC American positive law). After all, two similarly situated purposeful defendants might have wildly different reasons behind their intended action, and so have wildly different concern for their victim. And two similarly situated reckless defendants, who both tolerate the same subjective risk of harm to their victim, might have had wildly different thresholds, or ‘reserve prices’ of risk that they would have tolerated, and so again have wildly different concern for their victims. As you note, and as Larry Alexander and Kim Ferzan (among others) have argued, this will make it very hard to claim that the MPC mens rea hierarchy has any close correlation to the degree of malevolence or insufficient concern manifested.

   One way to deal with this seeming lack of fit between the ‘subjective’ conception of ‘manifestation’ and the work manifestation plays in criminal law, is that it is the wrong conception of ‘manifestation.’ So some recent criminal law theorists have begun advancing an alternative conception that we might call the signaling or epistemic conception of ‘manifestation (of which Alex is one of the most persuasive and thoughtful defenders). Roughly, an act manifests a certain mens rea, just in case the act is a certain kind of (reliable) epistemic indicator, or signal, of the culpable mens rea.

   My suspicion is that these different conceptions might explain some of the disagreement here about how much malevolence or disregard is manifested, or expressed, at T2, though I’m still working through how exactly the cases would play out according to these different conceptions.

   Take the ‘typical’ supposed corpse case, a la Thabo Mali. As Elise/Alex argue, a ‘corpse disposal’ does not typically signal or indicate the same kind or degree of disregard as ‘a killing’. So, if the signaling conception of the manifestation relationship is right, the mens rea dodge is going to be difficult to pull off. The defendants in ‘typical’ supposed corpse cases, a la Thabo Mali, will not express ‘the very same’ degree of malevolence On the other hand, if we are operating with the subjective conception of manifestation, it may be, as Gabe suggests, that the corpse disposal really is motivated by ‘the very same’ (token identical) malevolence which motivated the attempted killing at T1, since it is (at least in the typical cases) part of the very same plan.

   On the other hand, it may be that, because typically, a ‘post-killing corpse disposal’ is part of the very same plan, perhaps we should say it does signal or indicate the very same (token identical} malevolence as the T1 killing. Which means that even in the exceptional cases, like the reformed burier, where the defendant’s subjective quality of will toward the victim at T2 is much better than at T1, some version of the signaling conception of malevolence can explain why the ‘post-killing corpse disposal still manifests the very same malevolence as the killing at T1, despite the fact that the subjective reasons for which the defendant is acting at T2 are quite different. (though of course, there might be further complications here if we bring in the issue of lenity).

   Elise, we’ve spoken at least once about this ‘signaling’ conception of manifestation/correspondence, in relation to an earlier version of the paper. So, as with Alex’s comment above, I know you’ve got some interesting responses. But I wanted to bring it up to note the potential connections between the debates about manifestation and your supposed corpse cases in the hopes that (A) the distinction can help shed some light on the debates in the discussion so far and (B) that maybe your corpse cases can help bring some further clarity to the debates about what the right conception of ‘manifestation’ ought to be.

10. My time zone has me facing quite a backlog! My apologies for the delay in response, but I’m really thankful for your attention and will work my way through to respond to each of these posts this morning.

    First, to James! Thank you so much for offering the UK perspective.

    I appreciate your pointing out that decisions from courts are sometimes most plausibly explained by the moral intuitions of legal actors, rather than much anything deeper. A strong moral intuition that X defendant is guilty pushes a court like the Thabo Meli court to uphold a conviction, and then they write an opinion with some hand waving. It’s worth noticing that in the Thabo Meli decision, after claiming that it would be “impossible to divide up what was really one transaction,” the court quite explicitly throws up its hands about saying much more: “Their Lordships do not think that this is a matter which is susceptible of elaboration.” I’m sure this same kind of reticence has featured in decisions where courts find proximate cause without saying much by way of justifying that finding.

    While I agree that courts are often driven by moral intuition, I’m trying to do what I can to unearth what explains those moral intuitions. Why is it that so many, including the members of the Thabo Meli court, have the knee-jerk reaction that the Thabo Meli defendants are guilty? If we know the answer to this question, we know how the Thabo Meli court might have elaborated in its decision.

    I think this is a worthwhile project, because if we don’t come to better understand what grounds the moral intuitions behind the tough decisions, that raises both substantive and procedural justice concerns. Substantively: In the absence of a better story, there is no reliable way of sorting out those defendants who should be acquitted for lacking correspondence from those defendants who should not be acquitted for that reason. Procedurally: I think it’s a serious injustice when courts are unable to provide meaningful explanations for why a defendant’s conviction is being upheld.

    This brings me to a worry that you raise, James: that the intuitions I’m seeking to explain might not be particularly uniform. Already in this thread, we have seen an example of the diversity of intuitions that can materialize. Gabe thinks we should account for Reformed Burier’s liability even though he does not have sufficient mens rea at t2, and Mark expresses skepticism that we’ll have to do so. Unfortunately, with the philosophical methodology I’m deploying, there’s always a risk that conversation might bottom out with a: “I simply don’t share that intuition.” But if a theory captures core cases well, and it enjoys some other virtues like systematicity and simplicity and perhaps ability to play with other plausible adjacent theories, then perhaps taking a hard look at it will help drive consensus on considered judgment in the more controversial cases for which pre-theoretical intuitions diverge.

11. Anthony: Thanks so much for standing up for the transaction analysis. As you note, I’ve left room to say much more about it, because in the paper, I only tackle one potential elaboration of a transaction principle (that is, the principle that would enable us to determine just when an actus reus is “stretched” across multiple instances of conduct). Thanks, too, for offering a Canadian perspective! It’s neat to see us a cross-jurisdictional dialogue unfolding on this thread. And I’m grateful you are calling the Cooper and Bottineau cases to my attention. I didn’t know about them, but it sounds like I should have.

    You say that the Cooper court provides an alternative to the plan-based transaction principle: what you call the fact-based transaction principle. But I’m not sure you’d need an appeal to a different transaction principle to account for the Cooper result. Why not use a plan-based principle? Couldn’t we just say that: Cooper had a plan to strangle his victim? And that he was carrying out that plan when he strangled his victim, notwithstanding the fact that he was blacked out when he did so?

    If the court had let go of a drive to locate contemporaneity and instead focused on a non-deviant causal relationship between an intention to kill and the killing, I think the case is simpler yet. Cooper had an intention to kill. That led him to non-deviantly strangle his victim to death. I don’t think there’s any barrier to calling the killing intentional. I would think that the fact of Cooper’s blacking out part of the way through the strangulation doesn’t remove the applicability of Anscombe’s “Why” question.

    I would think that Bottineau, too, could be tidily accounted for on a Causal Theory. That the defendants unsuccessfully tried to undo the damage done from a long period of starvation at the eleventh hour doesn’t, I think, impact the determination that in virtue of having starved the child to death, the defendants engaged in an intentional killing. Imagine that D ties V to some train tracks. But as the train approaches, D has a change of heart. D tries his hardest to undo the knots, but they’re just too snug, and he can’t untie V in time. I don’t think that by trying hard to untie the knots, D makes it the case that V’s killing is not intentional. D killed V by tying him to the tracks such that he’s run over by a train, and that V was so tied to the tracks was a non-deviant result of his intention to kill.

    Stepping back from the particulars of these cases, I want to consider what you have said about the content of the fact-based transaction principle. You suggest that the principle glues together instances of conduct based on normative considerations: based “on whether we think it’s morally fair to ascribe the entirety of a series of acts to a defendant such that mens rea overlaps with a part of one transaction.” I have a general worry with relying on normative considerations to fix conditions of correspondence. Without being very careful about just which normative considerations are getting loaded into the analysis, we may well be moving in an explanatory circle. If the normative consideration doing the work really is just “so and so better count as culpable for having X’ed,” then we are using culpability to explain a determination of correspondence, and we won’t be able to use correspondence to explain culpability without moving in a circle. But using correspondence to explain culpability is, I take it, exactly what we want correspondence to be able to do for us. It’s one ingredient of culpability.

    There are non-circular possibilities. But this is just to say I’d want to hear more about what normative considerations your fact-based transaction principle relies upon. Count me intrigued!

12. Thanks, Elise, for this amazing paper. I made the mistake of only logging on for this for the first time this morning, on Pacific Time, so I’m very far from caught up! So, apologies if the issues I want to raise have already been discussed in detail and fully resolved. Feel free to just say, in reply, “See what I said to Gabe et al.”

    I wonder if we can pick up the thread about the principle of legality. One of the strategies under discussion involves asserting that D’s mental state at t2, when they uncontroversially (I think?) perform the act of killing, suffices for the mens rea for murder. Since it’s an intention that does not ceteris paribus suffice for that–it’s an intention to bury a corpse, e.g.–asserting that it suffices for murder’s mens rea involves insisting that all else is not equal. Various views on offer involve different ways of specifying what it is about events in the t1-t2 interval that make the t2 mental state different from the normal case. If I understood you right, Elise, in one of your responses above (to Mark Schroeder, I think?) you say that it would be a violation of the principle of legality to convict D under theories of this sort. (I only skimmed, I’m afraid, but I think Alex Sarch also knocks around closely-related stuff.). After all, the law says, “Don’t kill with intent to kill” and D didn’t do that, on these kinds of theories. Instead, they killed with an intention different from, but just as bad as, an intention to kill. They also had, at t1, an intention to kill, but the act of killing was not performed, at t2, *with* that intention. One great advantage, it seems to me, of your proposed daisy-chain view is that it convicts in Thabo Meli and Jackson and other paradigm supposed corpse cases without loosening the legality requirements.

    But, I wonder if your view, in the end, might have the same problems with the principle of legality, popping up with respect to the affirmative defense. You say a word above about affirmative defenses, and I think your guiding example is self-defense. If the defendant in Jackson, say, had made the victim unconscious and blue at t1 *in self-defense* and then, at t2, had chopped off her head, killing her, and while believing her dead, he would be owed a self-defense justification. Why? Well, the law says “You are shielded from liability for killing someone with intent to kill if you acted in self-defense.” It’s not enough that the conditions of self-defense are met; they also need to correspond with the crime. The daisy chain view gives you a better explanation for why the conditions of self-defense corresponds in the variation on Jackson just offered than many competing view: the conditions of self-defense bore the ordinary same-time relation to the lead mens rea element of the crime, the t1 intention to kill, which was linked to the later actus reus element by a daisy chain. It’s much harder to explain on the view under which the t2 mental state is the lead mens rea element. That view needs some way to link the t1 self-defense conditions to the crime, all of the elements of which were at t2. So far so good. And notice that you are in an even a better position when it comes to excuses at t1, in contrast to justifications like self-defense. If Jackson meets the standards of the insanity defense at t1, when he has and acts on the intention to kill, thus causing V to seem dead, and is cured before t2, when he chops off V’s head reasonably believing her dead, the conditions of the insanity defense line up in the normal way with a lead mental element of the crime and so shield him from liability. (Maybe there’s also no liability under the view that all the elements of the crime are in place at t2; maybe all else is equal at the point for assessing whether the t2 mental state suffices for mens rea, which it doesn’t ceteris paribus.)

    But, now let’s imagine, instead, that the conditions of an excusing affirmative defense are met at t2. So, for instance, acting on the intention to kill, Jackson makes V unconscious and blue at t1 in the absence of any shielding conditions. And then, by the time t2 rolls around he’s contracted a severe mental illness and meets the conditions of the insanity defense. At that time, acting with something other than an intent to kill, Jackson cuts off V’s head. My intuition is that he should not be shielded from liability on grounds of insanity. But I think he would be on the daisy chain view, unless we loosen the principle of legality. After all, he performed the crime in a condition of mental illness and the law says that you are not liable when that is so. It is very tempting at this point to defend the daisy chain view by saying that mental illness shields from liability only if the crime isn’t just as culpable in its presence as it would be in its absence. That’s appealing. But it’s a violation of the principle of legality to deny the shield from liability on those grounds. After all, the law says you’re shielded if you committed the crime *in* a state of mental illness. It doesn’t say you’re only shielded if your culpability was reduced thanks to mental illness.

    (Here’s a possibly obscure remark: You might say that by embracing the MPC view that the mens rea elements of the crime need to be given evaluatively neutral descriptive definitions, you reintroduce evaluatively-loaded conceptions of the affirmative defenses.)

    Like I say, you may have covered all this already! Feel free to just ignore this post. And congratulations on the paper! It’s a great step forward.

13. Alex: I really appreciate the careful work you’ve done in crafting a landscape of where we are in this discussion, and I think your questions are really incisive. Thank you. You have two clusters of questions: first, about how my strategy sits next to other strategies for explaining liability for murder, and second, about why we should be so concerned to locate liability for murder all, rather than something else. I’ll say something about each in turn.

    First, you’re right to stress my sufficiency claim. If I’m merely making a sufficiency claim about how liability for intentional murder can be explained by appeal to a daisy chain relation, then does my proposal really crowd out any others? And if it’s more complicated than other strategies we have access to, then maybe the other strategies crowd it out?

    One virtue of my strategy, I think, is that it provides one analysis of what looks to be a unified phenomenon. There’s something going on in each of the supposed corpse cases that is structurally similar: a defendant does what they think they already succeeded at doing, as a result of thinking they’ve already succeeded. Bifurcated strategies like Gabe’s carve up the collection of cases that are supposed corpse cases and provide very different analyses of e.g. the typical case, like Thabo Meli, versus Reformed Burier. But to me, it looks like there’s one sort of thing going on here that could submit to one explanation. That’s not to deny that in some instances, other explanations might not also be apt. But my explanation works for a cluster of cases, all of which look like they have something normatively significant in common.

    The availability of alternative explanations also takes us back to questions about whether mentes rea really should be recognized at t2, or whether actus rei really should be recognized at t1. I have my doubts about both as general strategies, as I sketched in my response to Gabe. And I have my doubts that appeal to construction or aggregation get us all that far. These are special devices that require building out—When is construction warranted? When is aggregation warranted?—and in building them out, we may well encounter the very same complications we encounter when building out a theory of correspondence. This is what I alluded to in pointing out that a search for an aggregative principle more or less returns us to a search for something like a transaction analysis.

    Second: you wonder about the value of securing liability for murder in supposed corpse cases. You say that a strategy that focuses instead on liability for attempted murder and then whichever crimes might be relevant at just t2 would result in more accurately labeling what’s going on in the supposed corpse context. I’m not sure I agree with the labeling point. If we work hard to take our legal hats off, I think that there’s a pre-theoretical intuition floating in the background that a supposed corpse defendant is not just responsible for a bad unsuccessful killing, but that he is responsible for a *killing.* He killed the guy! Moreover, it’s not like he’s responsible for the killing in the same way that people are sometimes responsible for accidents, if they were negligent about bringing the accident about. He really went about killing that guy!

    Now, maybe these comments pertaining to labeling just go to show that I’m in the grips of my own theory. But I think that there’s something to the fact that the supposed corpse defendant is a killer and not just an attempted killer (even a pretty bad one). To me, there appears to be an important locus of responsibility at t2.

    Finally, I’ll note for purposes of the forum what we’ve had a chance to discuss offline, which is just: that I agree with you when you say it might be a good idea to stick with simple actuation * as being good enough for legal purposes.* This takes us to a distinction I do not make in the paper between what we could call the correspondence requirement—what some jurisdiction adopts as a legal rule having to do with correspondence—and the correspondence relation—the exculpatory phenomenon that correspondence requirements reflect. The best version of the correspondence requirement need not perfectly track the correspondence relation. To harken back to what I said in response to Mark’s point about moral hazard: there might be good institutional design reasons to have a simple correspondence requirement that works pretty well, in most all of the cases. But assuming there’s value to tracking culpability, before picking our normatively ideal correspondence *requirement,* I would think we still want to have a good understanding of the correspondence *relation,* just so we can tell whether trading simplicity in for imperfect tracking is worth the cost of the imperfect tracking.

14. Greg: Thank you so much for your illuminating post. I’m finding it really helpful as I wrap my head around what the implications might be of pushing forward with a mens rea approach at t2, and why I feel resistant to that approach. I’ll just note that even if the existence of a plan brings with it implications for which deficiencies in transaction with reason can be manifested by which instances of conduct, I take it that Reformed Burier is special precisely because he does *not* follow the same plan at t2 that he had at t1. So we’ll need another mechanism to be able to appeal to the transaction of reason he was up to at t1 as relevantly manifested by his t2 conduct…perhaps a daisy chain? 😉

    A biographical note: In a very early draft of the paper, I tried to write a section to show that my daisy chain conditions were normatively significant because when they were satisfied, the defendant “manifested insufficient regard” in such a way so as to be culpable for the killing just as the defendant would have been had they intentionally killed. That kind of section seemed important, because if manifestation theory is the dominant theory of culpability these days, and if I’m trying to say that correspondence is an ingredient to culpability (and, in fact, sufficient for it, assuming no applicable defenses), then it would be neat to show that my theory of correspondence can play nice with one dominant understanding of culpability that has been defended by philosophers I very much respect.

    I ran into a lot of trouble trying to get the section off the ground. The trouble concerned a feature that the theories advanced by both Gideon and Alex have in common: an appeal to lenity, or charitability. As you’re aware, Greg, according to Gideon, an evaluator of culpability must treat defendants “as little different from the law-abiding citizen as possible, given his behavior,” (Gideon’s Age of Culpability, at 89-90) and according to Alex, the deficiency manifested by wrongdoing is fixed by the “least amount of error” in responding to reasons that would be necessary to explain the wrongdoing. (Alex’s 2017 paper, at 733-34).

    The trouble with a lenity-constrained manifestation theory as applied to a case like Reformed Burier’s is that the *most* charitable way of explaining the t2 killing would refuse to look past the transaction with reasons that is most proximate to the killing. And, at least for Reformed Burier, that t2 transaction with reasons looks pretty good; *not* particularly normatively deficient. But then, there’s a point of tension between manifestation accounts that proceed by way of a charitability constraint and what I’m trying to say about supposed corpse cases: that they’re culpable for killings in virtue of their intentions to kill. We have to understand manifestation as no longer constrained by what the most charitable evaluator of conduct would conclude about the defendant’s behavior. And instead of asking how supposed corpse defendants fit pre-existing manifestation theories of culpability, we might have to ask how manifestation needs to reimagined in light of supposed corpse cases.

15. I had intended to comment on a number of items in Elise’s and Gabe’s excellent discussions, but I got too involved in the one item that particularly interests me (the causal test of concurrence) to have any space/time to talk about anything else. I fear that my comments therefore are more in the way of my own musings on the topic of concurrence generally and less of a direct response to Elise’s daisy chain solution to the killing of supposed corpse cases or to Gabe’s amendments thereto.

    Restating the causal test of concurrence more completely

    Elise is surely right that of the three concurrence tests she distinguishes, only the causal test is at all plausible. Dressler’s (and the courts’) temporal test is only a proxy for the causal test, and a good proxy only when the relata of the relevant causal relations are temporally close. The “one transaction” test is just a cheat, and an ad hoc one at that, invitive as it is of Mark Kelman’s old “time-framing” problem. So I focus on giving the causal test its best statement, at least as I understand it.

    Step back to what is generally needed for blameworthiness for some bad state of affairs (“s”)

    [NOTE: Types of events or of states of affairs are capitalized, particular (“token”) events or states of affairs are designated in lower case.]

    in the world. Let us stay with intentions as the culpability in question, although we could do pretty much the same analysis with beliefs (predictive beliefs about consequences and present beliefs about circumstances, of either certain degree (foresight, knowledge) or some likelihood degree (recklessness)).

    Four here salient items occur in the sequence of four at least slightly different times, t-1 to t-4:

    t-1: actor X intends that some state of affairs of type S come about via some action or omission of his.
    t-2: actor X (via numerous mediate intentions about means, together forming a plan) intends to
    perform some instance of the type of most basic act (“BA”) he knows how to do (usually a
    movement of his body) as his chosen means to producing some state of affairs of type S, or X
    intends to refrain from doing any action that would prevent some state of affairs of type S to occur.
    t-3: actor X’s body moves, a basic act-token (“ba”), or there is an absence of any preventative act by X
    (“~BA”).
    t-4: bad state of affairs s occurs.

    For X to be morally responsible for s occurring, two sorts of relations must obtain:
    First, a contingent, non-analytic relation: either:
    (1) X’s intention-token at t-1 non-deviantly causes X’s intention-token at t-2, and
    X’s intention token at t-2 non-deviantly causes X’s basic act token at t-3, and X’s
    basic act token at t-3 non-deviantly causes state of affairs token s to occur at t-4; or
    (2) X’s intention-token at t-2 counterfactually depends on X’s intention-token at t-1, X’s omission
    at t-3 counterfactually depends on X’s intention-token at t-2, and the occurrence of s
    counterfactually depends on the absence of any preventative acts by X at t-3.

    [NOTE: Given my views (that causation and counterfactual dependence are metaphysically distinct relations and that either may suffice to ground moral responsibility for some bad state of affairs), there is a third possibility here consisting of intended actions (that is, not omissions) standing in a counterfactual but not causal relation; for simplicity I ignore here this third possibility. Given another of my views (that absences cannot be the relata of singular causal relations) there is no fourth possibility consisting of a causal relations between intended omissions and act-token ba or state of affairs s.]

    Second, an analytic relation of instantiation, namely:
    (1) s at t-4 is an instantiation of type S which was the object of X’s intention at t-1; and
    (2) ba at t-3 is an instantiation of type BA which was the object of X’s intention at t-2.

    [NOTE: The relation is instantiation of a type by a token of that type, not identity between tokens; this, because intentions (like predictive beliefs) can only plausibly take types as their objects, and this, because the tokens are unknown to the actor as she acts. We don’t literally intend (or predict) the fire that happened ln the warehouse yesterday, however idiomatic it is in English to say so after the fire has occurred and can be referred to as a particular; we only intend (or predict) that some instance of the type, fire, would occur. Thus, although the intentions at t-1 and t-2 are token intentions, their objects represent types of actions or states of affairs.]

    Correspondence, as I would analyze it, consists of both of these relations obtaining between the four items above (t-1 intention, t-2 intention, t-3 basic act, and t-4 state of affairs).

    A word about each relation. As to the first, the non-analytic relation, a lot of work is done by the notion of deviance.

    [NOTE: Deviance was developed with respect to the causal relation. But if one believes as I do that counterfactual dependence is a desert-determiner independently of causation, some notions of deviance are needed in order to cabin that relation to an intuitively acceptable scope of what we are responsible for.

    Often causal deviance in these contexts is divided up between upstream deviance (the t-1/t-2 and t-2/t-3 connections) and downstream deviance (the t-3/t-4 connection). Chisholm’s and Davidson’s 1950’s/1960’s examples of deviance were given in the context of assessing the causal theory of action; they were thus concerned with upstream deviance. E.g., Thebo Mali is an epileptic whose intent to kill his victim by striking him so excites and stresses him that he goes into a grand mal seizure the bodily movements of which strike and kill the intended victim. Downstream deviance is what the lawyers call proximate causation; wildly freakish, improbable sequences of events through which defendant’s bodily movements causally contribute to some bad state of affairs occuring are said to make defendant’s contribution to be “remote” and “insubstantial,” i.e., downstream deviant. Only downstream deviance is in issue in cases like Thebo Mali.

    It is right to distinguish these two forms of deviance in causal chains, for they differ in their analyses. More than 50 years ago Alvin Goldman and I both argued that it was a mistake to attempt to give a philosophical analysis of the criteria for upstream deviance; we argued that the “right way” (i.e., non-deviant way) for intentions to cause bodily movements so as to produce actions, was a matter for scientific discovery, not philosophical analysis. The neuroscience of voluntary motor movement starting with Sherrington has largely borne us out in this hope, for now there are detailed mappings of how intentions to move issue in the movements intended.

    Whereas it is implausible to hope for a similar physicalistic reduction for downstream deviance. Most of the theories of proximate causation is use in the law of today, for example, are not plausibly regarded as consisting of physicalistic or causal discriminations of any kind. Whether a harm was foreseeable to some actor, for example, is not a causal question; when the intentional action of a third party constitutes an intervening cause (and thus is said to “break the causal chain” between defendant’s act and the harm) is also not a causal discrimination but at best a moral one. Such tests largely treat downstream deviance as a matter to be resolved by policies extrinsic to causation, not as a matter to be resolved by discovery of a nature to causation or of some distinct kind of causation.

    Elise nonetheless spends a bit of time “trailing off into the foggy regions where casuists and lawyers ply their trade…: (Zeno Vendler’s 1984 characterization of the legal notions of proximate causation), speculating about whether the victim’s death by exposure was or was not foreseeable to Thebo Mail when he clubbed the victim on the head. But her heart is not in it because she sees too clearly the vacuous nature to foreseeability. If she wished a more extensive sojourn in these foggy regions, she might have asked Hart and Honore’s direct cause kind of question, viz, was the act of throwing the victim off the cliff an intervening cause breaking the connection between the clubbing of the victim and that victim’s death – perhaps not, because of the lack of both intentionality (of death) and of motivational independence for the allegedly intervening act of throwing?

    If we leave the law to the side and focus only on the morality of the matter, did Thebo Mali’s clubbing of the victim causally contribute (in a non-deviant way) to his death? Admittedly, there is an unusual nature to the way the causal chain worked — the clubbing lead to a belief that there was a body to dispose of, and this lead to the throwing of that “body,” which in turn lead to the exposure, and this in turn lead to the death – but is that more freakish than, e.g., the “hard way” cases of holes in one in golf where one gets credit, i.e. where one “makes a hole in one,” not by an accurately aimed square hit on the ball but by an inaccurate slice that sends the ball into a tree and then ricocheting into the hole?

    As to the second relation, the analytic one, things may be no easier than with the first relation for cases like Thebo Mali. First, however, let us be clear about the nature of these relations. To begin with, notice that these instantiation questions are not to be confused with another instantiation question that also must be asked and answered in order to find responsibility for some bad state of affairs. Laws and moral norms are general, the facts of particular cases are particular. To apply the law/norm to the facts of a case, one must therefore ask whether the act token done by the defendant instantiates the type of action prohibited, and one must also ask (for crimes requiring intention) whether the intention token with which defendant did his wrongful act, instantiates the type of intention required for conviction/guilt. In that latter question one is using as a base of comparison the statutory (or moral norm) description of the object of the prohibited intention, which will be quite general and unnuanced. Not so the question in which we are here interested: here the base for comparison is the psychological intention actually possessed by the defendant, and we ask how what he did matches up to that intention’s object. Thus it will not do to say (as is commonly said) that Thebo Mali intended the death of a human being (the statutory description) and what he got was the death of a human being. Rather, what he intended was to club his victim to death at one time and what he got was the exposure death of the victim at a later time. This last conclusion does not mean that therefore the state that Thebo Mali caused does not instantiate the type of state that he intended; for the match requirement here is not so precise in its demands of a “match.”

    Suppose defendant has the intent to disfigure required for conviction of mayhem. With that intention he hits his victim with a stick intending to put out the victim’s left eye; seeing the blow coming, the victim turns his head at the last minute to avoid the blow and his right eye is blinded. Few I suspect would think that this is not “close enough for government work,” i.e., that what the defendant caused was close enough to what he intended to cause that he can be held liable for intentionally (and intendingly) causing it. Mayhem, right, not negligent or reckless disfigurement?

    [NOTE: Many seem to find equally easy the classic cases of “transferred intent,” viz, cases where the only difference between what was done and what was intended was the identity of the victim to whom it was done. Anglo-American law has for centuries treated these as easy cases of there being a close enough match to sustain liability for intentional torts and intentional crimes even though the match of course is not perfect.]

    For comparison, take the facts of Regina v. Pembleton: what defendant intended was to strike another person with the rock he threw; what he did was to break the plate glass window behind that person with the rock that he threw. This is not an intentional (or intended) destruction of property but only at best a negligent or reckless destruction.

    Where is Thebo Mali along this continuum? What he intended was a beating death, what he got was an exposure to the elements death. Close enough for murder? It matters to me that part of the plan was not just the beating to death of the victim but also avoiding detection by throwing the body over the cliff;

    [NOTE: Cf. Edmund Hillary when confronted with the possibility that George Mallory had first summitted Mt. Everest: “I think that part of what one intends when one intends to climb a mountain is also to intend to get down alive.”]

    the latter makes what actually happened one step closer to what overall was intended, although the fit of course is still far from perfect.

    Why should morality (and a just criminal law built upon that morality) care about either the freakish causal routes determinative of the first relation or the mismatch of plans to results determinative of the second relation? Because I think we think it fair to blame people only for that over which they have sufficient control. When defendants obtain their intended results only by bizarre causal routes, or when they obtain results that do not closely match the results they intended to achieve, they didn’t control things to an extent making them responsible for the bad state of affairs that has occurred. We possess the limited control we do have over what happens in this world only by having mental states of desire, belief, and intention that represent the world as we want it to be, think that it is, or intend to make it, and by having the capacity to produce acts and states of affairs conforming to those representations. The worse the fit or the more tenuous the causal/counterfactual connection, the less control we have and the less we can be blamed for exercising.

16. To Gideon (I’m so sorry I’m still behind on the thread! My apologies to those trying to follow along with the zig zag): Thank you so much for your post and again, for the heroic amount of support you provided to me as I worked on this piece. And thanks too for bringing up affirmative defenses! They are the large can of worms I am still frightened to crack open!

    A point of clarification first. In response to Gabe, I offered what was just a preliminary worry with positing a mens rea substitution principle for finding mens rea at t2. The preliminary worry was that by convicting someone for killing at t2 in virtue of their mental state at t2, we could be convicting them absent violation of the law. But I didn’t think the principle of legality concern ran that deep. Of course, it only gains traction in systems that don’t have mechanisms for treating a supposed corpse defendant’s t2 mental state/s as mens rea states. For that reason, I don’t think it helps me out that much in the end. I took Gabe to be supposing that a legal system *could* recognize a mens rea state at t2 through legal recognition of such a mens rea substitution principle, and that a legal system indeed *should* do so, so as to account for the intuition that supposed corpse defendants are liable for murder in virtue of killings at t2. If a legal system did embrace and expand mens rea in this way, then that looks to remove what’s puzzling about supposed corpse cases, because the correspondence relation can remain plain old non-deviant causation, just at t2. Rather than appeal to the principle of legality, I think the stronger thing to say in response to Gabe’s mens approach would be to say that it is normatively indefensible for a legal system to go about endorsing the kind of substitution he proposes (or, at the very least, that it is normatively defensible for a legal system to *not* endorse such a principle), so the inculpatory phenomenon is better found as located away from the psychology at t2 and at a different spot—at the correspondence relation, eg.

    To turn to your examples: I’m imagining that satisfaction of the daisy chain alone won’t be particularly informative about the conditions under which a defendant should have recourse to an affirmative defense. I’m imagining that by satisfying the daisy chain, the defendant will be in a normatively equivalent position to the position that they would be in if their killing were intentional. Then we have to ask, of that quasi-intentional killing: is it the sort of thing that was nevertheless justified or excused, such that the defendant is not culpable for it, after all? And if a legal system had, as a rule for what constitutes a legal excuse, that: anytime an insanity condition co-occurs in time with an actus reus, the defendant is acquitted, then that might just be a pretty bad rule for a legal system to have, because it fails to track well the *moral* exculpatory phenomenon.

    I hope this clarifies, but please let me know if not, or if I’m missing the ball! Thank you again.

17. Michael: Thank you so much for your contribution. I’m very appreciative of your having spent the time!

    You’re tempted to conclude that there’s a non-deviant causal relationship between the Thabo Meli defendants’ intention to kill and the death of their victim. If that’s so, then Causal Theory need not be supplemented by any kind of fancier daisy chain that posits, as normatively equivalent to non-deviance, some *deviant* causal connection (as I’ve done in my paper). The question that occurs to me in digesting the proposal is just what we mean when we refer to non-deviance. Let me explain.

    Part of what I’ve done in my paper is construct what I take to be the most charitable version of Causal Theory on offer. Some people who have subscribed to Causal Theory have just used more words—words like “actuation” or “motivational concurrence”—to talk about the relation they mean, but that’s not exactly illuminating. So the first step in my discussion of the view was to add substance to the theory by supposing that what Causal Theory appeals to is the very same relation, whatever it is, that grounds ascriptions of intentionality when we say that a defendant intentionally X’ed in virtue of their intention to X. The way I put it in my paper, on this understanding of Causal Theory, the question of what constitutes the correspondence relation just is the same question that causalists of intentional action ask themselves when they ask what causal relation is required in order to make it the case that an action counts as intentional, in virtue of one’s intention. That would be great, as reducing the number of live questions counts as progress in philosophy!

    This is the relation I refer to when I refer to non-deviant causation. But I think that there’s pressure to say that the killing in a supposed corpse case *wasn’t* actually intentional. It sounds weird to call them intentional, at least. Anscombe’s “Why” question doesn’t look like it gains application, because a supposed corpse killer’s reasons for killing might not serve in the explanation for why it is that they kill, when they do. But then non-deviant causation understood as the grounds for *intentionality* looks to be just one particular sort of causation grounding *culpability,* and not the only kind.

    With that said: “non-deviant causation” is just a couple of words mushed together. If, *instead,* what the sophisticated Causal Theorist means in referring to “non-deviant causation” was just: whichever causal connection, obtaining between intention and result, is such that a defendant is morally responsible for the result in virtue of the intention, then some causal connections that fail to support intentionality could nonetheless be “non-deviant” in this modified sense. My daisy chain causal connection would then count as non-deviant rather than deviant.

    As an aside: You signal that part of why we might be comfortable finding a non-deviant causal relationship here is that the defendants in Thabo Meli were already planning to do as they did at t2: to toss their victim over the cliff. If we understand the defendants as having an intention to: kill-the-victim-and-then-toss-his-body-over-the-cliff, it feels a little ridiculous to say that they somehow escape responsibility for the death because the victim didn’t die exactly when they expected he would, but just, a little bit later on in the plan. I’d be curious how much work you feel appeal to a plan does in establishing a finding of non-deviance. What to make of Reformed Burier, who abandons his plan at t1 by the time he kills at t2? Does non-deviance still apply to cover his case?

    Thanks so much again, Michael!

18. Really enjoying all this discussion! Of the multitude of issues I’d love to follow up on, let me just pick up one and follow up on one thing that came up in the back-and-forth between Greg and Elise. [Very sorry this comment became so longer than anticipated.]

    I agree there are tight connections between the correspondence issue and the manifestation issues, so maybe one can illuminate the other. One thing this discussion has clarified is that if you like a lenity approach to manifestation of insufficient regard (calculating degrees of culpability), there is a question about whether lenity should also be applied to push us towards picking an act description that allows post-offense conduct to affect your overall culpability or whether it should not do so and culpability must be assessed for each bit of conduct considered separately (even if this leads to a harsher overall result for the defendant). I think I prefer the latter option and I’ll try to explain why.

    Here’s the comment by Elise in response to Greg, which I think raises this most clearly: “The trouble with a lenity-constrained manifestation theory as applied to a case like Reformed Burier’s is that the *most* charitable way of explaining the t2 killing would refuse to look past the transaction with reasons that is most proximate to the killing. And, at least for Reformed Burier, that t2 transaction with reasons looks pretty good; *not* particularly normatively deficient.”

    I am not sure this is right. Lenity might be applied in two ways — one where culpability assessments are complete immediately after the act being considered concludes (according to the most apt act description to be chosen), or another way where subsequent good conduct can mitigate the badness of the earlier conduct.

    (a) More specifically, one way to apply the lenity idea (which I think I probably prefer) would be to assess the culpability of the t1 conduct and the t2 conduct separately, and without allowing the assessment of the one bit of conduct to bleed into to the assessment of the other. By the defendant’s lights (i.e. taking as true the factual beliefs of the defendant — as I think we should in assessing culpability), the most charitable assessment of the quality of RB’s transaction with reasons would put them at same depths of depravity as an actual murderer. After all, at t1 they take themselves to have killed intentionally. And this is true even on the most lenity-infused interpretation of how they transacted with reasons at t1, where this conduct is considered in isolation. True, RB’s beliefs are mistaken, but this seems to be an actus reus issue — not pertinent to assessing their level of culpability. (At most it affects which offense they should be convicted of: murder or attempted murder.) When it comes to the t2 conduct, yes RB has experienced a change of heart and so there’s no additional culpability incurred from the t2 conduct. But that’s it: the t2 virtuous conduct does not make up for or reduce the culpability of what was done, or how RB transacted with reasons, at t2 according to this approach.

    (b). Another way to apply lenity — which I take to be implicit in Elise’s comment above (sorry if I misunderstood) — is to assess RB’s transaction with reasons at t1 and t2 separately, but to allow that RB’s morally good transaction with reasons at t2 CAN affect RB’s overall moral ledger. That is, RB’s good/remorseful conduct at t2 might be thought to partially make up for the badness of RB’s t1 transaction with reasons. After all, that’s what RB is TRYING to do: At t2 RB seeks to somewhat make up for the badness of what s/he did at t1. So if t1 incurs +100 units of culpability, for example, then maybe the t2 conduct contributes -20 units of culpability for a total of +80.

    [Note that there’s maybe a third way to go though it comes to much the same as (b). Greg’s comment suggested that in the original supposed corpse case, like Thabo Meli, we might consider the t2 conduct to be part of the same plan as the t1 conduct, and so fairly seen as an expression of the same underlying token malevolence. In the RB case, it’s not the same plan that’s at issue at t2, but maybe we still should consider the t1 and t2 conduct together as the one single transaction. Nonetheless, I take it this gives roughly the same overall assessment of RB as we had in (b). Here’s why. Call this option (c). If we go this route, then a lenity-constrained assessment of RB’s reasoning that motivated the t1+t2 conduct as a whole would also be something less than the very bad transaction with reasons that is on display in the t1 conduct considered by itself. Plausibly, this would be about the same as the overall culpability suggested in (b). That is, something less than +100 — plausibly involving a similar reduction of -20, to put the total at +80. Just like in (b).]

    Ok so now the question is whether the lenity idea as applied to manifestation (and thus the calculation of degrees of culpability) mandates that we select option (b) [or maybe (c) if it yields the same culpability result] rather than option (a). I took it that Elise’s comment quoted above suggested that lenity DOES require (b) [or (c)] rather than (a). But that’s what I’m not sure of.

    I think it’s at least plausible that there can be good institutional design reasons for the criminal law not to allow conduct carried out after the completed offense (i.e. an attempted murder at t1) to mitigate or reduce the culpability of that crime — at least in a way that affects one’s liability to conviction (i.e. whether one is guilty and how serious a crime one is guilty of). Granted it might matter morally or even appropriately shave a few years off one’s sentence. But to prevent defendants from opportunistically doing good things or dishonestly try to make it look like they’ve had a change of heart after they’ve completed their crime, probably at least the criminal liability rules (if not the more opaque downstream sentencing rules) should be largely insensitive to subsequent good conduct showing a change of heart. (Consider for example the lack of an abandonment defense to attempt and the very high bar to demonstrating a similar defense to a conspiracy.)

    If that’s right, then I worry about the comment above from Elise and so maybe there isn’t so much tension with the orthodox picture of manifestation after all. The reason is that when we consider just the culpability of RB (i.e. the insufficient regard manifested by their conduct even applying a lenity approach), then if we plunk for option (a), RB’s culpability remains just as high in virtue of her t1 reasoning/conduct as the culpability of the full-fledged murderer. [This picture assumes results don’t matter to culpability — only to labeling and perhaps sentencing for non-culpability/desert reasons.] If that’s right, then can’t the traditional manifestation picture get us pretty much everything you want in the RB case, Elise?

    Sorry for the unreasonably long comment — getting late here in the UK. Would love to hear your thoughts sometime, whether here time permitting or offline. Thanks!

19. Thanks so much for following up, Alex. Your sketch of the possibilities here is very helpful. I am afraid that in responding to Greg, I was resorting to a little bit of short-hand and I was not being very clear about my aspirations. My apologies for this! So let’s see if I can clarify.

    I agree that the culpability for t2 conduct shouldn’t impact the culpability for t1 conduct. So we are on the same page there. The trouble I was experiencing with an appeal to lenity-constrained manifestation had to do with a trouble in accounting for a supposed corpse defendant’s culpability for their *t2 conduct.* If a supposed corpse defendant is justifiably held liable for murder, which is to say, a kind of killing, and if that is so because a supposed corpse defendant is culpable for having killed, then I need a way to explain why the supposed corpse defendant was culpable for the t2 conduct *of killing.* Focusing just on culpability for t1 conduct, however grave it is, doesn’t get me culpability tied to the act that constitutes the actus reus for the crime I’m trying to say the supposed corpse defendant can justifiably be held liable for.

    To explain why a supposed corpse defendant is culpable for his t2 killing, I have to say quite the opposite of what you propose at (b). I have to say that a defendant’s culpability for his t2 conduct can be grounded in deficiencies having to do with an earlier, t1 transaction with reasons. But this is what it looks like lenity bars, because it looks like lenity would force us to look only locally, at the good transaction a supposed corpse defendant like RB exhibits at t2, generating the intention to bury their corpse.

    Note that my woes about how to handle lenity and manifestation in this arena need not turn into *your* woes, if you are happy to take the approach recommended in the second prong of your first post: if you give up on the project of explaining why the supposed corpse defendant is justifiably held liable for killing. But insofar as my aim is to explain justifiable liability for killing, and in virtue of culpability for a killing, I need to be able to look past the good proximate motives a supposed corpse defendant might have had for doing that which constituted the killing.

20. Elise, Gabe, and Gathered Chorus,

    This is really a fantastic discussion. Elise, your paper does a really wonderful job setting up the puzzle and working through a number of potential avenues, and I found the development of the Glanvillian mistake line of thinking helpful and provocative. And the resulting conversation has been so rich: I found many of my initial thoughts in the comment threads, either better developed than I’d had them or thoroughly answered (and often flummoxed).

    But let me push on the relationship between the intuitions about supposed corpse killers, the warrant for convicting supposed corpse killers, and principled versus institutional rationales for so convicting them. (Here, my thoughts will especially borrow from Alex, Mark, and Paul Robinson.)

    What would warrant convicting these supposed corpse killers of successful, intentional murder?

    First, if the good ways of thinking about what counts as success would include these cases, then we’ve got ‘em. But I take the dialectic in the paper to be 1) the existing ways of thinking about concurrence fail to capture these cases and so 2) here is the outline of an alternative to capture these cases. Because you are here arguing for a new alternative, I don’t think good ways of thinking about success can yet explain why we should convict these supposed corpse killers of successful, intentional murder. Indeed, the thrust of the article seemed to be that we should be attracted to this new way of thinking about success precisely to help us capture the thought that we should convict these supposed corpse killers—which goes exactly the other way.

    Second, we might be warranted in convicting these supposed corpse killers of successful, intentional murder if that best captures important intuitions. If it is intuitive that these are successful, intentional murderers, then we’ve got ‘em. But it is not clear that we have that intuition! I don’t have that intuition, for one. And, in the essay, the language in Section Two tells against that intuition. There, the intuition that is described is of equivalent culpability. That is, culpability the same as successful, intentional murder, rather than culpability for successful, intentional murder. Of course, we might challenge that intuition. How could this messy stuff have equivalent intuition to that one thing? Don’t we need some justification for things like aggregation or comparison? Down the road, maybe. For now, all we need to see is that, at least as far as I can see, there is no shared, stable intuition that these are successful murderers. We just have shared, intuitions that these killers are culpable like successful murderers.

    Because of these two thoughts, it seems to me that the argument cannot be from theory or intuition to the daisy chain story.

    Third, we might be warranted in convicting these supposed corpse killers of successful, intentional murder even if they aren’t actually successful, intentional murderers if 1) they are culpable in much the same way as successful, intentional murderers and 2) there are institutional design reasons that tell against complicating the criminal law in the ways that might be required to capture precisely what is going on.

    I think the combination of related attempted murder and negligent killing is more culpable than the sum of its parts, and I think that related combo is roughly as culpable as successful, intentional murder. We could have an independent crime that is the combo. But for proof and administrative reasons, I am perfectly happy to convict these folks of good, old ordinary murder. Why? There are probably few of the most puzzling cases, and the more common cases probably raise useless issues of proof. Imagine that A fires six bullets into B. The first five knock B down and stun B, and A thinks B dead. The sixth shot is intended to gloat and to warn others of how vicious A is—but it is actually the only killing shot. This is a supposed corpse killer, and yet I am perfectly happy if this person is convicted of murder, and I am even more happy if prosecutors and courts can avoid having to hash out that sort of evidence.

    As a bit of supporting argument: I think that this is an attractive way to make sense of culpable-ignorance cases more generally. Why should we convict a culpably ignorant smuggler of knowing smuggling? Because 1) the combo of culpable ignorance and negligent smuggling (or very culpable ignorance and innocent smuggling) is as culpable as knowing smuggling and yet 2) creating a bunch of additional crimes or disjunctive mens rea provisions is so much more complicated than adding one general culpable ignorance provision! If this practical rationale is attractive for culpable ignorance, and if you’re appealing to culpable ignorance, then I’m even more tempted to a practical rationale here.

    (What about supposed corpse killers whose second killing is not negligent? I find it very hard to fix my intuitions in those cases, especially because I suspect that I think that attempted murderers might have distinctive epistemic duties or even strict epistemic liabilities as to their would-be victims. If so, then the second killing will almost always be negligent. In any case,I haven’t yet been able to come up with a case that is seems to me to be clearly both non-negligent and also still fully culpable.)

    Elise, you have considered and responded to comments at least in the neighborhood of this thought above. And, consistent with Alex’s comments, I don’t take any of this to be reason to reject daisy-chain thinking—just that it takes away some potential support for daisy-chain thinking. So let me close with this:

    It is thus hard for me to see how to get out of the gate. If we have a plausible, grounded-in-practice explanation for how we do practice, and if our intuitions do not provide much argument for or against the in-practice explanation, then I am not sure how to start an argument aimed at deep metaphysics on the basis of contested and cloudy intuitions. So, perhaps meta-philosophically, is there a helpful way to force the argument to be on your turf?

21. Hi, Elise. Thanks for these helpful replies to my initial comment! I apologize for being slow to follow up – this page wasn’t re-loading for me so I didn’t see the comments after mine until now and I was puzzled why there wasn’t a more intense discussion going on.

    On the constraints of working within a model penal code-compatible picture; that seems like a really helpful methodological posture as a way to investigate, though not one that comes naturally to me, since I’m not a proper legal theorist. But at the same time, it’s not a priori obvious to me that the kinds of non-doctrinal reasons that you have to think that it is *right* to treat corpse cases as murders aren’t also putting pressure on those constraints. I’m not really sure how to think in a principled way about which structural features of a helpful framework like the MPC shouldn’t be in play when we’re theorizing normatively about what shape the law *should* have, given prior intuitive judgments about the shape it should have.

    I am tempted, as you note, to deny that the reformed burier case ought to go with the others, and agree that if I were not, the “especially serious attempt” picture would be more straightforwardly unappealing. I’m not sure to what extent, however, my intuition about this case is independent of my being a bit sympathetic to this idea and to the mens rea at t2 idea. I can’t say that I really like the “adding up” approach, myself; it feels too unprincipled whether the adding up is the right amount and the justification for departing from the clean connection between culpability and liability too messy.

    I agree with everything that you say about doctrinal design, but I do find this particular moral hazard to be troubling. It’s part of what makes me feel, despite some initial caution, like your guiding judgment about the main corpse cases has to be correct, even if the law is constructed in such a way as to give it a cleaner justification.

22. Craig: Thanks so much for your challenging set of comments. When I hit “meta-philosophically” I admit I shuddered a little bit, but I will do my best to be responsive.

    As I mentioned to James, I think what I’m after *is* in fact an explanation that accords with a pretty stable intuition–the intuition that Williams reports. The intuition is that even though courts like Thabo Meli can’t articulate good reasons for finding supposed corpse defendants liable for murder, the result is justifiable.

    The deeper intuition that I took to be underpinning this stable liability intuition is an intuition about culpability: that supposed corpse defendants can be justifiably held liable for murder because supposed corpse defendants are just as culpable for their killings as are supposed corpse defendants of the garden variety.

    If I’m understanding you right, you point to a different intuition that could be underpinning the liability intuition: that supposed corpse defendants can be justifiably held liable for murder because supposed corpse defendants are roughly as culpable as are supposed corpse defendants of the garden variety, when you consider their aggregate culpability.

    As I have mentioned in response to others’ posts, I think there’s some squirreliness with the underlying premise: that the culpabilities for t1 and t2 conduct are going to come out as both (a) summable and also (b) when summed, equivalent to the culpability of a garden-variety murder. And these are details that are going to matter a lot to the plausibility of the view. Say that a defendant perpetrates 100 counts of mail fraud and he also runs a red-light and kills a pedestrian. Presumably, it’s not as if there’s a total culpability here that would justify our holding *this* defendant liable for intentional murder.
    When understanding how to distribute criminal liability, it matters to us not just to what extent a defendant is culpable, but also, what a defendant is culpable for. My choice to vindicate the culpable-for-killing intuition rather than a culpable-full-stop intuition scores some points, I think, simply in virtue of the fact that if that intuition is vindicated, then the story about liability for act X appeals to culpability for act X and not something else.

23. Great discussion!

    Echoing you, Elise, I first want to emphasize how important I think your inquiry is: even if resultant moral luck is illusory; even if some attempted murder is as blameworthy as actual murder; even if attempted murderers should be punished just as severely—I still want to blame and punish supposed corpse killers for killing, not for trying. That’s why my preferred “dodges” don’t dodge the basic problem: they’re ways of inculpating supposed corpse killers for murder.

    Since you expressed uncertainty about the details of my mens rea “dodge,” I’ll take another crack at it here. The aim is to describe circumstances where we’re justified in believing a mistaken killer manifests a quality of will (“M”) equivalent to that of an intentional murderer. Begin with the (concededly debatable) premise that D manifests M if (i) D causes V’s death by an act D knows would likely kill a living person, and (ii) D would have performed this act even if D had known V was alive. Evidence for such counterfactuals is often hard to come by. But not in the typical supposed corpse case: that D tried wholeheartedly to kill V at t1 is evidence D would have killed V intentionally at t2 had D known V was alive—evidence reinforced by D’s callous disposal of V’s body, which implies persistence of the relevant culpability-constituting/counterfactual-grounding dispositions. Likewise when D tries to kill W at t1, mistakenly believes W dead, mistakes V for W at t2, and callously disposes of V’s body, killing V unintentionally. As I proposed in my précis: “[S]upposed corpse killers act at t2 with ‘sufficient mens rea to convict of murder’ if they cause their victims’ deaths by acts they know would likely kill a living person, if they believe their victims dead because they believe they killed them, and if they show every sign at t2 of being animated by the same malevolence that motivated their attack at t1.”

    You asked: What of the “defendant who hallucinates killing at t1 and then, with malevolence, buries the supposed corpse,” or the defendant who at t1 intends to kill in defense of others, exhibiting “laudable motives,” but at t2, “full of hatred,” buries the supposed corpse “unceremoniously”? Because neither defendant attempted to murder anyone at t1, we have weaker evidence that either would have killed their victim intentionally at t2 had they known their victim was alive, and thus weaker evidence that either manifested M when they mistakenly killed. (It still could be true that they did. No matter how we craft the doctrine, some killers will probably get away with (the moral equivalent of) murder.)

    On normative neutrality: while I don’t agree that the MPC’s culpability framework is normatively neutral (note MPC 2.02(2)’s “substantial and unjustifiable” risks), I see no reason why we couldn’t formulate a mens-rea-at-t2 standard in MPC-friendly terms. Instead of asking whether D showed every sign at t2 of being animated by the same malevolence as at t1, we could ask the jury: “Would D have cut off V’s head [or thrown V’s body off a cliff, etc.] even if D had known V was alive? In answering, consider the nature of D’s conduct toward V at t1 and t2 [or toward W at t1 and toward V at t2, whom D mistook for W].”

    On legality: to the extent that normative terms are vaguer than non-normative terms like “intention,” legality favors neutral over normative formulations, just as it generally favors sharp phrasings over vague ones. But (as you said to Gideon) legality doesn’t inherently favor one approach to supposed corpse cases over another; it all depends on how murder is defined.

    I think it’s worth pointing out that legality doesn’t favor the daisy chain theory where murder is defined as intentional killing (e.g., traditional U.S. first-degree murder as “willful, deliberate, and premeditated killing”; MPC murder as “caus[ing] . . . death . . . purposely[,] . . . knowingly . . . or . . . recklessly under circumstances manifesting extreme indifference”). Defining murder as intentional killing is nothing like defining a murderer as someone (a) who possesses the intention to kill, (b) who performs an action that causes death, and (c) whose intention and action “correspond.” If murder were defined the second way, the daisy chain theory would have the inside track, at least as regards legality. But where murder is defined as intentional killing, the legality-favored view would appear to be the causal analysis that locates mens rea and actus reus both at t1.

    I still like that analysis, even if you insist that “proximate causal claims” must be “ordinary causal claims.” If a second person, Parker, had decapitated the woman Jackson poisoned in Jackson v. Commonwealth, wouldn’t an ordinary causal analysis support the claim that both people played causal roles (Jackson at t1, Parker at t2), despite the “mismatch of plans to results” that Michael discusses? If so, why not say Jackson played a causal role at t1 in the actual case? How could Jackson’s t2 decapitation make his t1 poisoning not causal if, in the Jackson-Parker case, Parker’s t2 decapitation doesn’t make Jackson’s t1 poisoning not causal? I agree it “runs quite counter to ordinary notions of causation” to call Jackson’s t1 conduct the cause of death, especially in the one-defendant case. But the law needn’t call it that to classify it as the actus reus of murder—no more than in the Jackson-Parker example or in the case of the terrorist whose victim was buried alive by her family.

    Finally, I’m intrigued by your suggestion that so-called conduct crimes (which don’t involve causation) might yield supposed-corpse-like problems requiring daisy-chain-like solutions. But I want higher-stakes examples. My intuitions about Sloppy Stealer are banal: at t1, attempted theft (unless U.S. v. Oviedo controls); at t2, conversion. Am I being lazy? Maybe I’m just less indignant about a converter getting away with theft than a killer getting away with murder.

24. Thank you for circling back Gabe! This is such great stuff.

    I especially appreciate your spending more time unpacking the mens rea approach. I think it’s a significant challenge to my view and I clearly need to think more about it.

    The thing interesting me most about it is the role you see the t1 attempt playing in setting t2 mens rea. (I think this also ties in to the back-and-forth between Greg, Alex, and me about whether, and how, what happens at t1 can end up relating to what is ultimately manifested by an instance of t2 conduct). It sounds like according to your proposal, what the attempt gives us is very good evidence about what it is that the defendant would have been willing to do at t2, had they known that their victim were still alive. A supposed corpse defendant’s supposed killing—what we, knowing all the facts, know to be merely an attempt—may often be good evidence of what a supposed corpse defendant is willing to do at t2, but it may not be. We might have really good evidence that a defendant genuinely is reformed, such that they would not have performed the act of disposal or mutilation had they known that their victim was alive. On the flipside, there might be very good evidence *besides* what it is that a defendant has done earlier to suggest that a defendant who does *not* attempt to kill at t1 would be willing to kill at t2. We could round the facts out of my hallucination case to make it that way. But then is your proposal distinguishable from a proposal recommending culpability for concomitant ignorance, and if so, how?

    Your point about murder being defined in some places as an intentional killing, is well taken. I’ll note that in jurisdictions like this, we could understand what’s been done as implicitly settling on a correspondence requirement that may not reflect all instances of satisfaction of the correspondence relation. That would be so if we believe that there are some causal connections between intention and conduct that do not establish intentionality but that nonetheless are just as apt at grounding culpability (here, I’m appealing back to my response to Alex’s initial post).

    I think a lot of this discussion has circled around the uncomfortable task of finding the line between what is up for grabs in my theorizing about correspondence and what we decide to hold fixed as taken for granted because a feature in contingent, actual, systems. This takes us back to Mark’s follow-up point about my bias toward the MPC. If we are thinking like philosophers and not just lawyers, and so, we are thinking in terms of which kinds of structures *should* (or, weakening this if we want to go with something more like a negative retributivism, maybe just: *could permissibly*) be recognized by legal systems assuming there are no institutional design values to wobble design away from a culpability tracking system, then why isn’t how one sets mens rea more up for grabs in my paper?

    The more I think about it, the more I’m beginning to appreciate that maybe I’m just very much on Neurath’s boat here, and the plank I’ve decided to stand on is the MPC. I have been building out a correspondence relation that “plays nice” with the mens rea scheme set out in the MPC. But that involves placing a kind of bet: that there’s a normatively significant relation there to be found, because the MPC has established a scheme that is itself normatively defensible. If it turned out the MPC was indefensible, then a correspondence relation built to take up its mens rea states as relata would not be particularly interesting as a component to grounding claims of culpability. To provide a really concrete example of where the rubber might hit the road on my bet, consider negligence. If negligence was not the kind of mens rea state apt to ground culpability, then no matter how clever we get relating negligence to an actus reus so that, as Ken Simons once put it, the negligence enjoys “significance in action,” we won’t end up with the kind of thing that can ground culpability, because we won’t have started with the right ingredients. It would be like looking for a relation that can ground culpability for a killing between the state of feeling sad and a killing. It just won’t work.

    To tack on a nod to your final paragraph: I’m just beginning to think about how supposed-corpse-style patterns might exist in other contexts. I think the daisy chain is generalizable to conduct crime settings, and I would think that your mens rea approach might also be.

    Thanks again so much, Gabe.

25. I want to reiterate my thanks to everyone who has participated in this discussion. I learned a lot, and there’s a lot left here to chew on. I’m honored each of you has taken the time.

    Our discussion has focused exclusively on criminal liability. That’s for good reason: the principle of correspondence is self-consciously a creature of the criminal law. But I’m going to go out on a little bit of a limb here as our 48-hour window is coming to a close and cap things off by actually suggesting a way that things might open up. I want to indicate how I think trained attention on correspondence could have interesting implications for other areas of law, too.

    Certain doctrines in other fields of law are what Issa Kohler-Hausmann has called “criminal-law-like,” in that legal determinations are made to depend on both mental and physical components. These aren’t exactly what we’ve been here discussing as mens rea states and actus reus instances of conduct, because they could be set to be different than what they are in the criminal law, but they are analogous. For example, the way that equal protection doctrine has developed, the discriminatory impact of a law (the physical component) alone is insufficient to generate a constitutionally cognizable claim under the Equal Protection Clause. Rather, as the Supreme Court held in Washington v. Davis, there must also be discriminatory intent (the mental component). As the Court put it, “the invidious quality of a law claimed to be racially discriminatory must ultimately be traced to a racially discriminatory purpose.” How is it, though, that discriminatory intent must be related to the passage of a discriminatory law?

    This is a question about a relation akin to correspondence. And I think it’s a very open question. I think legal scholars and courts have so far missed that there must be an implicit correspondence principle at work in the organization of the equal protection doctrine—at least, now that the doctrine is organized so as to include a mental and a physical component. There’s a large literature on the discriminatory intent doctrine, but it focuses mainly on what “intent” is in this context, and on what types of evidence can be used to show it. This leaves untouched the question of how “intent” needs to be related to the discriminatory state policy in order to count as relevant to that policy’s constitutionality.

    With only a ham-fisted understanding of what constitutes the relation-akin-to-correspondence in this context, the Equal Protection Clause might well go under-enforced. Say that a policy passed for the purpose of oppressing a racial minority group at t1 is challenged in court. The state actor repeals its old policy and passes a new one at t2 with the same impact. In the legislative record at t2, there is no evidence of discriminatory intent.

    If we were looking for contemporaneity, as I think courts implicitly do, we’d consider the t1 intent, t1 policy, and effect of the t1 policy to be irrelevant to the constitutionality of the t2 policy. The t2 policy would pass constitutional muster given that it is not accompanied at that time by discriminatory intent.

    One reaction to this kind of scenario has been to say that because some such “tainted” t2 policies *should* be deemed unconstitutional, we should dispense with an intent requirement in equal protection doctrine. (W. Kerrel Murray airs this reaction in a recent Harvard Law Review piece). But this reaction springs from an ill-advised assumption. The assumption is that a criminal-law-like approach to a legal determination must come with it a non-severable contemporaneity requirement.

    We could instead just reject that the relation-akin-to-correspondence is contemporaneity and recommend a more nuanced correspondence requirement. If we recognize that there can causal connections between a t1 intention and a t2 policy that are such that the t1 intent comes to count as relevant to the t2 policy’s constitutionality, despite being strictly earlier to it, we could deem t2 policies unconstitutional without having to push for abolishment of the Washington v. Davis intent rule.

    That’s not to say the intent rule should stay; I think there are probably *really* good reasons to get rid of it. The best version of equal protection doctrine is probably not criminal-law-like at all, because probably, the question of whether discrimination is unconstitutional shouldn’t depend on the culpability of the state actor doing the discrimination. But with attention trained on the relation that is something-like-correspondence, I think we can appreciate that we could account for “taint” cases without abolishing the intent rule.

    This is all just to say that thinking about correspondence might bear important fruit in other contexts. I am excited by the possibilities!