Meant or Merely Foreseen: The Doctrine of Double Effect and the Power of Intention

The Hook: Two Bombers, One Outcome

Picture two fighter pilots in the same war. Both take off the same night, both drop bombs on an enemy city, and both, in the end, kill exactly one hundred civilians. The death toll is identical, the aircraft is identical, the explosive yield is identical.

The first pilot is a terror bomber. His mission is to break the morale of the civilian population. The dead civilians are not, for him, a regrettable collateral cost but the very point of the raid: only their deaths—and the fear of the survivors—will force the enemy to the negotiating table. Had the people miraculously survived unharmed, he would have failed his mission.

The second pilot is a tactical bomber. His target is a munitions factory on the city's edge. He knows with lethal certainty that the blast will collapse the neighboring apartment blocks and kill a hundred people. He does not want those deaths, he draws no benefit from them, and if all of them had climbed unhurt from the rubble the next morning, his mission would not have been diminished in the slightest—the factory would still be destroyed.

Now the decisive question: do the two act in a morally equivalent way? Almost everyone hesitates here—and most, in the end, say: no. The terror bomber does something the tactical bomber does not, even though the outcome—one hundred dead civilians—is the same. The difference lies not in the world but in the mind: one intends the deaths as his means, the other foresees them and accepts them as a side effect.

This exact intuition has a name and a history more than seven hundred years old: the doctrine of double effect (Latin duplex effectus, DDE for short). Its core claim is as simple as it is consequential: sometimes it is permissible to cause a grave harm as a foreseen but unintended side effect of a good action—even though it would be forbidden to bring about that very same harm as a means to the same good end. Intention, the claim goes, can make one of two outwardly identical acts permissible and the other wrong.

Unlike the trolley problem, which constructs a rare extreme situation, double effect sits at the heart of our everyday legal and moral sense—in the difference between murder and manslaughter, in palliative medicine, in the laws of war, and, as we shall see at the end, right inside the engine room where engineers decide which risks are acceptable.

The Core Concept: The Line Between Means and Side Effect

To grasp why double effect is so powerful and yet so contested, you have to see the distinction it rests on. It does not run between a good and a bad outcome—both bombers produce the same bad outcome. It runs between two ways a harm can be tied to our plan of action.

A harm is intended when it is either the goal of the action or a means on the way to that goal. The civilians' deaths are, for the terror bomber, a means: the causal chain that is supposed to lead to the war's end runs through them. Remove the deaths from the plan, and the plan collapses.

A harm is merely foreseen, by contrast, when it occurs as an accompaniment without itself being part of the plan. The civilians' deaths are, for the tactical bomber, a side effect: remove them from the world, and the plan—the destruction of the factory—works just as well. The harm is a by-product, not a cog in the machine.

A simple test helps separate the two cases: the test of the thwarted outcome (sometimes called the counterfactual test). Ask yourself: would the agent be disappointed if the harm failed to occur? The terror bomber would be dismayed if the civilians survived—his plan would have failed. The tactical bomber would be relieved—his plan would have succeeded and at a lower price. Whoever misses the harm when it fails to happen intended it. Whoever does not miss it merely foresaw it.

This test is more than a philosophical toy. It corresponds to the legal distinction between direct intent (dolus directus) and what we treat as recklessness or conditional intent, and it separates murder from negligent homicide. Double effect is thus the spelled-out core of an idea that virtually every legal system in the world relies on: what a person had in mind is often, for the moral and legal assessment of the act, just as important as what actually happened.

Part 1: The Origin—Thomas Aquinas and the Blow Struck in Self-Defense

Double effect is traditionally traced to Thomas Aquinas (1225–1274), more precisely to a single passage in his Summa Theologica (second part of the second part, Question 64, Article 7). The question he takes up there sounds concrete: may one kill an attacker in order to save one's own life?

That was no trivial question. Augustine, centuries earlier, had rejected killing in self-defense: it springs, he thought, from an inordinate self-love. Aquinas resolves the problem with a thought that would go on to shape Western ethics:

"Nothing hinders one act from having two effects, only one of which is intended, while the other is beside the intention. … Accordingly, the act of self-defense may have two effects: one, the saving of one's life; the other, the slaying of the aggressor." (Thomas Aquinas, Summa Theologica II-II, q. 64, a. 7)

The decisive move: whoever defends himself intends the saving of his own life; the death of the aggressor lies "beside the intention" (praeter intentionem). Not because it is unforeseeable—Aquinas was no naïf—but because it is not what the action aims at. The killing is permitted, provided it does not become the real purpose.

And Aquinas immediately adds a limit that is easily overlooked but became central to the later doctrine: the act must not be "out of proportion to the end." Whoever uses more force in self-defense than necessary acts unlawfully. Proportionality is, from the very beginning, not an add-on but a condition. Intention alone justifies nothing; it must be accompanied by measure and necessity.

It is worth noting that Aquinas himself never formulated a "doctrine" with numbered conditions. He resolved a concrete case. It was the late-medieval and Baroque moral theologians—the casuists—who forged his observation into a general principle. The classic four-condition formulation we cite today took its polished shape from the Jesuit Jean Pierre Gury in the nineteenth century and was canonized by authors such as Joseph Mangan (1949) and the New Catholic Encyclopedia.

Part 2: The Four Conditions

In its classic form, one may perform an act with both a good and a bad effect only if all four of the following conditions are met simultaneously. The table below summarizes them and shows, in each case, where our two bombers fail or pass.

Two things are remarkable about this structure. First, the second condition—the intention criterion—is the real innovation; all three others appear, in some form, in ordinary weighing-of-consequences ethics too. Second, the fourth condition, proportionality, is a consequentialist element planted in the middle of an intention-centered doctrine. Double effect is therefore not a pure ethics of disposition: it demands that you weigh the consequences and adopt a particular inner stance toward the harm.

A common misunderstanding can be cleared up right here: double effect does not claim that "not intended" automatically means "permitted." The tactical bomber who kills a thousand civilians out of pure convenience in order to hit an insignificant shack acts wrongly despite lacking any intention to kill—he fails on proportionality (condition 4). And every serious version additionally requires that the agent seek to minimize the harm rather than shrug and accept it. The philosopher and military ethicist Michael Walzer sharpened this: whoever accepts civilian casualties as a side effect must be willing to shoulder additional risk or forgo some benefit in order to keep those casualties small. Intention without diligence is no excuse.

Part 3: The Canonical Cases

The force of double effect shows less in its definition than in a series of case pairs that all follow the same pattern: two actions lead to the same harm, but one seems permissible and the other wrongful—and the only difference is the role the harm plays in the agent's plan.

Euthanasia vs. pain-relieving treatment. A doctor who injects a terminally ill patient with a large dose of morphine in order to bring about death commits, on the traditional view, a killing. A doctor who gives the same dose in order to relieve pain, merely foreseeing a possible shortening of life, acts permissibly. Same syringe, same dose, same physiological effect—different intention. (That the underlying medical assumption—that correctly dosed opioids shorten life—is largely a myth, we will return to.)

Abortion vs. hysterectomy. A pregnant woman has uterine cancer. If the doctor removes the cancerous uterus to save her life, the fetus dies as a foreseen side effect—permitted, on the doctrine. A direct abortion that kills the fetus as a means of rescue would, by contrast, be forbidden. It was precisely this example that in 1967 led the philosopher Philippa Foot to engage with double effect—and, in doing so, to invent by accident the most famous thought experiment in moral philosophy.

Self-defense vs. pre-emptive killing. Whoever kills an attacker in a fight acts permissibly, even foreseeing the fatal outcome. Whoever kills that same person in cold blood because he knows the man is plotting an attack later kills intentionally—and acts wrongly.

Suicide vs. the soldier on the grenade. A soldier who throws himself on an exploding grenade to shield his comrades intends their rescue and merely foresees his own death. He is not a suicide, though he certainly loses his life. Again, intention decides, not outcome.

These cases have astonishing intuitive pull. They explain why double effect has survived so long: it puts into words a pattern of judgment that most people already share, long before they have ever heard of Thomas Aquinas.

Part 4: Foot, the Trolley Problem, and the Road into the Modern Era

When Philippa Foot wrote her 1967 essay "The Problem of Abortion and the Doctrine of the Double Effect," she actually meant to test whether double effect really does what is attributed to it. To test it, she devised a pair of cases that stripped away every cultural charge of the abortion debate and left only the bare skeleton of the problem.

The first case: the driver of a runaway tram can only choose between two tracks. On one stand five workers, on the other, one. If he steers away, he kills one instead of five. Most people find: steering is permissible. The second case: a judge can prevent a riot that will cost five lives only by having an innocent man executed. Most people find: that is murder. In both cases one is sacrificed to save five—so why do we judge so differently?

Foot's answer ran, at first, through double effect: the tram driver does not intend the death of the one, he merely foresees it; the judge, by contrast, uses the death of the innocent as a means to pacify the mob. But Judith Jarvis Thomson in 1985 sharpened the scenario into what everyone now knows as the trolley problem: not the driver but a bystander at a switch decides. And she paired it with the notorious footbridge case: the same observer could save the five by pushing a large man off a bridge into the trolley's path. Arithmetically identical—one dies, five live—yet almost no one thinks the pushing is permissible.

Double effect offers an elegant explanation: in the switch case the death of the one is a side effect (remove it, and five are still saved); in the pushing case the man's death is the means (only because his body stops the trolley are the five saved—remove his death and the plan fails). The test of the thwarted outcome separates the cases cleanly.

Yet Thomson herself became, over the years, a skeptic. In "Turning the Trolley" (2008) she recanted her earlier view and argued that a mere bystander may not throw the switch at all—one may not force a person into a lethal position into which no one else had put him. The debate over whether double effect really explains the trolley intuitions, or whether we merely retrofit the principle onto our feelings, remains open to this day. And it is at exactly this seam that the modern critique attacks.

Part 5: The Critique—Where the Doctrine Cracks

As intuitive as double effect seems, the objections are just as stubborn. Three are especially sharp.

The closeness problem. Where exactly does the line run between "harm as a means" and "harm as a closely adjacent side effect"? Consider a caver stuck in a narrow exit while the water rises and his companions risk drowning. They blast him free with dynamite. Do they intend his death—or only his dismemberment and removal from the path, with death a side effect? That sounds like hairsplitting, and that is precisely the charge: whoever describes it cleverly enough can recast almost any intended killing as "mere blasting-free with a lethal side effect." In the footbridge case one can say that one does not intend the large man's death, only that his body stops the trolley—death being a side effect. If the doctrine can be evaded so easily, what does it still accomplish?

The side-effect effect (the Knobe effect). The experimental philosopher Joshua Knobe showed something unsettling in 2003: our judgment about whether something happened intentionally itself depends on whether we morally approve of it. In his famous experiment a manager says, "I don't care about the environment, I just want profit." If the program harms the environment, ~82% of respondents say he brought the harm about intentionally. If the same indifferent program helps the environment, only ~23% say he helped it intentionally—with identical indifference. Moral judgment thus colors the attribution of intention, not the other way around. This is a serious danger for double effect: if we call a harm "intended" merely because we condemn it, then intention can no longer be the neutral basis on which the judgment rests. It would be just a label we stick on afterward.

Quinn's direct vs. indirect agency. The philosopher Warren Quinn tried to rescue the sound core of the doctrine by shifting the line: what matters, he argued, is not "means vs. side effect" but whether the agent actively involves the victim in his plan (direct agency) or whether harm merely befalls the victim on the side (indirect agency). But this too wobbles: the soldier who throws himself on the grenade involves his own body very directly in the rescue plan—yet his act counts as permissible. On closer inspection, the clean categories dissolve.

The most fundamental objection, though, runs: perhaps double effect is not a single principle at all, but only an umbrella term for a heap of individual exceptions to a general prohibition on killing. The critic Alison McIntyre argues that in every concrete case substantive considerations—medical diligence, patient consent, the conventions of war, proportionality—do the real justificatory work. The intention distinction would then be just a veneer over a far more variegated moral landscape.

Part 6: The Empirical Turn—Is Double Effect Wired Into Our Minds?

Here modern cognitive science enters the stage, and it poses a fascinating question: if double effect is so contested as a philosophical theory—might it nonetheless be a psychological fact? Do people around the world actually act on it, whether they know it or not?

Cushman, Young, and Hauser (2006) tested, in Psychological Science, three possible principles that might underlie our moral judgments: (a) harm caused by action is worse than by omission; (b) harm as a means is worse than harm as a side effect (exactly double effect); (c) harm involving physical contact is worse than harm without it. The result was doubly revealing. First: subjects did indeed judge in line with all three principles—including double effect. Second, and this is the real finding: when asked to justify their judgment, they could name the action principle and the contact principle—but not the intention principle. They judged according to double effect but could not say that or why they did so. This is called a dissociation between judgment and justification: the principle operates, but it operates unconsciously.

Hauser and colleagues (2007) buttressed this with data from the "Moral Sense Test," a web-based study with participants from over a hundred countries. The judgments followed the double-effect pattern with remarkable stability—largely independent of gender, age, education, religion, or nationality. This suggests that the means/side-effect distinction is not a piece of specialized Catholic-theological knowledge but something that appears to be deeply anchored in human moral cognition.

Joshua Greene's dual-process theory supplies the neural narrative. On the basis of fMRI studies, Greene argues that our brain processes moral dilemmas with two systems—an idea closely related to the two-system view of cognitive science. A fast, automatic, emotional system fires especially in "personal" harms such as pushing the man off the bridge and produces the intuitive "No!" A slow, deliberative, more calculating system weighs costs and benefits and, in the detached switch-throwing, leans toward the utilitarian "Yes." Greene's provocative thesis: our double-effect intuitions may be less a deep moral truth than the echo of evolutionarily ancient emotional alarm bells that ring louder at physical, direct violence.

Double effect thus stands at a strange crossroads. As a normative principle—as a prescription for what we ought to do—it is philosophically battered. As a descriptive account of how people actually judge, it is empirically remarkably robust. I am of the opinion that this very split is the most important lesson: the fact that we all feel according to a principle does not yet prove the principle valid. The Knobe effect and Greene's alarm bells warn against mistaking a merely deep-seated intuition for a moral truth—a caution that moral luck and population ethics teach us as well.

Part 7: Where Double Effect Really Counts Today

The debate sounds academic, yet it decides matters of life and death in four very real fields.

Palliative medicine. Double effect has for decades been the ethical backbone of end-of-life pain therapy: a doctor may give opioids to relieve suffering, even foreseeing a shortening of life. But it is worth looking at the facts here, since scientifically grounded claims matter. The premise that expertly titrated opioids typically shorten life is, on current research, largely a myth. Review studies (e.g., Sykes and Thorns 2003 in the Lancet Oncology; Fohr 1998) find no evidence that carefully dosed opioids dangerously suppress respiration. This has consequences: the widespread fear of the "lethal final dose" leads to pain being under-treated. Double effect was long the justification here for something that needs no such dramatic justification at all—and its false premise may have caused more suffering than it prevented.

Terminal sedation. Things get sharper with the continuous deep sedation of dying patients. In 1997 the U.S. Supreme Court invoked double effect in Vacco v. Quill to distinguish it from (there prohibited) active euthanasia. But the physicians Timothy Quill, Rebecca Dresser, and Dan Brock argued in the New England Journal of Medicine that double effect, consistently applied, ought actually to forbid terminal sedation: unlike with an opioid, where death is only a possible side effect, withholding fluids and nutrition from a deeply sedated patient leads inevitably to death—and is often done with exactly that intention. A principle that forbids a common and humane practice can hardly be its foundation. Here the doctrine reaches its limit.

The laws of war. The distinction between terror bomber and tactical bomber is cast into international humanitarian law. The rules of customary law, as codified by the International Committee of the Red Cross, prohibit attacks on civilians (intention) and at the same time require the minimization of civilian collateral harm (proportionality, advance warning, precautions). Double effect supplies the conceptual skeleton—but, as the ICRC shows, by no means the whole justification: without the duty to warn and to minimize harm, mere "non-intention" would be far too cheap an excuse.

Autonomous vehicles and AI. And finally the seven-hundred-year-old question lands in source code. When a self-driving car swerves away from an unavoidable accident, does it "intend" the harm it accepts in doing so? Programmers who write collision algorithms are, in effect, making double-effect decisions—only that here no spontaneous intuition applies but an explicitly coded rule. This forces a clarity that philosophy managed to duck for centuries: one must define what "intended" even means for a machine. That links double effect directly to the question of whether machines can have anything like intentions or understanding at all—the question of the Chinese Room.

The Central Takeaway

The practical core of double effect is not a theological relic but a sharp thinking tool that any professional who decides about risks and harms can use: separate the intended purpose of your action from the harms you merely foresee along the way—and examine each of them separately.

To do so, apply the doctrine's three questions in sequence. First, the test of the thwarted outcome: would my plan have failed if the harm had not occurred? If the answer is yes, the harm is your means, and you bear a far heavier burden of justification. Second, proportionality: does the intended benefit really outweigh the foreseen harm? Third—and this is the one most easily forgotten—the duty to minimize: have I done everything to keep the harm small, even at a cost in time, money, or convenience?

This grid is surprisingly useful far beyond medicine and war. Whoever shuts down a system and foresees a data loss stands in a different position from someone who brings about the data loss as leverage. Whoever discloses a security vulnerability to protect users, accepting a short-term risk of abuse, acts differently from someone who weaponizes the flaw. In a blameless post-mortem culture, exactly this separation is decisive: we assess the intention and diligence of those involved, not merely the accidental outcome.

At the same time, the critique counsels humility. The Knobe effect shows that we tend to call a harm "intended" precisely because we condemn it. The intention line is not a neutral ruler we simply read off—it is itself tinted by our value judgments. I am of the opinion that the most honest use of double effect is to treat it as a questioning grid and not as a license: it helps us ask the right questions but never replaces the substantive weighing—diligence, proportionality, consent, convention—that does the real work in the concrete case.

A Question to Reflect On

Think of a professional decision in which you accepted a harm—a migration that predictably locked out some users; a feature that drew resources away from somewhere else; a deadline met at the expense of diligence. Ask yourself honestly, in hindsight: was that harm really only a foreseen side effect of your goal—or was it, if you look closely, a means on which your plan secretly depended? And if the harm had been magically removed, would you still have had your success—or would you, secretly, have been disappointed?

Cross-References in the Vault

• The Switch in the Machine: The Trolley Problem and the Ethics of Autonomous Vehicles – The trolley problem arose from Foot's very examination of this doctrine; here it is applied to autonomous vehicles.
• The Child in the Road: Moral Luck and the Limits of Responsibility – The tension between intention and outcome from another angle: how much should chance be allowed to decide our guilt?
• The Billions Barely Alive: Parfit's Repugnant Conclusion and the Ethics of Future Generations – Another example of how robust moral intuitions collide with a consistent theory.
• The Chinese Room: Searle and the Question of Whether Machines Can Understand – Can a machine "intend" anything at all? The precondition without which double effect runs empty for AI.
• The Predictive Brain: Predictive Processing and the Illusion of Perception – The two-system model of mind underlying Greene's dual-process account of morality.
• The Two Boxes: Newcomb's Paradox and the Battle Over Rational Choice – A related conflict between two equally compelling principles of rational choice.

Sources

1. Alison McIntyre, "Doctrine of Double Effect," Stanford Encyclopedia of Philosophy (rev. 2023) – https://plato.stanford.edu/entries/double-effect/
2. Thomas Aquinas, Summa Theologica II-II, q. 64, a. 7 ("Whether it is lawful to kill a man in self-defense") – https://www.newadvent.org/summa/3064.htm#article7
3. Fiery Cushman, Liane Young, Marc Hauser, "The Role of Conscious Reasoning and Intuition in Moral Judgment: Testing Three Principles of Harm," Psychological Science 17 (2006), 1082–1089 – https://journals.sagepub.com/doi/abs/10.1111/j.1467-9280.2006.01834.x
4. Marc Hauser, Fiery Cushman, Liane Young, R. Kang-Xing Jin, John Mikhail, "A Dissociation Between Moral Judgments and Justifications," Mind & Language 22 (2007), 1–21 – https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0017.2006.00297.x
5. Timothy E. Quill, Rebecca Dresser, Dan W. Brock, "The Rule of Double Effect – A Critique of Its Role in End-of-Life Decision Making," New England Journal of Medicine 337 (1997), 1768–1771 – https://www.nejm.org/doi/full/10.1056/NEJM199712113372413
6. International Committee of the Red Cross, Customary International Humanitarian Law (Rules 14–24, proportionality and precautions) – https://ihl-databases.icrc.org/en/customary-ihl
7. Vacco v. Quill, 521 U.S. 793 (1997), U.S. Supreme Court – https://www.law.cornell.edu/supct/html/95-1858.ZO.html

Note: This article touches on euthanasia, end-of-life pain therapy, and sedation. It is intended as a philosophical and historical account and is not a substitute for medical or legal advice.

← All articles