Plans and their Accomplishment:

A New Version of the Principle of Double Effect

Alexander R. Pruss

 

Department of Philosophy

Georgetown University

 

December 28, 2006

 

1. Double effect

           Suppose that one initiates a causal sequence leading to a basically evil state of affairs, but does not intend the evil effect, and the good effects of the action are proportionate to the bad.  A state of affairs is a “basic evil” provided it is evil in virtue of itself and not in virtue of its connection with other states of affairs.  The classic form of the Principle of Double Effect (PDE) can be taken to state that then the action that initiated the causal sequence is not wrong on account of the basically evil state of affairs it produces.  The action might still be wrong for some other reason, say by violating a promise, or causing some other basically evil state of affairs, or being a malum in se.  The PDE, however, provides the agent with a complete defense against the accusation that the action she did was wrong on account of the basically evil state of affairs produced. 

If I am a drug manufacturer and make a rubella vaccine knowing that a few patients will die of side-effects, I satisfy the PDE if the public health benefits are sufficiently large, and hence the side-effect deaths are not something I am culpable for.  However, the action may still be wrong on other grounds: I might be exploiting staff, breaking promises, or procuring tissue derived from illicitly aborted fetuses to culture the vaccine.

           Notice that in my formulation I did not use the classic phrase “either as an end or as a means” in saying that one does not intend the bad effect.   For that is not the business of the PDE properly understood, but of an account of intention: the PDE should limit itself simply to saying that the effect is not intended.  Now, I do not dispute the classic catchphrase that he who intends the end, intends the means.  Hence indeed for the PDE to apply, the basically evil effect must not be intended either as an end or as a means. 

However, it may also turn out that there are also other ways of intending than as means or as ends, and if so, then the PDE’s formulation will also require that the bad effect not be intended in such other ways.  To make plausible the possibility of other kinds of intending, suppose that I offer to pay you ten dollars for sending a nerve signal to your hand.  You think about it for a moment, and you clench your fist, knowing that the clenching of the fist will have resulted from a nerve signal from your brain to the hand.  But it is not right to say that the clenching of your hand is a means to the sending of the nerve signal, because the sending of the nerve signal is what causes the clenching of the hand.

2. Counterexamples

           There is a cottage industry of apparent counterexamples to the PDE, begun by Jonathan Bennett who argued that you can reformulate your intentions so as to avoid seeming to intend evils.  Thus, rather than intending to terrorize the enemy population by killing civilians, which is impermissible, you drop a bomb on the civilians in order to make them seem to be dead, which is not a basic evil.  That they die is foreseen, regrettable but unintended, it seems, just as it is foreseen, regrettable but unintended that some patients die from the rubella vaccine.  But such an action is intuitively wrong, and hence the PDE is false.

           Now there is a simple answer in this case: the chosen means for making the civilians seem to be dead is making them be dead, and hence the bomber does intend their deaths.  However, this will not answer more sophisticated cases.  An eccentric, literalistic but always truthful magnate tells you he will donate billions to famine relief, saving a million lives, if and only if you follow his directions to the iota.  You will purchase a gun, sneak at night into a zoo owned by the magnate, and kill the first mammal you see.  You follow his instructions to the letter, but unfortunately the first mammal you see is the zookeeper.  When you are charged with murder, you argue that you did not intend to kill the human there, but only to kill the mammal.  After all, it was irrelevant to you that the mammal was human—the only relevant fact was that it was the first mammal you saw, and it just happened to be human.

           In general, the problem of closeness is that for many actions that are intuitively made wrong by the fact that a basic evil is used as a means, it seems to suffice for making sense of the action to attribute to the agent an intention for something close to the evil but not itself basically evil. Instead of intending to cause death, you intend to cause apparent death.  Instead of intending to kill a human, you intend to kill the first mammal you see in the zoo. 

           But now there is a simple argument that such cases are not effective counterexamples to the PDE.  For what makes these examples so plausible is precisely because we all recognize the sophistry of saying “I intended to kill the mammal and not the human” or “I did not intend to kill them, but only to make them seem dead.”  The person who intended to kill the mammal while knowing the mammal to be human, it seems intended to kill a human.  And the person who bombed people to make them seem dead surely intended to kill them, unless she was under the false impression that she was dropping a knockout gas on them.  But if these observations are right, then we do not here have a good argument that there are counterexamples to the PDE as I have formulated it, since it seems the PDE simply does not apply: the basic evils are intended, and the actions hence are wrong.

           Many defenders of the PDE have said that tactical bombing can be justified when civilian casualties are not a means to the tactical goal.  This is supposed to be a paradigmatic example of the PDE.  But now note that the person who says that in bombing a populated city one cannot but intend civilian deaths also is not objecting to the PDE as I have formulated it.  She at most is objecting to the claim that one can intend to bomb a populated city without intending civilian deaths, and hence is objecting to an application of the PDE.

           To provide a counterexample to the PDE along one of the above lines, one would have construct a situation where an action is wrong on account of a basic evil that is produced and then positively argue that the basic evil in the situation is nonetheless unintended.  One could try to do that, but the difficulty in such a strategy is that intuitions to the effect that the basic evil is unintended at the very same time make more plausible the claim that the action was morally acceptable, while intuitions to the effect that the action was wrong, also make more plausible the claim that the basic evil was intended.  The challenge for the PDE’s opponent, then, is to argue simultaneously that the evil is unintended and that the action is wrong, without the argument for the one claim weakening one’s commitment to the other.  Simply stating counterexamples is not enough.

           Nonetheless, the challenge can be met if the PDE is left in its classical formulation.  Something is one’s intention provided that the hope of bringing it about contributes in the right way to explaining one’s action.  Take once again the case of killing the first mammal one sees in the zoo.  I go in, and shoot the first mammal in the heart.  The action as described is fully explained by my intention to kill the first mammal I see and my belief that shooting it in the heart is an efficient way to kill it.  Now, let us suppose that I also know that the mammal was human.  However, as far as the story goes, this additional fact about my knowledge would not contribute to the explanation of my action: I want to kill the mammal because it is human, but because it is a mammal;  I shoot it in the heart not because it is human, but because it is a mammal, and all mammals can be killed by being shot in the heart.  Had I abstained from killing the zookeeper because he is human, or killed him out of a personal vendetta, his humanity might have been explanatorily relevant.  But as it is, it was not.  This is a serious problem for the classical formulation of the PDE.

           Now, one might try to say that on Aristotelian grounds, being human is prior to being a mammal, and so one is only killing a mammal because one is killing a human.  But that does not seem relevant to the question about intentions.  After all, if one did not know that the mammal was human, one could kill it without intending to kill a human.  But knowledge only changes one’s intentions when that knowledge is explanatorily integrated into the action.  But if the person shooting the zookeeper does not at all care whether the mammal is human or not, then the knowledge that the mammal is human is explanatorily irrelevant, and hence the intentions are the same whether the shooter knows the zookeeper to be a mammal or not.

3. Identity, intention and accomplishment

           There is an obvious intuitive thing to say about the case of killing the mammal.  The intended death of the mammal is identical with the death of the human.  However, “intends” creates an intensional (with an s) context, and indeed must do so in order to figure as it does in the explanations of our actions.  After all, one might have intentions (with a t) that completely fail to match the world.  Marcus might intend to kill Cicero in order to benefit Tully by reducing the competition Tully was facing as a writer, and Marcus’s action of stabbing Cicero only makes sense when the intention is described in this way: it would make no sense to kill Tully to reduce the competition Tully is facing.  One might try talking of intention de re, but that would be an odd concept indeed.

           The mammal case shows we need to describe the objects of action in the PDE in a more extensional way than intention language allows.  There is also an independent, albeit vague, objection to be made.  The PDE as it stands is too interiorized.  What is right or wrong is in the first instance an action, something coming from the will but typically causally reaching out to outside the will.  Of course it is only an action, and indeed the kind of action that it is, when it is intentional.  However, the danger of focusing on intentions is that it is apt to mislocate the source of wrongness.  The wrongness of murder comes from the harm that the murderer does to the victim.  Now, granted, an attempted murder is also wrong, but it is wrong precisely because it is an attempt at doing what is wrong.  The moral character of an attempt depends on the moral character of a success.  It is an important lesson in moral philosophy that one’s theorizing should be based as much as possible on the normal cases.  And the normal case is of a successful action that is right or wrong on account of what has been accomplished, and an agent who is fully responsible for this action.

           Another way to put this is to observe that what is in the murderer’s will, is the concept of death, but what happens to the victim is a particular, concrete death.  And it is the particular death, not the concept of death, that is the evil here.  Having the concept of someone death as a goal in one’s will is bad precisely because it would be bad for that person to die, and it is bad to have one’s will attempt to accomplish a bad.  So it is that which is accomplished which should be at the forefront of moral investigation.

           Now this reasoning may seem inimical to Double Effect.  However, “accomplishment” depends on intention.  When we act, we have an action plan which includes the end or ends for which we act as well as other things to be accomplished for the sake of the end or ends, all given in more or less detail.  The action plan need not be explicit, but it determines whether what has actually happened counts as a fully successful action or not.  If I intend to nourish myself by a bologna sandwich and accidentally eat a salami sandwich, I have achieved my end, but I have not accomplished a fully successful action, because a part of the plan has not been fulfilled.  The fulfillment of the end is an accident rather than an accomplishment, because I have failed to accomplish a part of the plan.  Whether something in the world counts as my accomplishment in a particular action depends on what my plan was, i.e., on whether the thing was a part of the actualization of the plan.

Further effects beyond the ends of the action are irrelevant to the success or failure of the action, and hence are not accomplishments.  For the occurrence of such a further effect does not help explain why one counted as successful in acting.  These effects do not hold failure at bay.

           I will use “accomplish” and “accomplishment” extensionally.  What one accomplishes is an accomplishment, and if x and y are identical, and x is an accomplishment, so is y.  Thus, when Marcus killed Cicero in order to benefit Tully, what he actually accomplished was also the death of Tully.  Marcus’ plan was accomplished only in part: Cicero was killed, as planned, but Tully did not benefit.  Nor are the effects, even foreseen ones, that are beyond the ends in the plan an accomplishment.  Thus, Cicero’s death led to some worms being nourished.  But the feeding of worms is not an accomplishment of Marcus’.  Were the worms to have disliked Cicero, the degree to which Marcus’ plan was accomplished would have been no less. 

           What is accomplished, being extensional, can as it were fill in the details in one’s action plan.  When a CEO says “Do whatever it takes to win this contract” and means it literally, and the subordinate bombs the competitor’s offices and thereby gains the contract, the bombing is something accomplished by the CEO, even if she did not expect that to be the means used.  The CEO’s action plan was that she would give an order which would initiate a process leading to the winning of the contract.  The intention involved a universal, a process, and the accomplishment of that intention involved a particular, the process.  This particular process was in fact a bombing.  And so the action plan was accomplished by the bombing.  The CEO’s excuse that she did not expect a bombing to happen would not remove her responsibility.  Granted, if she had a very poor imagination, her culpability may be in doubt, but nonetheless she was responsible, and the bombing was something she accomplished, by means of the subordinate.  In fact, in general, what one intends to produce at least tends to be a universal, and what one accomplishes is always a particular.

           It is worth coordinating the notions of accomplishing, intentionally bringing about, action plan and intending.  One proposal that appears faithful to much ordinary language is to say that you intentionally bring A about provided that you accomplish A while believing that A might be a part of your accomplishment of the action plan.  Thus, saying that someone intentionally brought something about is a conjunction of a claim about what the person intended and a claim about what the person believed.  What is accomplished is always a partial or complete fulfillment of the action plan, and in successful cases the accomplishment is at least a part of the truthmaker of the claim that this action plan succeeded.  Accomplishment and intentional bringing about depend on the world.  The content of the action plan does not—it is entirely a matter of will.  Likewise, intending does not depend on the mind-independent aspects of the world.  Someone who is a brain in a vat can intend all kinds of things that are completely out of touch with what she could produce in the world.  What one intended is not affected by whether and how one succeeds.  But what one accomplishes obviously is.

One intends something provided it is a part of one’s action plan.  The CEO who intends that the contract be got by any means necessary did not intend the bombing of the competitor’s offices, since a bombing was not a part of the action plan.  However, a bombing was in fact a part of the fulfillment or accomplishment of the plan.  Now one might think that in the case where the CEO knows that bombing will be the means chosen by the subordinate, the bombing will in fact be intended.  However, this is mistaken.  Only what is explanatorily relevant is a part of the action, and simply adding knowledge to pre-existing intentions need not modify the intentions, because it need not modify the explanation of anything one does.  The agent might simply not care about the additional knowledge.  To the ruthless CEO, learning that the subordinate will use a bombing might be just as irrelevant as learning that Fermat’s Last Theorem is true.  Thus the bombing is not intended even when the CEO knows it will be the subordinate’s means of choice.  This, of course, means that the PDE will have to be modified from its classical form.

4. Double effect (again)

           Let us, then, take seriously the intuition that we should focus on the case of successful action and that moral evaluation should begin with what is done, and let us formulate the PDE in terms of successful action.  Here, then, is a formulation of the PDE that takes these lessons to heart:

(PDE2) An action is intrinsically wrong on account of resulting in a basic evil e if and only if e is accomplished in the action.

Classical formulations talk about proportionality.  This is unnecessary here.  For the question is whether an action is intrinsically wrong, is the sort of action that it is never permissible to do on deontological grounds.  Proportionality is a question of consequences and circumstances, costs and benefits.  If proportionality is lacking in any action, the action will be wrong, but it will not thereby be intrinsically wrong.  Moreover, note that we have strengthened the initial PDE which was a permissive principle into now giving a necessary and sufficient condition for an action to be intrinsically wrong on account of resulting in a basic evil.

           In the case of a successful action, everything intended is also accomplished.  Thus talking about accomplishment instead of intention in the PDE will yield a more restrictive principle in the case of successful actions.  Conversely, everything accomplished was a part of what one intended under some description or other.  Hence we do not depart far from the classical forms.  (We could even extend the original to get this if we made intentions be extensional, but that is deeply implausible.)

5. Action failure and culpability

           The intention formulations of the PDE had the advantage that they applied directly in cases of failed actions.  Thus, when Fred intended to kill Bob but failed, the wrongness was seen as inhering in the intention that Bob be dead.  However, on the accomplishment account, no basic evil (other than perhaps the action itself, and whether that is evil is the question at hand) was accomplished.  We need, then, an additional moral principle for such cases of action failure. 

Fortunately, one is not hard to find.  It is wrong to try to do wrong.  This principle is ambiguous between two cases.  One is where one is trying to do something that one believes, correctly or not, to be wrong.  The other is where one is trying to do something that is wrong, whether or not one believes it.  The principle is correct, however, in both cases:

(Try1) It is wrong to try to do something accomplishing which one believes would be wrong.

(Try2) It is wrong to try to do something accomplishing which would be wrong.

It would be elegant if one could subsume the two cases under a single moral principle.  I do not at present know how to do that, however, except by using ambiguous scope as in my initial formulation “It is wrong to try to do wrong.” 

           This account yields a plausible order of explanation: The attempt is wrong because the success would be wrong.  Insofar as talking of intention is talking of an attempt, the classic formulation of the PDE obscures the order of explanation, and is apt to make one think that what makes it wrong to succeed in murdering someone is that it was wrong to have tried.  The modified PDE together with Try1 and Try2 leaves open the question whether to attempt murder is just as wrong as to commit it, and as such does not foreclose the question whether having the moral luck of failing at the murder reduces the wrongness of the deed, or merely reduces the harm of the deed.  The account is compatible with either answer to this question.

           The question of culpability, however, falls apart from the question of wrongness.  Suppose that Fred tries to kill the noisy mammal next door, falsely believing it to be a rat.  He proceeds by introducing chlorine gas under the door, thereby killing his neighbor Jennifer, who was in fact the noisy mammal.  Fred has, perhaps, accomplished the death of the noisy mammal[1], and this death is identical with Jennifer’s death, and so Fred has accomplished Jennifer’s death.  Nonetheless, Fred did not believe that the death of a person would be something he would accomplish, and so he did not believe himself to be violating the modified PDE.  Thus, at least if the ignorance was not culpable, he is not culpable for killing Jennifer (he may be culpable for reckless endangerment of human life, of course).  Nonetheless, the modified PDE lets us say that what he did was in fact wrong.  This explains why it is that Fred is the worse off as moral agent for having done the action.

6. Event individuation

           In his critique of the PDE, Jonathan Bennett considered an account on which one is prohibited from bringing about an event that is identical with a prohibited event.  His grounds for dismissing this was a dilemma: event individuation is either coarse-grained or fine-grained.  On a coarse-grained analysis of event individuation, turning a light switch, causing an explosion, causing a noise, etc., may all be the same event.  On a fine-grained analysis, there are at least as many events as non-synonymous descriptions of events. 

A fine-grained analysis will not help double effect, it seems, because it would make the PDE too permissive.  For instance, a terror bomber could be claimed not to be intending the deaths of the civilians, but the deaths of the civilians as a means to ending the war, which is a non-synonymous description, and hence a different event.  But clearly the terror bomber is not acting rightly.

On the other hand, a coarse-grained analysis will not permit the paradigmatic case of the tactical bomber who bombs an enemy military facility in order to bring about an end to the war, despite knowing that civilians in the vicinity will die.  For then the event of the bombs exploding is identical with the event of the bombs killing the civilians.  Since the bomber intends the bombs to explode, the killing of the civilians is likewise intended.  But, it is assumed, the tactical bomber is acting rightly.

The same problem would seem to infect the present account.  If we count accomplishments in a fine-grained way, then we will allow terror bombing since the deaths of the civilians will not be accomplished, but only the deaths of the civilians as a means to peace, while if we count them in a coarse-grained way, then we will disallow tactical bombing, since in accomplishing the exploding of the bombs one accomplishes the identical event of the killing of the civilians.

But let us look at this more closely.  I will use the coarse-grained approach.  It is plainly incorrect to say that the exploding of the bombs is identical with the deaths of the civilians, since the exploding causes the deaths.  Rather the exploding of the bombs is identical with the cause of the deaths of the civilians.  Thus, it follows on the proposed account that the tactical bomber accomplishes the cause of the deaths of the civilians.  But this is a distinct claim from saying that he accomplishes the deaths of the civilians.  An effect, after all, is always distinct from its cause.[2]  This is true even in cases where the cause is described in such a way as to entail the effect.

To defend tactical bombing, it has to be claimed that the cause of the deaths of the civilians need not be a basic evil.  But this is clear.  A basic evil is evil in and of itself.  Now some causes of death will be basic evils: for instance, torture or decapitation.  However, just about any event can serve as the cause of death under appropriate circumstances—one might die from drinking a cup of water or from writing a letter of recommendation (the cases are easy to flesh out).  The cause of death as such is bad because of the death that it causes, and hence it is not a basic evil, but an instrumental evil.  The mistake in the Bennett line of reasoning was to confuse instrumental with basic evils. 

Now one might oppose this line of thought as follows.  The tactical bomber accomplishes the cause of the civilians’ deaths.  But the deaths of the civilians are a logical consequence of the occurrence of the cause of the civilian deaths.  Surely if one accomplishes something, one accomplishes all its logical consequences.  Hence, the deaths of the civilians are an accomplishment of the tactical bomber.  But the logical consequence principle is simply false.  I have never accomplished that 2+2=4, even though it is logically entailed by every report of every action of mine.  One might try to weaken the principle by restricting it to more relevant entailments.  But even that will not do.  Suppose I accomplish the event that Sarah in light of her knowledge that p wishes for.  It certainly does not follow that the truth of p was accomplished by me, even though that p is true is intuitively relevantly entailed by the fact that some event that happened was wished for by Sarah in light of her knowledge that p.

A variant objection would be that one cannot accomplish the cause of A as such without thereby accomplishing A.  But this confuses the extensional sense of “accomplish” that I have introduced with an intensional (with an s) sense that is more closely aligned to intentions (with a t).  The confusion is clear from the phrase “as such” which is an interloper in extensional contexts like this one.  To accomplish the cause of A as such is indeed to accomplish A as caused by that cause.  But that is not the right way to describe the accomplishment.  If the cause is C, an explosion, say, then it is C that is accomplished, not C-qua-cause-of-A, since “qua” and “as such” do not belong in extensional contexts.

Thus, that the tactical bomber accomplishes the cause of the civilians’ deaths does not imply that he acted wrongly.  But what about the fact that he did not just accomplish the cause of the civilians’ deaths, but the killing of the civilians?  Here a distinction needs to be made.  The killing of the civilians can be thought of in two ways: either as an event that causes the deaths or as an event that includes the deaths.  If the killing includes the deaths, the event is no longer identical with the exploding of the bombs on anybody’s account of event individuation, since the bodily states of the civilians are not a part of the exploding of the bombs.  Hence it no longer follows that the bomber has accomplished the killing of the civilians.  If, on the other hand, the killing is seen as the cause of death, then the previous argument applies.  The killing of the civilians can be accomplished without the deaths of the civilians being accomplished.

7. Closeness and danger

           The account given so far handles the case of killing the mammal.  However, what about other cases of closeness?  There is a class of putative counterexamples to Double Effect where the agent intends the proximate cause of a basic evil, apparently without intending the basic evil itself.  These cases are most persuasive when the agent needs to intend that proximate cause under a description that makes it clear that it is in fact the proximate cause of a basic evil. 

For instance, consider the case of throwing a large bystander in the path of a runaway trolley, reasoning that the bystander’s body will absorb the trolley’s kinetic energy, and save the lives of several people down the path.  Intuitively, the action is wrong.  The action plan of an agent who does this must contain the fact that a trolley with high kinetic energy impacts the body of the bystander.  However, Double Effect seems to justify this action.  This seems to be true even on the present accomplishment account.  For suppose that the plan is successful.  Then, what has been accomplished is that the body of the bystander has absorbed the kinetic energy of the trolley.  This absorption of kinetic energy is the cause of death, and hence is not identical with the death itself.

           However, the phrase “Double Effect justifies this action” is incorrect.  For what Double Effect licenses is not the conclusion that the action is right, but that it is not wrong on account of the basic evil produced, in the trolley case the death of the bystander.  The action in these closeness cases may nonetheless be wrong on other grounds.  I would like to suggest one such ground, specifically in the cases where the basic evil is the death of an innocent bystander.  In these cases, among the accomplishments of the agent there is the proximate cause of death.  The proximate cause of death is not a basic evil, as we have seen.  However, the proximate cause of death always is a grave danger to life.  Thus in these closeness cases, a grave danger to the life of an innocent bystander is accomplished.  The bystander is endangered, and this endangering is accomplished by the agent.

           Now, gravely endangering a person is not intrinsically wrong.  It is acceptable to throw oneself on an exploding grenade to save one’s comrades, even though in doing so one intends to absorb kinetic energy, an absorption that is a grave danger to one’s life, even though it is wrong to accomplish one’s own death.  However, typically, one has acted wrongly when one has accomplished a non-consenting innocent party’s being grave danger, say because one used an enemy civilian as a human shield.  The “typically” is needed here, because we are not dealing with the accomplishment of a basic evil, and so there may be additional cases where the action is permissible, e.g., ones where consent can be presumed.

           While we cannot apply Double Effect when basic evils are not at issue, nonetheless reasoning similar to that behind Double Effect is applicable here.  Generally speaking, one has acted wrongly if it is an accomplishment of one’s action that a non-consenting innocent is in danger.  However, if the danger is merely caused by what one has accomplished, without one’s accomplishments being innately dangerous to the innocent party, then one’s action will be immune to criticism on account of the endangering if proportionality holds.

           The distinction that considerations of accomplishment yield is relevant.  Consider, for instance, the difference between redirecting the trolley away from five people onto a path where there is only one, and throwing the one into the trolley’s path.  In the case of throwing, the absorption of kinetic energy by the one person is the grave danger that she is under, and it is accomplished by one’s action.  In the case of redirecting, all that is accomplished is that the trolley takes a particular path.  This is not an innately dangerous situation.  It is only dangerous when we add the fact that there is an innocent person in the way, which is not a part of one’s accomplishment.  Hence in the redirection case, one has accomplished not the danger itself, but a cause of the danger.

It may be—and this suggestion is due to Mark Murphy—that a number of cases that are seen as problems for Double Effect are actually questions not about Double Effect but about justice and authority.  It is morally praiseworthy to jump in front of a trolley or on a grenade, but not to toss an innocent bystander.  The present approach has the advantage, then, of focusing the discussion exactly where it should be. 

8. Metaphysics and Conclusions

On the accomplishment account, applications of Double Effect are tied to metaphysics.  One needs to figure out what accomplishments are in fact identical with what.  For instance, suppose that we have a metaphysics on which persons are not animals, and the person dies because the associated human animal dies.  Since the person is distinct from the animal, the death of the person is distinct from the death of the animal.  On such a view, it makes some sense to suppose that the death of the animal might be an accomplishment without the death of the person being one.  Resistance to such a conclusion is, I suspect, mainly due to the fact that we do not really accept such a metaphysics.  In this case the moral issue is easier to solve than the metaphysical problem: one is clearly a murderer even if one is only trying to kill the human animal.  When one accepts PDE2, this yields an argument against the metaphysical view on which human persons are distinct from animals and human persons die because their associated animals die.

In other cases, the metaphysics may be more tractable than the ethics.  The deaths of civilians are distinct from the explosions of bombs, and the explosions come first, and so in accomplishing the explosions of bombs, one has not necessarily accomplished the deaths of civilians. 

It might be surprising to some that moral questions should depend so much on metaphysics.  But surely what is right and wrong depends on how things really are.

There is one crucial question for future investigation that I have skirted here, and this is the ontology and individuation of accomplishments.  Certainly events can be accomplishments, but events do not exhaust the class of accomplishments, at least if something like “a lasting peace” is not an event.  Accomplishments may include tokens of relations, too.  Thus, in the example of setting the bomb-sights by the civilians, the token nearness between the explosion and the civilians is an accomplishment.  We can say confidently that accomplishments are not individuated as finely as propositions, since then Double Effect would justify almost everything.  Moreover, while “that” clauses can be used in some reports of accomplishments, nonetheless I suspect that not all accomplishment reports can be reduced to sentences of the form “x accomplished that p”.  The details here are difficult to fill in.  But that should be no surprise, given the difficulty of the moral issues.



[1] The “perhaps” is because Fred might have intended to kill the noisy mammal by killing a rat. Insofar as he didn’t kill the rat, the death of the noisy mammal was not accomplished by him.  However, we may imagine that Fred believed it to be a rat but did not care that it was a rat.  In that case, he is guilty of not caring what he is killing, but not guilty of murder.

[2] People like Descartes who dispute this in general should not dispute this in the case at hand.