{
  "id": "double-effect",
  "title": "Principle of double effect",
  "lang": "en",
  "source_lang": "en",
  "summary": "The principle of double effect distinguishes intended harm from foreseen side effects when assessing actions that produce both good and bad consequences.",
  "category": "philosophy",
  "subcategory": "ethics",
  "type": "concept",
  "aliases": [
    "Doctrine of double effect",
    "Rule of double effect",
    "Double-effect reasoning"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/double-effect/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/double-effect/",
      "json": "https://aiwiki.page/zh/double-effect.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "The principle of double effect is a principle in ethics concerning actions that produce both a good effect and a harmful one. It holds that causing a foreseen harm can sometimes be permissible when that harm is neither the agent’s objective nor the means of achieving the good effect. Permission also depends on further conditions, particularly the moral character of the action and the existence of a sufficiently serious reason for accepting the harm. The principle therefore does not hold that good intentions excuse any harmful action.\n\nHistorically associated with Thomas Aquinas and Catholic moral theology, double-effect reasoning also occupies an important place in secular moral philosophy. It is used to examine self-defense, civilian casualties in war, medical treatment with serious side effects, and the sacrifice of one person to save others. Its central distinction—between harm that is intended and harm that is foreseen but not intended—has generated sustained disagreement about the nature of action, the limits of permissible harm, and the relationship between an agent’s purposes and other people’s rights."
    },
    {
      "heading": "Historical origins and development",
      "text": "The passage most commonly identified as the origin of double-effect reasoning occurs in Aquinas’s Summa theologiae, II–II, question 64, article 7, written during the thirteenth century. The article asks whether it is lawful to kill another person in self-defense. Aquinas argues that one action can have two effects: preserving the defender’s life and killing the attacker. The moral assessment depends on what is intended and on whether the defensive force is proportionate to the threat. An act directed toward preserving life can become unlawful if it employs greater violence than necessary.\n\nAquinas’s argument belongs to a wider account of voluntary action, natural law, and the moral significance of an action’s object, end, and circumstances. His distinction draws on an intellectual background shaped by Aristotle, particularly the analysis of choice, deliberation, and action for an end. Nevertheless, Aquinas did not present the familiar modern four-condition formula under the name “principle of double effect.” His discussion supplies a foundational example rather than a complete, separately organized doctrine applicable to every mixed-consequence action.\n\nThe self-defense passage also contains qualifications that are sometimes lost in simplified presentations. Aquinas distinguishes private defensive action from killing undertaken under public authority. In his account, a private individual may defend life but may not simply adopt killing as the intended purpose of that defense. His position therefore cannot be reduced to the contemporary claim that any intentional killing in self-defense is permissible. The historical text combines a distinction about intention with particular assumptions about authority, justice, and the preservation of life.\n\nLater theologians in the traditions of scholasticism and Catholic moral theology extended related reasoning to cases involving medical treatment, pregnancy, warfare, and other conflicts of duty. The familiar lists of conditions emerged through this later development rather than appearing as a single unchanging formula from the thirteenth century onward. Nineteenth- and twentieth-century moral manuals helped establish the principle as a recognizable method for evaluating actions with good and bad effects.\n\nDuring the twentieth century, the principle became an important topic in analytic philosophy. G. E. M. Anscombe’s work on intention and action, Philippa Foot’s discussion of abortion and conflicting duties, and Judith Jarvis Thomson’s analyses of trolley cases shifted attention toward the conceptual structure of intended and unintended harm. The principle consequently acquired a life beyond its theological origins. Contemporary discussion includes religious defenses, secular rights-based interpretations, revisions of its traditional conditions, and arguments that the distinctions it emphasizes cannot support the moral conclusions commonly drawn from them."
    },
    {
      "heading": "The traditional conditions and their logical structure",
      "text": "A common formulation gives four conditions for an action with both good and bad effects to be permissible. First, the action considered in its relevant moral description must be good or at least morally permissible. Second, the agent must intend the good effect rather than the bad effect, although the bad effect may be foreseen and accepted. Third, the good effect must not be achieved by means of the bad effect. Fourth, there must be a proportionately serious reason for permitting the bad effect. Some formulations add an explicit requirement to avoid the harm or reduce it where reasonably possible.\n\nThe first condition prevents the principle from functioning as a general excuse for wrongful conduct. An agent cannot begin with an independently prohibited act and justify it simply by pointing to beneficial consequences. Yet identifying the action’s moral character is itself difficult. Describing conduct as “saving patients,” “killing a person,” or “removing an organ” can emphasize different aspects of the same event. A satisfactory application therefore requires a defensible account of what the agent is doing, not merely a favorable label attached to the action.\n\nThe second and third conditions distinguish two ways in which harm can enter a plan. Harm may be the objective pursued, or it may be an intermediate step selected to reach some further objective. Both count as intended in standard double-effect reasoning. A person who injures a hostage to compel someone else’s cooperation cannot claim that the injury was unintended simply because cooperation, rather than suffering, was the ultimate goal. By contrast, a harmful consequence may lie outside the chosen route to the good effect even when the agent knows that it will occur.\n\nThis distinction can be represented schematically. Let \\(A\\) be the action, \\(G\\) its beneficial effect, and \\(H\\) its harmful effect. In a typical side-effect structure, \\(A\\) produces \\(G\\) and also produces \\(H\\). In a means structure, \\(A\\) produces \\(H\\), and \\(H\\) is instrumental in producing \\(G\\). The diagrammatic difference helps explain the doctrine, but it is not a complete test. Questions remain about which causal features belong to the agent’s plan and which descriptions correctly identify the relevant harm.\n\nThe conditions are normally understood as cumulative rather than interchangeable. A sufficiently important benefit does not automatically cancel the prohibition on intending harm, and lack of harmful intention does not remove the need for proportionality. Double effect is therefore a conditional permission, not a requirement to perform every qualifying action. Other duties, rights, commitments, or institutional rules may still rule an action out. The principle also does not establish that every action failing its conditions is impermissible under every ethical theory; that conclusion depends on the broader theory within which it is used."
    },
    {
      "heading": "Intention, foresight, and the description of action",
      "text": "The central distinction concerns intention rather than mere preference. To intend an outcome is ordinarily to incorporate it into a practical plan, either as an end or as a means. Foresight, by contrast, concerns what the agent expects to happen. A consequence can be highly probable, or even believed inevitable, without being adopted as part of the agent’s purpose. Conversely, an intended outcome can be uncertain: a failed attempt to injure someone remains an attempt even though the injury never occurs.\n\nIntention is also distinct from desire or emotional attitude. Someone may reluctantly intend a harmful step while disliking it intensely. A person who destroys a competitor’s equipment to obtain a contract may regret the destruction but still select it as a means. Conversely, an agent might welcome an incidental consequence without having acted in order to produce it. Such attitudes may affect the assessment of character or motivation, but they do not by themselves settle the structure of the action. Double-effect reasoning therefore requires more than the assertion that an agent “did not want” the harm.\n\nA frequently proposed diagnostic asks whether the agent would still pursue the action if the harmful effect could be prevented while preserving the benefit. An affirmative answer can suggest that the harm is incidental. However, this counterfactual test is not decisive. An agent might welcome a harmless alternative while nonetheless choosing harmful means in the actual circumstances. A coercer would prefer effortless compliance but may still intentionally inflict pain when compliance is otherwise unavailable. Hypothetical preferences cannot replace an examination of the plan actually adopted.\n\nA further difficulty concerns the level at which an intended effect is described. If an attacker plans to destroy an occupied building, is the death of its occupants necessarily intended? Some accounts distinguish destroying the structure from killing the people inside it, even where their deaths are virtually certain. Critics argue that excessively narrow descriptions can separate aspects of an action that are practically inseparable. The resulting problem is sometimes called the problem of closeness: when is a harmful consequence too closely connected with the chosen act to count as merely incidental?\n\nThese issues connect double effect with philosophy of mind and the philosophy of action. Intention is not normally treated as an inaccessible inner declaration that overrides all contrary evidence. Planning documents, instructions, selected methods, responses to alternatives, and the role a consequence plays in achieving the objective can all bear on its identification. Nevertheless, different theories of action can interpret the same evidence differently. The principle’s practical reliability therefore depends partly on a theory of intention that neither equates knowledge with purpose nor permits agents to redescribe deliberately chosen harm out of existence."
    },
    {
      "heading": "Proportionality, alternatives, and responsibility",
      "text": "The proportionality requirement addresses a question different from intention: even if harm is incidental, is accepting it justified? A small convenience cannot ordinarily justify a grave injury merely because the injury is unwanted. Relevant considerations include the importance of the benefit, the seriousness and likelihood of the harm, the availability of less harmful alternatives, and the agent’s obligations toward those affected. The assessment concerns the concrete circumstances, not an abstract comparison between an unspecified good and an unspecified evil.\n\nProportionality in double-effect reasoning does not necessarily mean maximizing aggregate welfare. A utilitarian analysis may compare options by their expected contribution to overall well-being. A defender of double effect can instead hold that certain actions are excluded before consequences are compared and that proportionality governs only the remaining options. The principle thus combines constraints on the means of action with an assessment of consequences. Its place within deontology reflects this refusal to let a sufficiently large benefit automatically justify every method of obtaining it.\n\nAlternatives matter because an avoidable harm is harder to justify than an unavoidable one. If two equally effective actions impose substantially different risks on innocent people, the availability of the safer action weakens the case for choosing the more harmful one. However, formulations differ about the strength of this requirement. Some demand necessity or the least harmful feasible means; others incorporate avoidability into a broader proportionality judgment. An alternative may also be less effective, more uncertain, slower, or harmful to different people, making comparisons more complex than simply identifying a harmless option.\n\nUncertainty introduces another layer. Agents frequently choose under incomplete information about causal pathways and outcomes. Probability and expected value can help represent anticipated benefits and losses, but numerical estimates do not settle questions about rights, permissible means, or the distribution of risk. Nor does a favorable eventual outcome necessarily vindicate an unreasonable decision. Conversely, a justified decision can have a disastrous outcome. Assessment of the decision must therefore distinguish what actually happened from what could reasonably have been anticipated when the action was chosen.\n\nThe absence of harmful intention also does not eliminate moral responsibility. Agents can be responsible for negligent, reckless, disproportionate, or avoidable side effects. A person who knowingly exposes others to severe danger for a trivial benefit cannot appeal to lack of intent as a complete defense. This point distinguishes double effect from the much weaker claim that unintended harm is morally irrelevant. Its defenders instead maintain that intention is one morally significant feature among several, while acknowledging that foreseen harms continue to require explanation, restraint, and, where appropriate, accountability."
    },
    {
      "heading": "Trolley cases and the limits of causal tests",
      "text": "The trolley problem became a central testing ground for double-effect reasoning after Philippa Foot’s 1967 essay “The Problem of Abortion and the Doctrine of the Double Effect.” In a familiar version, a runaway trolley threatens five people, and an agent can divert it onto another track where it will kill one person. The case invites comparison with situations in which saving five requires deliberately killing one. Its purpose is not simply to solicit a numerical preference but to isolate features that might explain different judgments about otherwise similar outcomes.\n\nIn the standard diversion case, the one person’s death appears to be a side effect of redirecting the threat. If that person escaped, the diversion would still save the five. By contrast, in the footbridge case, associated with Judith Jarvis Thomson’s work, a bystander could stop the trolley by pushing a large person into its path. Here the person’s physical presence is used to stop the vehicle. Defenders of double effect often explain the difference by saying that the first action redirects a threat while the second incorporates serious harm to a person into the rescue plan.\n\nThe explanation immediately encounters a refinement. In the footbridge case, it may be the person’s body, rather than the person’s death, that stops the trolley. If the person survived, the plan could still succeed. It follows that a test focused only on whether death is causally necessary may miss the relevant issue. The intended means may instead be forcibly placing someone in a position of violent impact. The moral object under discussion is then not merely death but an imposed bodily invasion or harmful use of a person.\n\nLoop-track cases create another difficulty. Suppose a diverted trolley will travel around a loop and return toward the five unless it is stopped by striking the one person on the alternate track. The diversion now appears to rely on that person’s presence. Some judgments remain more permissive than in the footbridge case, despite the changed causal structure. Such examples suggest that intuitions may also respond to redirecting an existing threat, initiating force against someone, bodily contact, or the distinction between doing harm and allowing it.\n\nAs thought experiments, trolley cases reveal the need to separate these factors rather than treating any single response as proof of a doctrine. A rights-based theory might condemn using an uninvolved person as a resource without making intention the fundamental explanation. A consequentialist might reject the standard contrast altogether or explain intuitive resistance through the wider costs of permitting such conduct. The cases are most informative when their assumptions are explicit: who owns the threatened resources, whether rescue duties apply, what alternatives exist, and which causal facts the decision-maker knows."
    },
    {
      "heading": "Medical applications and disputed examples",
      "text": "Medical ethics has supplied many influential examples of double effect because beneficial interventions can impose serious burdens or risks. The traditional illustration concerns treatment intended to relieve severe pain while carrying a risk of shortening life. Double-effect reasoning distinguishes symptom relief accompanied by a harmful side effect from administering a substance in order to cause death. The distinction concerns the purpose and structure of the intervention, not simply the physical identity of the medication or the fact that death follows treatment.\n\nThis illustration must be separated from clinical claims about what properly administered treatment actually does. The ethical example assumes a life-shortening risk; it does not establish that ordinary pain relief necessarily shortens life. In palliative care, the rationale, dose, timing, monitoring, and intended therapeutic effect matter to the interpretation of an intervention. Philosophical discussion can become misleading if it treats every use of strong analgesia near the end of life as an instance of knowingly accelerating death.\n\nPalliative sedation raises related questions because its intended immediate effect includes a reduction of consciousness to relieve otherwise refractory suffering. The relevant distinction is not between intending every physiological change and intending none. Rather, the analysis asks which effects are selected as therapeutic means, which burdens are accepted, and whether any additional harmful consequences are proportionate. Informed consent and patient autonomy introduce independent requirements. A treatment does not become permissible merely because its harmful consequences are unintended, and consent does not automatically justify every way of bringing about death.\n\nCatholic discussions of pregnancy have also used double-effect reasoning, especially in distinguishing different interventions for ectopic pregnancy. A common analysis treats removal of a pathologically affected fallopian tube as directed toward removing dangerous tissue, while embryonic death is considered an unavoidable consequence. The classification of other interventions, including those that act more directly on embryonic or trophoblastic tissue, has generated disagreement within Catholic bioethics. These debates concern how the act and its means are identified; they should not be presented as settled descriptions of either medical necessity or universally accepted moral distinctions.\n\nDecisions to withhold or withdraw treatment require additional analysis. An intervention may be refused because it is excessively burdensome or lacks proportionate benefit, even when death is foreseen. Whether that decision involves an intention to cause death cannot be inferred simply from the omission of treatment. The role of euthanasia, assisted dying, refusal of treatment, and symptom relief varies across ethical frameworks and legal systems. Double effect can clarify one part of the assessment, but it cannot by itself establish a diagnosis, determine treatment effectiveness, identify valid consent, or resolve every question about professional obligations."
    },
    {
      "heading": "Warfare, public authority, and legal analogues",
      "text": "In just war theory, double effect is commonly invoked to distinguish an attack on a military objective that foreseeably harms civilians from an attack that uses civilian suffering as its means or objective. The standard contrast is between striking a military installation despite anticipated civilian casualties and attacking civilians to terrorize a population into surrender. Even if both actions seek an eventual military advantage, civilian harm plays a different role in the respective plans.\n\nThis distinction imposes more than a requirement that military personnel declare a permissible goal. If civilian deaths are chosen to break morale, compel displacement, or exert pressure, they cannot be reclassified as side effects simply because the ultimate objective is victory. Conversely, the fact that a commander anticipates civilian casualties does not alone establish that those casualties are intended. Identifying the operative plan requires attention to target selection, weapons, timing, available intelligence, and the causal role assigned to the civilian harm.\n\nInternational humanitarian law addresses related problems through rules on distinction, proportionality, and precautions in attack. These rules are not identical to the philosophical principle of double effect. Article 51(5)(b) of Additional Protocol I of 1977 describes as indiscriminate an attack expected to cause incidental civilian harm excessive in relation to the concrete and direct military advantage anticipated. Article 57 addresses precautions, including verification of targets and measures to avoid or minimize incidental harm. These provisions establish legal requirements rather than merely restating a test of private moral intention.\n\nThe difference matters because legal assessment does not turn exclusively on whether an attacker regards civilian casualties as unintended. An attack may violate applicable rules because it is indiscriminate, disproportionate, or conducted without required precautions. A claimed beneficial objective does not remove those failures. Equally, the legal meaning of a military objective and the relevant standards of evidence must be established through the governing law, not derived solely from a philosophical account of intended and foreseen effects.\n\nDomestic criminal law contains another partial analogue in mens rea, the mental element of an offense. Legal systems distinguish among intention, knowledge, recklessness, and negligence, but their definitions vary. In some contexts, awareness that a result is virtually certain may support a legal finding of intention even where the result is not the actor’s desired objective. There is therefore no general legal exemption called “double effect” that follows automatically from the philosophical doctrine. Its distinctions may illuminate legal reasoning, while responsibility, defenses, evidentiary standards, and institutional authority remain governed by additional rules."
    },
    {
      "heading": "Philosophical objections and competing explanations",
      "text": "A major objection asks why intention should change the permissibility of otherwise equivalent harms. If two actions foreseeably kill the same number of people and save the same number, critics argue that the victims’ losses do not become smaller because one agent incorporates death into a plan while the other treats it as incidental. On this view, intention may affect blameworthiness or character without determining whether the action itself is permissible. The objection challenges a central claim of double effect: that the relation between an agent’s plan and a harmful outcome can alter what may be done.\n\nA related criticism distinguishes the wrongness of conduct from the moral assessment of the person performing it. A beneficial rescue might remain permissible even if the rescuer acts from hostility toward someone who will incidentally be harmed. Conversely, an admirable motive cannot make an unjustified attack permissible. Defenders respond that the doctrine is not fundamentally about benevolent feelings. It concerns what is chosen as an end or means, which helps constitute the action itself. The disagreement therefore depends partly on whether intention merely accompanies an action or helps identify its morally relevant nature.\n\nAnother objection is that action descriptions can be manipulated. Almost any harmful means might be redescribed as a neutral bodily movement followed by unfortunate consequences. This is sometimes called the problem of redescribing or “closeness” in double-effect reasoning. A credible defense needs constraints on permissible descriptions: the agent cannot exclude from the plan a harmful intervention that explains how the benefit is to be obtained. Yet attempts to specify those constraints may rely on independent judgments about rights, bodily integrity, or wrongful use, raising the question of how much explanatory work double effect itself performs.\n\nConsequentialist approaches generally deny that intended harm is intrinsically less permissible than equally serious unintended harm, although they may recognize the practical value of rules against intentional injury. Rights-based approaches may instead explain the relevant cases through claims against assault, coercion, or being used without consent. Virtue ethics can emphasize practical judgment, character, and the quality of deliberation rather than a fixed list of conditions. These alternatives can converge on particular verdicts while disagreeing about their underlying explanation.\n\nThe empirical study of moral judgment adds a further question: whether ordinary responses reliably track the intention–foresight distinction or instead respond to factors such as personal force, emotional salience, proximity, and perceived responsibility. Such findings can test claims about how people judge cases, but they do not independently establish which judgments are justified. Contemporary philosophical work therefore continues to examine whether double effect is an independent moral principle, a useful expression of deeper constraints on harming others, or an imperfect heuristic whose apparent successes are better explained by other features of action."
    }
  ],
  "links": [
    "ethics",
    "thomas-aquinas",
    "natural-law",
    "aristotle",
    "scholasticism",
    "analytic-philosophy",
    "counterfactual",
    "philosophy-of-mind",
    "utilitarianism",
    "deontology",
    "probability",
    "expected-value",
    "trolley-problem",
    "thought-experiment",
    "palliative-care",
    "informed-consent",
    "patient-autonomy",
    "international-law",
    "consequentialism",
    "virtue-ethics"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}