Ethics
Doctrine of Double Effect
When a foreseen harm is permitted, but an intended harm is not
The Doctrine of Double Effect (DDE) holds that an action with both a good and a bad effect can be permissible if the bad effect is foreseen but not intended, the bad is not the means to the good, and the good is proportionate. Aquinas formulated it in Summa Theologica II-II Q.64 to justify killing in self-defense; today it underwrites the moral and legal distinction between palliative sedation and euthanasia, between collateral damage and terror bombing, and between flipping the trolley switch and pushing a man off the bridge.
- OriginatorThomas Aquinas, Summa Theologica II-II Q.64 a.7 (c. 1270)
- TraditionCatholic moral theology; deontological ethics
- ConditionsFour (act, means, intention, proportionality)
- Modern reviversAnscombe (1958), Foot (1967), Quinn (1989), Kamm
- Stock test casesTrolley switch vs footbridge; palliative sedation; tactical vs terror bombing
- Main rivalConsequentialism (denies intention/foresight matters)
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
How the doctrine works
The doctrine answers a recurring shape of moral problem: an action you would otherwise want to take produces, as a side effect, a harm you would never directly cause. May you act? Pure consequentialism says compare the totals and pick the larger sum. Pure pacifism says any act causally tied to a death is forbidden. Double effect carves a middle path: the act may be permissible, but only if the harm is genuinely a side effect — not used as the lever that produces the good.
Four conditions, in their now-standard scholastic form, must all hold:
- The act itself must be morally good or at least morally neutral. An intrinsically wrongful act (lying, blasphemy in older formulations) cannot be redeemed by good consequences.
- The bad effect must not be the means of producing the good effect. If you achieve the good through the harm, the harm is intended in fact, even if you would prefer a miracle. Pushing the large man on the footbridge fails here: his death is the brake.
- The agent must intend only the good effect; the bad effect must be merely foreseen. Intention is not just a private feeling. It is structurally what your plan needs to succeed.
- There must be proportionate reason — the good achieved must be commensurate with the harm risked. Saving one life by foreseeing the death of millions fails proportionality even if the other three conditions hold.
Together the conditions express a single thought: a person must never be used as the instrument of someone else's good, but a person may be harmed as the unavoidable cost of a separate, legitimate aim.
Aquinas's original case: self-defense
In Question 64 Article 7 of the Summa, Aquinas asks whether it is lawful for a private person to kill in self-defense. His answer reshapes Western ethics. "Nothing hinders one act from having two effects, only one of which is intended, while the other is beside the intention. Now, moral acts take their species according to what is intended, not according to what is beside the intention." The defender, in raising a sword to repel an attacker, intends his own preservation. The attacker's death is a foreseen consequence — a tragic side effect of a justified act of defense — not the goal pursued.
Aquinas adds a proportionality clause that modern textbooks sometimes omit: the defender must use no more force than necessary. Lethal force may not be intended even if foreseen as likely, and a clearly disproportionate response (killing a slap-thrower) loses the doctrine's protection. The argument is the seed of the just-war tradition, the rules of engagement that distinguish combatants from civilians, and the modern law of self-defense.
Worked example: the trolley problem
Philippa Foot introduced the trolley case in The Problem of Abortion and the Doctrine of the Double Effect (1967) precisely to test DDE. Set out the two variants side by side:
| Switch case | Footbridge case | |
|---|---|---|
| Action | Pull lever to divert trolley onto side track | Push large man off bridge into trolley's path |
| People killed | 1 worker on the side track | 1 large man on the bridge |
| People saved | 5 workers on the main track | 5 workers on the main track |
| Numerical outcome | +4 lives | +4 lives |
| Is the death the means? | No — track would still be diverted if it were empty | Yes — without the man's body the trolley does not stop |
| DDE verdict | Permissible (good means, foreseen harm) | Forbidden (harm is the means) |
| Survey intuition (Mikhail, Greene et al.) | ~85% say switch is permissible | ~12% say pushing is permissible |
The numbers are identical. A consequentialist must either declare both permitted or both forbidden. DDE — and most respondents — distinguishes them by the role the death plays in the agent's plan. The footbridge death is needed; the switch death is regretted but irrelevant to whether the diversion works.
Worked example: tactical bombing vs terror bombing
The cleaner laboratory case, often credited to Anscombe, comes from World War II ethics. Two pilots fly missions over an enemy industrial city; both kill exactly the same number of civilians.
- Tactical bomber aims at a munitions plant. Civilians living next to it die in the blast. The bomber would, given a costless way to do so, prefer the civilians be evacuated; the deaths add no value to her plan.
- Terror bomber targets civilians directly to break the enemy's morale and force surrender. He would, given a costless way to do so, want to confirm the deaths; without them his plan does not succeed.
By DDE, the tactical bomber may proceed if the proportionality and means conditions are met; the terror bomber acts impermissibly. Anscombe's Mr Truman's Degree (1956) used exactly this framework to argue that the United States' atomic bombings were murders, not legitimate acts of war, and to oppose Truman's honorary Oxford doctorate.
Worked example: palliative sedation
A patient with terminal cancer is in agony. Morphine at a high dose will relieve the pain; it may also depress respiration and shorten life by hours or days. May the doctor administer it?
- The act (administering analgesia) is medically appropriate.
- The earlier death is not the means of pain relief; the morphine relieves pain by binding opioid receptors, and the patient's pain would be relieved equally if she happened to live longer.
- The doctor intends pain relief and merely foresees the risk to respiration.
- Relieving severe terminal suffering is proportionate to the risk of hastening an already-imminent death.
All four conditions hold; the action is permitted. Most jurisdictions track this analysis legally — the same dose given with the explicit intention to kill (active voluntary euthanasia) is criminal in places where palliative sedation is routine. DDE is the philosophical scaffolding under that distinction.
Variants and refinements
| Position | What it adds | Why |
|---|---|---|
| Quinn's "doctrine of doing and allowing" (1989) | Distinguishes harmful direct agency (the victim's harm contributes to the agent's end) from harmful indirect agency (it does not) | Replaces the foggy "intended/foreseen" line with a clearer structural one that survives redescription |
| Kamm's principle of permissible harm | Harm to a victim is permissible only if it is an aspect or effect of a greater good's flowing to others, not a means | Refines case-by-case judgments where Quinn's binary is too coarse (loop trolley variants) |
| Counterfactual test (Bennett, modified by Quinn) | Ask: would the agent welcome a miracle in which the bad effect didn't occur but the good still did? Yes = side effect; no = means | Operationalizes intention without appealing to private mental states |
| Closeness rule | Effects too tightly causally bound to the means count as the means even if formally distinct | Blocks the Strategic Bomber dodge ("I intended only the rubble") in cases where rubble entails death |
| Threshold deontology | DDE applies up to a catastrophic threshold; beyond it, even intended harms can be permissible | Avoids the conclusion that DDE forbids stopping a genocide if doing so requires intending one death |
Counterarguments
The consequentialist challenge. Jonathan Bennett, Peter Singer, and Shelly Kagan argue the intended/foreseen distinction is morally inert. Two acts with the same outcomes, agents, and victims should be evaluated alike; mental gymnastics about which effect was "the goal" cannot turn a wrongful act into a permitted one. If the terror bomber kills 200 to save 2,000 and the tactical bomber kills 200 to save 2,000, the difference between them is purely psychological.
The redescription problem. Almost any harm can be redescribed as a side effect by selecting the right description of the act. The terror bomber can claim he intended only "to drop bombs that produce noise that demoralizes the population" with the deaths being a regrettable byproduct. Defenders of DDE respond with the closeness rule and counterfactual tests, but critics argue these patches are ad hoc.
Anscombe's reply. Anscombe argued the redescription move misunderstands what intention is. From Intention (1957): an action's intentional descriptions are constrained by the agent's reasons for acting and the structure of her plan, not by any verbal label she chooses. The terror bomber needs the deaths; that need is part of the plan; so the deaths are intended whatever he says.
The "queasy" objection. Some critics charge DDE is just a way to keep one's hands clean while permitting the same outcomes a forbidden plan would produce. Bernard Williams worried it lets agents preserve a sense of moral integrity at the cost of refusing actions that, all things considered, are right.
Common confusions
- Intention is not desire. A doctor may not desire her patient's death and may grieve over it; what matters is whether the death is part of her plan. Intention is structural, not emotional.
- Foresight is not indifference. Permitting a foreseen harm under DDE doesn't mean treating it lightly. Proportionality (condition 4) requires that the good be worth the foreseen cost; the harm is permitted, not discounted.
- DDE is not "the ends justify the means." Its main thrust is the opposite: the doctrine forbids using a person as a means to an end. Only side effects, not means, are eligible for justification by good outcomes.
- DDE doesn't always permit; it sometimes forbids. When the bad effect is too disproportionate, or the bad is the means, or the act itself is wrongful, DDE rules out actions a naive cost-benefit analysis might endorse.
- The four conditions are conjunctive. Failing any one is fatal. Many beginner mistakes (especially in medical ethics) treat the conditions as a checklist where strong scores on three compensate for failing the fourth — they don't.
Why it matters
- Just war and rules of engagement. DDE underwrites the legal distinction between combatants and civilians, between proportionate collateral damage and prohibited deliberate targeting.
- End-of-life medicine. The legality of palliative sedation in most jurisdictions rests on a DDE-shaped argument.
- Self-defense law. The "necessary force" doctrine is Aquinas's proportionality clause in modern dress.
- Trolley intuitions. DDE is the leading theoretical account of why most people judge the switch case differently from the footbridge.
- Counterterrorism debates. Targeted killings, drone strikes, and interrogation policy all turn on whether harms are means or side effects.
Frequently asked questions
What are the four conditions of double effect?
The standard formulation: (1) the act itself must be morally good or neutral; (2) the bad effect must not be the means of producing the good; (3) the agent must intend only the good effect, not the bad; (4) the good effect must be proportionate to (worth) the bad. Failing any condition makes the action impermissible. The means condition (2) does the heaviest lifting — it blocks "using" a person's harm as a tool.
Where does DDE come from?
Thomas Aquinas, Summa Theologica II-II, Question 64, Article 7 (1270s). Asking whether killing in self-defense is lawful, Aquinas writes that one act "may have two effects, of which one alone is intended, while the other is beside the intention." He concludes the defender intends self-preservation, not the attacker's death; the death is a foreseen but unintended side effect. The four-condition formulation crystallized later in Catholic moral theology and was refined by analytic philosophers in the 20th century.
How does DDE handle the trolley problem?
Switch case (divert trolley from five to one): the death of the one is a foreseen side effect of redirecting; you would prefer the track be empty. Permitted by DDE. Footbridge case (push a large person onto tracks to stop the trolley): the death is the means — you need their body's mass. Forbidden by DDE. This asymmetry matches most people's intuitions in surveys, which is why DDE is often cited as capturing a real moral distinction rather than just a philosophical artifact.
What is the closeness problem?
Critics like Jonathan Bennett argue you can re-describe almost any "means" as a "side effect" if you pick the description carefully. In the strategic-bombing vs terror-bombing case, a clever bomber could claim "I intended only the rubble; the deaths were foreseen." Defenders reply that the test is counterfactual: would you welcome a miracle in which the bad effect didn't occur? If yes, it's a side effect; if no — if the bad effect is needed for your plan to work — it's the means.
How is DDE used in medical ethics?
Palliative sedation: a doctor administering morphine to relieve a dying patient's pain may foresee that the dose hastens death. Under DDE the action is permissible because the doctor intends pain relief, not death; the death is a side effect; relief is proportionate to the foreseen risk. The same logic distinguishes legal palliative care from euthanasia in many jurisdictions. The hysterectomy of a cancerous uterus in a pregnant woman, killing the fetus, is the textbook Catholic-medical case.
What does Anscombe contribute?
Elizabeth Anscombe's "Modern Moral Philosophy" (1958) and "Mr Truman's Degree" (1956) reactivated DDE in analytic ethics. She argued the atomic bombings of Hiroshima and Nagasaki were murder, not permissible acts of war, because civilian deaths were the means of compelling surrender, not a side effect. Anscombe's Intention (1957) is the philosophical backbone: actions are individuated by what the agent does under a description, so "intention" is a substantive, non-arbitrary feature of acts.
What's the strongest objection to DDE?
Two leading challenges. First, consequentialists (Bennett, Singer) argue the intended/foreseen distinction is morally irrelevant — outcomes are what matter, and DDE produces a "queasy" double standard. Second, the redescription objection: the line between "means" and "side effect" is too pliable to do real moral work. Defenders (Quinn, Kamm) reply with refined versions — Warren Quinn's distinction between "harmful direct agency" and "harmful indirect agency" is the canonical 1989 update.