The Doctrine of Double Effect:

Its Relevance to Collateral Damage Cases


CPT Chris Mayer

Instructor, U.S. Military Academy

Phone: 845-938-3368

Fax: 845-446-2562




Political and military leaders, knowingly and unknowingly, use the Doctrine of Double Effect  (DDE) to justify military actions that cause collateral damage.[1]  There are, however, critics of the DDE who claim that it does not provide a morally relevant distinction in many situations, to include cases involving collateral damage.  Much of this criticism is focused on the abuse of the DDE and its alleged incoherence.  However, it is possible to answer most of this criticism by identifying the moral principles that serve as the DDE’s foundation and by recognizing the problems caused by proximate means and the closeness of the intended means of an act and its foreseen but unintended side effects.  In this paper I will attempt to develop a version of the DDE that is true to its underlying moral principles and avoids the problems associated with proximate means and closeness.  Consequently, I will argue that this version of the DDE reliably provides an acceptable moral distinction in many cases in which military action causes collateral damage.[2] 

1.  Methodology. 

            I will examine various collateral damage cases to determine if the DDE’s judgment concurs with our intuitions and is consistent with the moral principles that serve as the DDE’s foundation.  This will make it possible to determine if the DDE provides a morally relevant distinction for these types of cases.  By relying on intuition to evaluate the DDE’s assessment of particular cases, it will be possible to make a prima facie determination as to whether the DDE makes the correct judgment in a given case.  While intuition cannot serve as the only assessment of the DDE’s judgment, it does provide a good initial evaluation.  I will also determine whether the DDE’s judgment about collateral damage cases is consistent with the moral principles that serve as the DDE’s foundation.

If it can be successfully argued that the DDE’s judgments fits with our intuitions and adheres to its moral principles in many collateral damage cases, it will then be possible to accept the DDE’s judgment in these types of cases.  In cases where the DDE provided distinction does not provide a morally relevant distinction, I will explain what is happening and what can be done when faced with these types of cases.[3]  However, before an alternate version of the DDE can be developed and applied to collateral damage cases, it is necessary to present the standard formulation of the DDE. 

2.  The Doctrine of Double Effect. 

2.1 The Standard Conception of the Doctrine of Double Effect. 

            The DDE has practical applications in that it provides a set of conditions that can be applied to certain types of cases to determine if an action is morally permissible.  However, it is important to recognize that the purpose of the DDE is to ensure that certain moral principles are adhered to.  That is, the DDE is simply a heuristic method that can be used to determine whether a particular course of action adheres to a set of moral principles.  If an action meets the requirements established by the DDE’s foundational principles, then the action is morally permissible.  Consequently, if by applying the DDE to a case we determine that an action is morally permissible, yet this action actually violates the DDE’s foundational principles, then the DDE has failed to serve its purpose.  Therefore, in cases where the DDE does not provide clear guidance, it may be necessary to appeal directly to the foundational principles.  Evaluating a case in light of these principles will allow us to determine whether the DDE has made the correct judgment and is serving its purpose. 

The first moral principle that serves as the DDE’s foundation is that we should not intentionally harm innocent people.  This action serves as a constraint on our actions towards other people.  It is necessary to specify innocent people because there are instances, such as self-defense, where it is morally permissible to intentionally inflict harm. [4]   The second principle, which is related to the first, is that people possess a right to self-defense.  That is, people have a right to protect themselves against those that seek to harm them.[5]  The final principle is that a person’s intentional actions should be able to be characterized as good.  People must not aim at evil ends or means, and the bad that they cause unintentionally must not outweigh the good produced by their action.  Adhering to this principle does not necessarily limit us to a consequentialist perspective because we can, as Quinn proposes, assume “that goods of different kinds (for example, preservation of life and relief from suffering) can be compared and at least roughly summed up, and that in cases of conflicting rights we can make at least a rough comparison of the overall good protected by the rights on each side of the conflict” (Quinn 161).  When determining whether an action can be characterized as good, we can look at the types of rights it violates or adheres to, and this goes beyond a simple utilitarian calculation.  So, the DDE’s foundational principles create a duty not to harm innocent people, a right to self-defense, and a duty to ensure that one’s actions are good.[6]   These duties and rights serve as the basis for the DDE’s guidelines.

The DDE has traditionally employed four conditions that must be met before an action can be considered morally permissible.  These four conditions help determine whether an action adheres to or violates the DDE’s foundational principles.  The first is that “the intended final end must be good”.  The agent’s goal must be good; his final end cannot be bad.  Secondly, the “intended means to it (the final end) must be morally acceptable”.  Immoral means cannot be used to achieve a morally good end.  Third, “the foreseen bad upshot must not in itself be willed (that is, must not be, in some sense, intended)”.  If the act has a morally bad side effect, this side effect must not be intended as a means or an end.  Finally, “the good end must be proportionate to the bad upshot (that is, must be important enough to justify the bad upshot)” (Quinn 189).  The evil brought about by the morally bad foreseen but unintended side effect should be proportionately less than, or equal to, the good accomplished by the act. 

By examining the four conditions mentioned above we can see that the DDE, as it relates to harm, makes the claim that it is sometimes morally permissible to allow a harm to occur as a foreseen but intended side effect, while it is not morally permissible to bring about the same harm as either an intended end or as a means to achieve an intended end.  The DDE places restrictions on what the intended ends and the necessary means of our actions may be, however, it does permit us to cause evil by our actions as long as we do not intend the evil as an end or a means.  By applying the four traditional conditions of the DDE to a possible course of action, we can usually determine whether or not the action is permissible.  However, applying the DDE is often difficult, and it is the problems associated with its application that I will focus on in the next section. 

2.2 Problems Associated with the DDE and Potential Solutions. 

Critics charge that many potential problems can arise when using the DDE to evaluate actions involving harm.  One such problem is the confusion as to what actually counts as a means to an end.  In certain types of situations it may be unclear as to whether harm was used to achieve the intended end.  We can resolve this problem by closely examining the means used so that we can identify when harm is simply a more proximate means by which the intended means are achieved.  Harm must not bring about the desired end.  Being able to identify proximate means will reduce the possibility that the DDE will judge an action to be permissible when our intuitions, and the principles of the DDE, indicate that the action is not permissible.

Imagine a case where Jones steals Smith’s wallet, and Smith fires a gun directly at Jones with the intended end of getting his wallet back.[7]  Smith may claim that he did not need to kill or harm Jones, because if Jones dropped the wallet Smith would no longer attempt to shoot him.  Therefore, Smith would claim that his intended means of getting his wallet back was to stop Jones, or at least force Jones to drop the wallet.  And, because his intended end was to get his wallet back from Jones, and not to intentionally harm Jones as a means or end, Smith might claim that his actions were morally permissible under the guidelines established by the DDE.  Yet it is clear that Smith is intending to harm Jones by shooting him.  Smith’s actions are not morally permissible because the proximate means of stopping Jones, or causing him to drop the wallet, involves killing or seriously harming him.  And the means of killing or seriously harming Jones is shooting him.  Smith is stopping Jones by shooting and harming him, and this makes harm a proximate means.  This violates the constraint of not intentionally harming an innocent person.[8]  Therefore, to reduce the problems associated with proximate means, and to ensure that the DDE provides an acceptable moral distinction when harmful proximate means may be involved, the means used to achieve the intended end must be thoroughly examined.  When harm is a proximate means, the action is not morally permissible.   

Another problem associated with the DDE is that of closeness between the intended means and the foreseen but unintended side effects.  Richard Norman argues, “The boundary between intended actions and unintended side-effects is not a sharp one” (Norman 89).  It is often difficult to determine if the intended means and the foreseen but unintended side effects are separate.  One example that illustrates the problem of closeness is the Shooting Range case.  You are at a range shooting at targets.  After a short break you return to your firing position and take aim at your target, however, you notice that someone has tied a child to your target.   Despite the presence of the child, you fire your weapon because your intended end is to hit the target and your mean of doing this is to fire your weapon.  You claim that harming the child is a foreseen side effect because you do not intend to harm the child.  Therefore, you believe that the DDE’s guidelines allow you to fire at your target.  However, firing at the target is too close to harming the child; the intended means are too close to the foreseen side effect.  To avoid the problems associated with closeness, one must ensure that the intended means are truly separate from the foreseen but unintended side effects.  This ensures that the harm is not intentional.[9]  If the means used to achieve the end are too closely related to the harm, then it cannot be claimed that the harm is a foreseen side effect.  This makes the action impermissible.  However, not every case where closeness is an issue will be as clear-cut as the Shooting Range case.

When the four conditions of the DDE do not provide a clear judgment for a particular case due to proximate means or closeness, it is necessary to appeal to the DDE’s foundational principles, and the duties and rights produced by these principles.  The DDE’s requirement that our intentional actions should be able to be characterized as good is a perfect duty. Thus, our intentional actions should always be good.  But, there are situations when the other two principles (the duty not to harm innocent people and the right to self-defense) conflict making it impossible to adhere to both.  In most cases the duty not to intentionally harm innocent people has precedence over the right to self-defense.  However, there are circumstances that would allow an agent to exercise his right to self-defense, even when this caused him to intentionally harm an innocent person.  That is, the duty not to intentionally harm innocent people is not absolute, and there is a threshold at which the right to self-defense takes precedence.

Thresholds recognize that “if enough good is at stake, then the constraint is no longer in force, and it is permissible to harm the person” (Kagan 80).  Therefore, it may become necessary to harm an innocent person in order to exercise a right to self-defense and meet the requirement that intentional actions can be characterized as good.  For example, it seems permissible to cut off a person’s foot when she is caught on a railroad track and a train is coming towards her.  We are intending to save her life, however, we are also intentionally harming her by cutting off her foot which seems contrary to the DDE’s guidance.  However, a threshold has been crossed and the good of saving the person, and the value of her life, outweighs the harm caused by cutting off her foot.  We are helping the person exercise her right to self-defense by intentionally harming her.  Additionally, in certain situations adhering to the duty not to intentionally harm innocent people makes the right to self-defense inconceivable and makes it difficult to characterize the resulting action as good.  In these situations it is permissible for a person to harm innocent people in order to defend himself.[10]  Although we cannot intentionally harm innocent people as an end in itself, there will be situations where one can harm innocent people as a means to exercise the right to self-defense. 

The concept of a threshold, as I described above, is often confused with the involvement of a third party.  That is, many think that the third party involvement is doing the work when in fact what has happened is that a threshold has been crossed.  For example, in the Shooting Range case we would not shoot our weapon for fear of harming the child.  We would not want to intentionally harm the child even if someone else is responsible for putting him there.  However, if a terrorist held a hostage, we would consider shooting through the hostage if killing the terrorist was deemed truly necessary.  In this situation it is clear that we are intentionally aiming at the hostage, even if we do not require her death to achieve our end.  Our means of killing the terrorist is shooting through and harming the hostage.  In this case the third agent involvement makes only a minor difference, but when combined with a belief in the existence of thresholds it is possible to resolve the difficulty.  If not shooting the terrorist meant that the terrorist could escape and set off a nuclear device, we would definitely shoot the terrorist, even though it causes the death of the hostage (who is an innocent person). 

In both cases (Shooting Range and Terrorist) a third agent is responsible for placing an innocent person in harm’s way.  However, in the Shooting Range case we do not shoot at the target because the good of hitting the target does not outweigh the harm caused to the child (even though we may be on a time limit or desperately need the practice) and our right to self-defense is not at stake.  The fact that the third agent is responsible for placing the child in front of the target does not, by itself, make it permissible to shoot through the child to hit the target.  In the Terrorist case (where the terrorist’s escape means that he will set off a nuclear device), it may be permissible to shoot through the hostage to kill the terrorist.   The difference between these cases is that a threshold has been crossed in the Terrorist case; the right to self-defense is inconceivable in this situation and is being violated in a fundamental manner.  It is impossible to imagine possessing a right to self-defense when a nuclear weapon is detonated.  Additionally, the rights of a large number of people are at stake.  These people are completely defenseless against someone who intends to inflict serious harm on them unless the hostage is harmed.  Also, allowing the terrorist to destroy the city cannot be characterized as good.  Therefore, there will be times when it is permissible for people to exercise their right to self-defense even when the means of doing so involve intentionally harming an innocent person.  While the prohibition against intentionally harming an innocent person as an end is absolute, there is only a prima facie prohibition against intentionally harming an innocent person as a means.  It would be permissible, in certain types of cases, to intentionally harm an innocent person as a means of protecting the right to self-defense.   These types  of cases will typically arise only in war. 

2.3 Formulating the DDE for Collateral Damage Cases.

In war, the DDE’s constraint against intentionally harming innocent people is transformed into a constraint against intentionally harming noncombatants.[11]  However, it is virtually impossible to fight a modern war without unintentionally harming noncombatants.  If military forces were restricted from performing actions that unintentionally harmed noncombatants, most acts of war would be considered morally impermissible.  Francisco De Vitoria, a just war theorist, has written, “sometimes it is right to slay the innocent, even knowingly, as when a fortress is stormed in a just war, although it is known that there are a number of innocent people in it” (Christopher 56).  Just as in storming a fortress, it is often permissible in war to unintentionally cause noncombatant deaths.  The very nature of war makes it likely that noncombatants will be knowingly, but unintentionally, harmed. 

The DDE establishes the limits of permissible violence in war, and yet it also provides military forces with flexibility as they fight in war.  The principle that promotes this flexibility is that nations have a “reasonable right to self defense against unjust aggression”, and it is this principle that justifies the existence of a threshold (FitzPatrick 10).  This principle transforms the individual right to self-defense into a principle of self-defense applicable to states within the international arena.  It ensures that nations are able to effectively defend themselves against aggression. 

Taking into consideration the problems discussed in section 2.2, and the adjustments that must be made for military situations, I will attempt to formulate a version of the DDE that differentiates moral from immoral actions in war (specifically collateral damage cases) and promotes the DDE’s foundational principles.  The principles underlying this formulation are: the absolute prohibition against intentionally harming noncombatants as an end and the prima facie prohibition against intentionally harming noncombatants as means; the right of nations and individuals to defend themselves against aggression; and the requirement that actions in war should be able to be characterized as good.[12]   

The first condition of the reformulated DDE is that the intended end of an action should not be the intentional harming of a noncombatant.  It is always prohibited to intentionally harm a noncombatant as the end of one’s actions.  Secondly, harm to noncombatants should not serve as the means, or proximate means, to achieving a good end unless a threshold has been crossed.[13]  The presence of a threshold preserves the rights of self-defense for nations and the individual.  It also avoids the indefensible argument of those who claim that when they shoot through the hostage to kill the terrorist they did not intend to harm the hostage.  We can now say that the harm to the hostage was the means of killing the terrorist and is permissible only if a threshold has been crossed. 

The third condition is that the bad foreseen side-effects of an action must not be intended.  If the side-effects are too close to the means, the only way for the action to be permissible is if the threshold has been crossed.  The final condition is that the good caused by achieving the intended end should equal or outweigh the harm caused by the foreseen but unintended side effects.  On balance the action should bring about good, or at least not bring about morally bad effects.  For the actions of military forces to be characterized as good, military forces must pursue victory through the destruction of military targets while attempting to minimize collateral damage.[14]  It is this version of the DDE that I will use to analyze collateral damage cases.             

3.  Applying the DDE to Collateral Damage Cases. 

            By applying the revised version of the DDE to collateral damage cases, it is possible to determine if the DDE makes morally relevant distinctions that fit our moral intuitions and are true to the  DDE’s  foundational principles.  I will begin with collateral damage cases in which a moral distinction is easily made and then evaluate cases where making a moral distinction becomes increasingly difficult.

Collateral damage often occurs when military targets are attacked during war.  For example, imagine a situation (Strategic Bombing) where launching an air attack on a major chemical/biological weapons factory (that supplies thirty percent of the enemy’s chemical/biological weapons) would likely cause debris that would destroy three neighboring houses (killing fifteen noncombatants).  The intended end of the action is the destruction of the chemical/biological weapons factory; the destruction of the houses and deaths of the noncombatants are not part of the intended end.  Because the action does not aim at intentionally harming the noncombatants as an end, it meets the requirements set by the first condition of the revised DDE.  The harm to the noncombatants is also not a means, or a proximate means, of achieving the intended end.  And, the harm to the noncombatants is outweighed by the good accomplished by achieving the intended end.  The act can be characterized as good, and its end is the destruction of a legitimate military target.  Thus, the action also meets the requirements set by the second, third, and fourth condition and is therefore morally permissible.  If destroying the factory only eliminated a small percentage of an enemy’s chemical/biological weapon production (say less than five percent), while unintentionally killing ten thousand noncombatants, it would not meet the proportionality requirements set in the fourth condition.  However, because this action does meet the proportionality requirements, it is permissible. 

Another unambiguous situation occurred at My Lai during the Vietnam War when American soldiers attacked noncombatants. 

There was no sign of the Vietcong battalion and no shot was fired at Charlie Company all day, but they carried on.  They burnt down every house.  They raped women and girls and then killed them.  Pregnant women had their stomachs slashed open and were left to die.  There were mass executions.  Dozens of people at a time, including old men, women and children, were machine-gunned in a ditch.  In four hours nearly 500 villagers were killed (Glover 58). 


Because the action intentionally aimed at harming noncombatants as an end, it is clear that the DDE would not regard the actions of these soldiers as morally permissible.  Even if it was argued that the killing of noncombatants was a means, it is clear that this case would not even approach the threshold.    There were other methods available for the soldiers to defend themselves and the nation. Killing 500 noncombatants was not an act of self-defense; the villagers were not a threat to the soldiers, nor did their presence in the village make self-defense inconceivable for the United States or the soldiers.  The actions in My Lai violated the constraint against intentionally harming noncombatants and cannot be characterized as good.  Additionally, the soldiers were not seeking to destroy legitimate military targets.  Clearly, then, the actions at My Lai do not meet the conditions of the DDE and are morally impermissible. 

In a case called Platoon Defense, a platoon defending an area comes under attack by an enemy unit.  Because of the direction that the enemy is attacking, some of the platoon small arms defensive fire will likely hit noncombatants located in a town to the front of the platoon position.  The platoon’s intended end is the defeat of the enemy, with the death of the enemy soldiers as a means to accomplish this.  The foreseen but unintended side effect is the death of some of the noncombatants in the town.  The difficulty in analyzing this case is that in order for the soldiers to hit the enemy, they have to aim their weapons in the direction of the town.  While it is clear that the soldiers in the platoon are not seeking to harm the noncombatants as ends or means, the means and the foreseen but unintended side effects are very close.  However, the right of nations to defend themselves against unjust aggression, and the right of soldiers to self-defense, seems relevant to this situation.  The right to self-defense is not even conceivable unless the soldiers fire at the enemy soldiers.  Firing at the enemy soldiers meets the conditions of the revised DDE, thus making it permissible for the platoon to defend itself against the enemy unit.  Yet, even though the action is permissible, the platoon should take actions that reduce the likelihood of noncombatant deaths.  This will allow the platoon to, as much as possible, adhere to the constraint against intentionally harming noncombatants.  They are only allowed to do what is minimally required for self-defense. 

            The case of Area Bombing involves the attack of an area that contains both combatants and noncombatants.  An enemy unit retreats into a town, and the entire town is bombed because the precise location of the enemy soldiers is not known.  The harm to the noncombatants is not aimed at; the intended end is the destruction of the enemy unit that retreated into the town.  However, the harm to the noncombatants could be viewed as the intended means thus making it difficult to classify the noncombatant deaths as a foreseen but unintended side effect.  By bombing the town, the agent seems to intend to kill as many people as possible and only hopes that the enemy soldiers are among the dead.  It does not seem possible to separate the intended end of harming the combatants from the harm caused to the noncombatants.  The town must be destroyed in order for the enemy unit to be destroyed.  Additionally, the threshold is not crossed because the right to self-defense is not at stake.  Therefore, because of the closeness of the foreseen harm to the intended means, and the fact that the case is not near the threshold, this action is not permissible.  We do not even need to consider the fourth condition (proportionality) because the first three conditions are not met.  From this analysis we see that most area bombing cases clearly do not meet the four conditions of the DDE.  However, there are other cases that are more difficult.  

During war some nations may place noncombatants at strategic military targets to deter the enemy from destroying these targets.  The opposing military force must decide whether to destroy the target or forego its destruction because of the presence of the noncombatants being used as human shields.  For instance, if twenty noncombatants were placed on the roof of the chemical/biological weapons factory discussed above in Strategic Bombing, a Human Shield case would exist.  For the missile to hit the factory it also must hit the noncombatants; the agent must aim at and harm the noncombatants (just as the hostage must be aimed at and harmed to kill the terrorist).  Applying the DDE in this case is difficult because the intended means and the foreseen but unintended side effects are extremely close, and a third agent is involved.  Aiming at the factory appears to be the same as aiming at the noncombatants and would seem to violate the constraint against intentionally harming noncombatants.  However, the noncombatant deaths are not needed; it is only necessary to kill the noncombatants because they are occupying the same space as the factory. 

Could the fact that a third agent placed the noncombatants on the factory make this action permissible?  If third agents put the noncombatants on the roof of the factory, it might make us more likely to destroy that factory instead of the one that contained noncombatants who had simply sought shelter from the rain because in the first case we would hold the third agent at least partially responsible for the deaths of the noncombatants.  Third agent involvement slightly lowers the threshold and shifts some of the responsibility for the deaths, however, it will not make a significant moral difference by itself.  What is more important in both of these cases is that the right to self-defense is fundamentally at stake.  Therefore, even though we are intentionally harming noncombatants as a means to destroying the military target (or at least the unintended consequences are very close to the  means), it is morally permissible to bomb the factory if failure to do so would mean giving up the right to self-defense.  A historical example may illustrate this point.

            During World War II the Germans attempted to transport heavy water on a ferry from Norway to Germany so that German scientists could develop atomic weapons.[15]  Civilians, who did not know about the presence of the heavy water, were also on board.  The presence of the noncombatants on the ferry had to be considered before any decision was made in regards to sinking the ferry (Glover 91).  Because the sinking of the ferry (a legitimate military target) and the killing of the noncombatants were so close, the DDE would forbid sinking the ferry unless the agent could argue that a threshold had been crossed.  Third agent involvement may be a slight factor in this case, however, the more important issue is whether adhering to the noncombatant prohibition makes it impossible for the allies to defend themselves.  If the Nazis possessed atomic weapons, they would have likely used them resulting in hundreds of thousands of deaths.  Because of the potential effects of the Nazis possessing atomic weapons on the Allies’ ability to defend themselves, the case is above the threshold.  Therefore, the DDE’s judgment that sinking the ferry was permissible concurs with our intuitions and the DDE’s principles.

            In Terror Bombing cases a military force attacks the civilian populace to decrease citizen morale.  In one version an agent claims that terror bombing is permissible under the DDE because he only needs for the citizens in a city to appear dead, and he is not intending their death (even though he is bombing the city).  This rationale, however, does not justify the terror bomber’s actions because it appears that civilians are being intentionally targeted.  This type of action is not permissible because the harm to noncombatants is serving as a proximate means to the means of making the civilians appear dead.  Also, our right to self-defense is not at stake, as it is in the previous example.  Therefore, in this version it is clear that the DDE provides an acceptable moral distinction and would determine that the bombing was not permissible. 

In another version of the case promoted by David Lewis, “the bomber knows that it will be enough to accomplish his task to explode his bombs over the target city.  Strictly speaking all that he intends is for his bombs to explode over the city.  He does not (or need not) intend any effects on the city itself (or its population)” (Woodward 218).  In this case it can be said that while the action may aim at lowering the morale of the civilian population, it is not intentionally seeking to harm the noncombatants (because the missiles are exploded in the air and are only meant to scare the noncombatants in the city).  The harm to the noncombatants in the city (there will likely be deaths) can be categorized as foreseen but unintended.  While the foreseen side effect is close to the intended means, it is not as close as the first terror bombing example.  Additionally, an appeal can be made to the fourth condition.  If it can be demonstrated that the good achieved outweighs the foreseen harm, then this action may be deemed permissible by the DDE.  However, classifying this action as morally permissible seems wrong because noncombatants are being involved in a military action.  The bomber “deliberately involves his victims in something to further his purposes and this involvement, as he foresees, is harmful to them” (Woodard 219).  Because a military target is not present (as it is in strategic bombing), the action seems to target noncombatants and cause them harm.  Yet, because the bomber isn’t directly aiming at the noncombatants, it is difficult for the DDE to make a morally relevant distinction.  Therefore, this type of case is outside of the DDE’s jurisdiction; there seems to be a moral factor present in this case that the DDE does not recognize.  This factor is that the agent is involving noncombatants in a military action, even though the noncombatants are not actually targets and the harm is not intended.  Thus, while the DDE has trouble making a determination in this case, we recognize that it contains a unique factor and that another set of guidelines would be needed to make a proper distinction.[16]

4.  Potential Criticism. 

One possible criticism of my formulation of the DDE is that the introduction of a threshold diminishes the role of the DDE in collateral damage cases.  However, despite the fact that the threshold alters how the DDE is applied in certain cases, the standard formulation of the DDE is still doing most of the work in determining whether or not an action is morally permissible.  The presence of a threshold seeks to prevent the DDE from judging an act to be permissible, when in fact this act violates the DDE’s fundamental principles.  Thus, introducing a threshold to the DDE, and understanding the problems of proximate means and closeness, ensures that the DDE is true to its principles.  In difficult cases, the DDE is still doing the work, however, the threshold appeals directly to the foundational principles.  Appealing to these principles ensures that the DDE is accomplishing the purpose that it was intended to accomplish. 

It also might be argued that the use of thresholds seriously weakens the prohibition against intentionally killing noncombatants.  Thresholds could possibly justify too much and make the constraint against intentionally killing combatants extremely weak.  This is a valid criticism, and care must be taken when using thresholds.  However, thresholds, as I have defined them, will be very high.  The constraint against intentionally killing noncombatants will take priority in most cases, and thresholds will only be considered in cases where the right to self-defense is at stake or is being fundamentally violated. 

The use of a threshold recognizes that it is difficult to maintain an absolute prohibition against any action, including the prohibition against intentionally harming noncombatants.  The DDE recognizes that the lives of noncombatants are important, but so is the right to self-defense.  People have the duty not to intentionally harm innocent people, but they also have the right to defend themselves.  In war, military forces exercise a political community’s right to self-defense, but, in certain cases, this right to self-defense could be threatened or violated in such a way that it can no longer be said to exist.  It is not only that the people the military forces are defending may be harmed, but that it no longer is possible to even talk of defending them; self-defense is inconceivable and impossible.  In many of these cases the right to self-defense takes priority, but even when this happens, we should still try to minimize harm to noncombatants and adhere to our duty not to harm noncombatants as much as possible.  It is still required that our actions are able to be characterized as good.  Consequently, in extreme cases we have two options.  We can completely give up the right to self-defense in favor of the prohibition against intentionally harming noncombatants.  Or, we can exercise our right to self-defense while still adhering, as much as possible, to the duty not to intentionally harm noncombatants.  Some examples may illustrate when the right to self-defense is, and is not, at stake.

Imagine that Saddam Hussein has placed noncombatants (as human shields) at Iraq’s only nuclear missile silo, and the silo is just hours away from launching a nuclear missile at the United States.  It seems permissible to attack the silo (even though noncombatants will be killed) because failing to do so would be giving up our right to defend ourselves as a nation.  In this situation, adhering to the prohibition against attacking noncombatants would make the right to self-defense inconceivable.  At the other extreme imagine that one noncombatant is tied to a retreating tank.  While it would be militarily beneficial to attack the  tank, the right to self-defense is not at stake.  It is still possible to conceive of self-defense without attacking the retreating tank, even though the tank might one day be used in offensive operations.  The tank example would approach the threshold only if a noncombatant was tied to all Iraqi tanks, and these tanks were part of a large scale offensive.    

Another reason that thresholds do not seriously weaken the constraint against killing noncombatants is that the DDE with thresholds does not allow combatants to direct military action exclusively against noncombatants.  As I discussed above, military forces should only seek the destruction of military targets, and their ends should never be harm to noncombatants.  In the Terror Bombing case we see the distinction that the DDE makes when applied to cases where the military targets civilians.  The DDE either judges this type of action to be impermissible or, in certain situations, it cannot make a distinction at all which forces people to rely on their intuitions or another moral principle.  Therefore, even with thresholds, military forces are still required to aim at military targets.[17]  Thresholds just recognize that aiming at the military target sometimes entails aiming at noncombatants, so in some cases it might be permissible for the harm to noncombatants to serve as a means to destroying legitimate military targets .  The thresholds, as contained in the DDE, will not allow for intentional harm to noncombatants as an end, as in My Lai or Terror Bombing, but instead will allow for the destruction of a justifiable military target, as in Human Shield or the Ferry. 

Another potential criticism is that the closeness of the intended means and foreseen harm, and the problem associated with proximate means, will still create problems for the DDE.  Despite the fact that the DDE cannot make a distinction that fits with our intuitions and the DDE’s principles in all collateral damage cases, the fact that it does in many cases is all that is necessary for the DDE to be useful.  There will be cases where it is impossible to interpret how close the intended means and foreseen harm actually are, or what is and what is not a proximate means.  These cases will be outside the jurisdiction of the DDE, and it is these types of cases that must be evaluated using another moral principle or theoretical framework. 

5.  Conclusion. 

            In this paper I have attempted to demonstrate that the DDE provides an acceptable moral distinction that fits with our moral intuitions, and its foundational principles, in many cases that involve collateral damage in war.  The DDE with a threshold prohibits intentional harm to noncombatants as an end, allows intentional harm to noncombatants as a means only when a threshold has been crossed, and requires that our actions be able to be characterized as good.  In many cases, these requirements allow one to separate permissible actions from impermissible actions.  Additionally, by closely examining actions, one can reduce the confusion caused by proximate means and closeness of intended means and foreseen but unintended side effects.  While these problems cannot be eliminated, recognizing that they exist, and taking steps to reduce the confusion caused by them, will make the DDE distinction more in line with our intuitions.  In those cases that are outside the jurisdiction of the DDE, it is clear that other factors are involved besides not intentionally harming noncombatants, performing good actions, and the right to self defense.  To evaluate these types of cases, another framework is required. 

Examining the DDE’s foundational principles, and including a threshold in its formulation, recognizes that the DDE is simply a heuristic method.  Standard versions of the DDE can sometimes betray the principles in which they were developed to promote, which is a general problem with heuristic methods.  To avoid the problems associated with “rule worship”, and to ensure that a heuristic method accomplishes its purpose, it is necessary to examine cases to ensure that a heuristic method accurately reflects its foundational principles.  Also, it is reasonable to assume that most moral principles are not absolute, thus allowing, in rare instances, for the intentional harm to noncombatants to serve as a means of destroying a military target.  This does not seriously weaken the prohibition against harming noncombatants and ensures that the DDE accomplishes its purpose. 

6.  References. 



Christopher, Paul.  The Ethics of War and Peace: An Introduction to Legal and Moral

Issues.  New Jersey: Simon & Schuster, 1999. 


FitzPatrick, William.  “A Puzzle about the Doctrine of Double Effect”.  Philosophy 5334 Reader.  Virginia

Tech, August 31, 1998


FitzPatrick, William.  “Normative Ethics: September 4, 2001 Handout”. Virginia Tech,



Glover, Jonathan.  Humanity: A Moral History of the Twentieth Century.  United States: Yale University

Press, 2001.


Kagan, Shelly.  Normative Ethics.  Colorado: Westview Press, 1998. 


Quinn, Warren S.  “Actions, Intentions, and Consequences: The Doctrine of Doing and Allowing”.  Ethics:

Problems and Principles.  Edited by John martin Fischer and Mark Ravizza.  Texas: Holt, Rinehart,

and Winston, 1992. 


Quinn, Warren S.  “Actions, Intentions, and Consequences: The Doctrine of Double Effect”.  Ethics:

Problems and Principles.  Edited by John martin Fischer and Mark Ravizza.  Texas: Holt, Rinehart,

and Winston, 1992. 


Regan, Richard.  Just War: Principles and Cases.  Washington D.C.: The Catholic

University of America Press, 1996.


Walzer, Michael.  Just and Unjust Wars: A Moral Argument with Historical Illustrations. 

New York: Basic Books, 1977.


Woodward, P.A.  “The Importance of the Proportionality Condition to the Doctrine of Double

Effect: A Response to Fischer, Ravizza, and Copp”.  The Doctrine of Double Effect.  Edited by P.A. Woodward.  Indiana, University of Notre Dame, 2001. 





[1] Collateral damage is usually thought of as harm to nonmilitary targets to include people and property.  I will be focusing on harm to people. 

[2] By ‘acceptable moral distinction’ I mean that the DDE is able to serve as a principle that, when applied to a case, allows one to determine if an act is morally permissible.  That is, it usually able to distinguish morally permissible actions from morally impermissible actions.

[3] It would be great if the DDE could provide an acceptable moral distinction in all cases, but like other moral principles it cannot.  It is enough that it is usually able to provide acceptable moral distinctions and that we can recognize when it is not able to do so.  The doctrine of doing and allowing, or a focus on rights, will be useful in many of the cases that are out of the DDE’s jurisdiction.  I would like to thank Bill Fitzpatrick for his assistance with this point. 

[4] From this point forward when I refer to harm I will be referring to serious harm.  My definition of harm will not include trivial harm, but will include actions such as killing, serious injury, or torture.  Extending the DDE constraint against intentional harm to include trivial harm would make it very difficult to apply the DDE because of the many activities that could be considered harmful. 

[5] I will not debate the point of whether a guilty person has a right to self-defense.  Hobbes claims that “no man can transferre, or lay down his Right to save himselfe from Death, Wounds and Imprisonment and therefore the promise of not resisting force, in no Covenant transferreth any right, nor is obliging” (Hobbes XIV, 69-70).  I am inclined to agree with Hobbes on this point, but this is not important for my paper.

[6] While the DDE may have more than the three foundational principles that I listed, the three principles that I mention seem to capture the DDE’s view concerning harm.  I will not offer a thorough defense of these principles because various ethical perspectives support principles of this sort, even if the ethical perspectives do differ on how the principles should be applied, or which has priority.    

[7] This example was given by Bill Fitzpatrick during the Fall 2001 Normative Ethics seminar at Virginia Tech.

[8] Jones, of course, is not truly innocent, however his theft does not make it permissible for Smith to kill or seriously injure him.  Retrieving one’s stolen wallet does not allow one to inflict serious harm on the thief.     

[9] Jonathan Glover presents two tests for evaluating closeness.  However, he finds that “the weaker test (‘were the consequences wanted?’) allows too much [while] the stronger version (‘were the deaths so bound up with the act that we cannot say they died by accident?’) allows too little” (Glover 108).  The use of thresholds and the flexibility provided by the right of nations to defend themselves against unjust aggression clears up some of the difficulty associated with closeness.

[10] Because of the importance of the constraint against intentionally harming innocent people, thresholds must be placed at high levels.  The constraint not to intentionally harm an innocent person must be significantly outweighed by the right to self-defense, and the burden is on the agent to show this. 

[11] The distinction between combatants and noncombatants has been recognized throughout history.  “The term combatants refers primarily to members of the armed forces, but can include certain political leaders who are engaged in planning and carrying out the war effort as well as civilians who are working on behalf of the military such as those working in munitions factories and radar sites.  Shopkeepers, farmers, and other members of what Grotius refers to as the ‘peaceable population’ are not considered combatants” (Christopher 160).  Prisoners of war, surrendering soldiers, and those disabled from fighting can also be considered noncombatants. 

[12] In war, soldiers maintain the right to self-defense, however, they can now harm and be harmed by other soldiers.  That is, combatants are not considered innocent because in war it can be rightfully assumed that enemy combatants intend to harm you.  Consequently, because soldiers maintain the right to self-defense and their duty not to intentionally harm innocent people does not apply to enemy combatants, it is morally permissible for a soldier to intentionally harm enemy combatants (unless the enemy combatants give up their combatant status). 

[13] Thresholds recognize the importance of not intentionally harming a noncombatant, while also recognizing the importance of the DDE’s other two principles.  The threshold allows that, in certain types of cases, it is permissible to harm noncombatants  when failure to do so makes the right to self-defense inconceivable. Crossing the threshold does not make it permissible for one to aim at killing noncombatants as an end.  Military action should be aimed at military targets and noncombatants should not be viewed as military targets.  Therefore, crossing the threshold often makes it permissible to attack a military target, even though the foreseen harm of noncombatants is close to the intended end of the attack, or the harming of the noncombatants is a means to reaching the military target. 

[14] Michael Waltzer, in Just and Unjust Wars, further refines the DDE by adding the condition that the agent “aware of the evil involved [from the foreseen but unintended side effects], seeks to minimize it, accepting costs to himself” (Walzer 155).  I believe Walzer makes a good point. However, I believe that because the DDE directs us to perform good intentional actions, and to adhere to the duty of not intentionally harming noncombatants, agents will already seek to minimize foreseen but unintended collateral damage.  Therefore, this additional constraint is implicit in the factor of promoting the good and does not need to be included in the DDE formulation. 

[15]Heavy water is essential to the production of atomic weapons.

[16] Perhaps a principle such as Quinn’s harmful agency would provide a relevant moral distinction in this case.

[17] Walzer has proposed that a “supreme emergency” may justify terror bombing, as in the British attack on German cities during World War II (Walzer 252).  I will not examine this argument, however I will note that the thresholds that I have included in the DDE only apply to cases where noncombatants are harmed as a means of destroying a legitimate military target.