Legal Causation

The Question of Causal Attribution in Law

Legal causation concerns the conditions under which an event, action, or omission is treated as causally connected to a consequence that matters for legal responsibility. Consider a driver who violates a safety rule, a pedestrian who is injured, and a second driver whose conduct also contributes to the collision. A factual history may establish that each driver changed the course of events. The law must then identify the injury relevant to the claim, the conduct relevant to the legal prohibition, and the connection sufficient to make the defendant answerable for that injury. These three selections cannot be made from a chronological narrative alone. They require a theory of causation, an account of the relevant legal norm, and rules governing proof.

The fundamental question is therefore precise: what makes an outcome attributable to a legally relevant act? An answer must distinguish the occurrence of a consequence from its description, a cause from the conditions that made it possible, a causal explanation from the evidential demonstration of that explanation, and causal involvement from the reasons for imposing a legal remedy or punishment. The inquiry is central to tort and criminal law, appears differently in contract and public law, and reaches beyond doctrine into metaphysics, epistemology, and moral philosophy. It is a philosophical problem because disputes persist even when the events are agreed: two sufficient fires destroy one house; an injured patient refuses treatment; a third person deliberately exploits a hazard; a risk-generating exposure is proved but the exposure that produced an illness cannot be identified.

The major tradition of Anglo-American discussion examines factual causation and the limits of legal responsibility together while disagreeing on whether these are distinct questions. Hart and Honoré defend the explanatory importance of ordinary causal concepts, including abnormal conditions and voluntary interventions. Moore defends causation as a genuine relation connecting acts to effects and argues that the law's many policy considerations should be separated from that relation. Wright reconstructs factual causation through necessary elements of sufficient sets. Stapleton insists that the task for legal theory is to specify the precise information conveyed by the term and to keep questions of historical involvement apart from the scope of responsibility (Hart & Honoré, 1985; Moore, 2009; Wright, 1985; Stapleton, 2008).

The argument of this monograph develops a layered account of legal attribution. Causal involvement is a relation to be investigated with the best available evidence; the identification of the legally relevant risk and consequence is governed by the substantive norm; and liability depends on further conditions, including fault where required, defenses, burdens of proof, and institutional choices about remedies. The layers are connected because law determines which causal question needs an answer. They remain analytically distinguishable because an action can be a cause without generating liability, and a legal system can exceptionally allocate loss under a risk-based rule where individualized causation remains unproved.

Definition, Scope, and the Objects of the Inquiry

The expression legal causation is used both narrowly and broadly. In a narrow doctrinal sense it often designates proximate cause, remoteness, or scope of liability: an evaluative limit placed on consequences for which an actor may be held liable after factual causation has been established. In a broader jurisprudential sense it designates the whole legal inquiry into the connection between conduct and consequence, including causal facts, selection of a relevant event, intervening acts, and the justification of responsibility. The present investigation uses the broad sense and marks the narrow sense explicitly. This prevents a verbal ambiguity from settling substantive questions before they have been examined (Moore, 2024).

A causal claim has relata: the things said to stand in the relation. In litigation these may be actions, omissions, particular events, changes in conditions, bodily injuries, financial losses, risks, or descriptions of those phenomena. Saying that an industrial exposure caused mesothelioma makes a claim about a disease in an individual. Saying that exposure increased the risk of mesothelioma makes a claim about a probabilistic change. Both may be warranted; they have different truth conditions. Likewise, the claim that a driver caused a collision differs from the claim that the driver's manner of driving caused death. The description of the result matters because the law attaches significance to specified harms rather than to every subsequent state of the universe.

Four questions organize the subject. First, did the relevant conduct make a causal difference to the actual consequence? Second, what causal account explains its contribution, including cases where several factors were sufficient? Third, is the consequence within the range of harms or interests addressed by the operative legal duty, offense, or statutory prohibition? Fourth, has the requisite connection been established under the rules of proof appropriate to the proceeding? Answering the fourth question cannot replace answering the first: a judgment based on a legal presumption may establish liability without showing that the court discovered the actual physical process that produced the harm.

Legal causation thus has an ontological dimension, because the world contains events and processes that may be causally related; a semantic dimension, because legal propositions individuate conduct and damage; an epistemic dimension, because causal history is inferred from evidence; and a normative dimension, because legal institutions choose which consequences warrant attribution and remedy. The central methodological task is to explain their interaction rather than derive a single all-purpose test from the ambiguous word cause.

Historical Foundations: From Causes to Juridical Responsibility

Reflection on causation long precedes modern liability law. Aristotle's analysis of material, formal, efficient, and final causes in the Physics provided a vocabulary for explanation (see Physics, Book II, §§3 and 7); efficient causation subsequently became especially important to discussions of physical production. His framework does not supply a ready-made doctrine of negligence. Its value is to show that questions framed as “why?” can demand different kinds of answer. The purpose of legal compensation, the physical mechanism of an injury, and the description of the act are explanatory questions at different levels. Translating every Aristotelian aitia into a modern sufficient-condition test would suppress these distinctions.

The Roman legal tradition developed forms of liability for damage, including the lex Aquilia and later juristic interpretation of wrongful loss, that made the relation between conduct and damage a practical question of classification (see Digest, 9.2, Ad legem Aquiliam). Roman doctrine was shaped by procedural forms and legal actions; its categories should not be read as a fully articulated modern theory of counterfactual dependence. Medieval analysis of voluntary conduct, intention, foreseeability, and responsibility continued to inform legal thought through moral theology and theories of agency. The historical significance is a gradual differentiation of the consequences an agent brings about, those the agent intends, and those for which a legal community calls the agent to account.

In early modern philosophy, Hume's analysis of necessary connection shifted attention toward regularity, experience, and the inferential character of causal judgment. His Enquiry Concerning Human Understanding, especially Sections IV and VII, investigates what warrants an expectation that one event will follow another. Hume's problem is epistemological and metaphysical: the relation is not established simply by perceiving two successive events. Litigation faces a related difficulty when a causal process cannot be directly observed and must be reconstructed through testimony, experiments, and statistical evidence (Hume, 1748, §§IV, VII).

John Stuart Mill's A System of Logic (1843), especially Book III, Chapters V–VIII, analyzes the law of causation, the composition of causes, and methods of experimental inquiry. Mill treats effects as arising within complexes of conditions and seeks methodological rules for identifying causally relevant factors. Modern lawyers frequently ask whether one action was a necessary condition of an outcome, but Mill's analysis warns against imagining that a complex effect possesses one solitary antecedent. A legal demand for the cause already presupposes a relevance selection within a field of conditions (Mill, 1843, Book III).

The modern law of negligence, industrial injury, mechanized transport, and complex markets brought these questions into sharp focus. Multiple actors and long sequences of consequences made a simple rule of direct physical contact inadequate. By the twentieth century courts had developed tests of factual causation alongside debates about proximate cause, remoteness, intervening acts, and foreseeable harm. The resulting doctrine arose within adjudication and legal policy, and its philosophical reconstruction must preserve that institutional setting.

The Classical Modern Synthesis of Hart and Honoré

H. L. A. Hart and Tony Honoré's Causation in the Law, first published in 1959 and substantially revised in a second edition in 1985, is a defining attempt to explain legal causal reasoning through the conceptual structure of ordinary causal explanation. Their thesis resists treating every judicial judgment about causation as an unarticulated policy choice. Ordinary causal explanations distinguish a normal background from something exceptional: a fire is traced to the dropped match, while oxygen is treated as a standing condition; a flood is explained by a breached levee rather than by the mere presence of gravity. Such selections convey information about a change in a situation, although selection for explanatory prominence differs from the totality of the causal history (Hart & Honoré, 1985).

The distinction between cause and condition is context-sensitive. Oxygen may itself be the unusual factor in a laboratory chamber. Normality therefore cannot mean that the environment is causally irrelevant. It means that an explanation directed to a particular practical question selects one factor against an assumed field. The philosophical value of Hart and Honoré's approach lies in recognizing why legal disputes frequently focus on abnormal interventions, culpable acts, and unexpected developments. A court answers a practical causal question for a specific norm; it does not record the entire state of nature.

Their account of voluntary human intervention is particularly influential. Independent, deliberate, informed action can become a reason to treat a subsequent harm as attributable to the intervening actor rather than the earlier actor, although the legal treatment depends on the purpose of the duty and the relevant relationship. The House of Lords expressly invoked Hart and Honoré when discussing autonomous self-administration of drugs in R v Kennedy (No 2) ([2007] UKHL 38, para. 14](https://www.bailii.org/uk/cases/UKHL/2007/38.html)). The precedent demonstrates both the influence and the bounded character of their theory: what breaks the chain in one criminal offense cannot be presumed to break it in a statutory pollution regime or a duty specifically designed to prevent the very intervention that occurred.

The principal objection is that explanatory prominence and legal responsibility do not always track one another. Abnormality may explain why an event attracts attention, but it does not by itself explain whether a party should bear a loss. A rare, harmless background condition may be extraordinary without being blameworthy; routine institutional practices may cause harm even though they are normal. Hart and Honoré supply a richer grammar of causal selection than an unqualified but-for test, while the law still requires a distinct account of why one selection has legal force.

Counterfactual Dependence and Its Reach

The familiar but-for inquiry asks whether the injury would have occurred if the defendant's conduct had been absent, with relevant background conditions held suitably fixed. Its appeal comes from the direct comparison between an actual sequence and a counterfactual alternative. If a driver passes through a red light and strikes a pedestrian who would otherwise have crossed unharmed, the driver's movement makes a difference to the injury. But-for reasoning is particularly tractable when a specific mechanism is known, one candidate factor is decisive, and the comparison can be described without changing the character of the whole situation.

David Lewis's 1973 analysis of causation via counterfactual dependence placed this intuition in a more systematic metaphysical framework. A causal chain may connect events even where the effect does not depend counterfactually on a single immediately preceding event; relations among possible alternatives are central to assessing dependence. The framework also exposes the need to specify how alternative worlds are held similar to the actual world. A lawyer comparing a negligent driver with a competent driver must say what replaces the negligent act: careful driving along the same road, no journey, or a radically different series of events? The selection may affect the result (Lewis, 1973).

Legal counterfactuals often make an implicit normative substitution. If an employer failed to install ventilation, the relevant contrast is usually compliance with the duty, not the employer's total disappearance. If a surgeon failed to warn, the hypothetical question concerns a properly informed patient and what that patient would have chosen. Such comparisons cannot be constructed solely from physical science because the legal norm helps determine the alternative conduct. This does not make the historical causal facts matters of legal fiat. It means that the law asks a focused causal question about what compliance would have changed.

The but-for test also generates false negatives in overdetermination. Two independently sufficient fires simultaneously destroy one house. Remove either fire and the house still burns. Applied separately, the strict necessity test calls neither fire a cause, although each was capable of producing the destruction and both contributed to the actual history. This defect is not repaired by simply announcing that judges may disregard the test when the result seems unfair. A philosophical theory must explain why the circumstances are structurally different and offer a causal criterion appropriate to them.

Mackie's INUS Conditions and the Anatomy of a Causal Complex

J. L. Mackie described a cause in many ordinary cases as an insufficient but necessary part of an unnecessary but sufficient condition, commonly abbreviated INUS. A short circuit may be insufficient to produce a fire without combustible materials and oxygen; within one sufficient complex it is necessary, while another complex—such as a deliberate flame in suitable surroundings—might independently have produced a fire. The analysis captures why one factor can be correctly identified as a cause even though it was neither sufficient by itself nor necessary across all possible ways the effect could occur (Mackie, 1965).

The distinction between local necessity within a complex and unconditional necessity for an outcome matters for law. Suppose a defective heating element, dry furnishings, and an open door combine to start a blaze. The defective element can be a crucial component of the sufficient mechanism that actually operated even though other possible mechanisms existed. Legal investigation should therefore reconstruct which factors were actually present and how they jointly produced the damage before asking whether a factor would have been indispensable in a different world. The method places a premium on mechanistic evidence rather than loose appeals to probability.

INUS analysis does not itself settle which conditions count as sufficient or how finely events should be described. A set defined as sufficient only because it includes the outcome would be circular. Likewise, if background conditions are arbitrarily specified, every factor can be made necessary within a contrived set. The adequacy of a causal complex depends on independent knowledge about how its components combine to produce the effect. Physical experiments, medical evidence, engineering analysis, and credible behavioral models can constrain the selection.

A further difficulty arises when actual causal processes preempt one another. One would-be cause prevents an alternative sufficient process from operating; the alternative remains a threat but never contributes to the result. Simply identifying two sufficient plans would exaggerate actual causal contribution. Legal theory needs a distinction between a factor that participated in the productive history and one whose mere potential sufficiency is known after the fact. This is the central task addressed by necessary-element accounts and by their critics.

Wright's NESS Theory and Multiple Sufficient Causes

Richard W. Wright's NESS account analyzes a cause as a necessary element of a sufficient set of actual antecedent conditions. In a scenario of duplicative causation, each of two independently sufficient contributions may be necessary within a particular sufficient set even though neither is necessary for the outcome considered against all actual contributions together. This revises the rigid but-for test and attempts to preserve an objective distinction between causal connection and ultimate responsibility. The landmark presentation is Wright's 1985 study of tort causation, which also critiques the indeterminate use of the label substantial factor (Wright, 1985).

A proper NESS argument must specify the sufficient set and demonstrate that its members actually operated. Imagine two factories release independently sufficient amounts of a toxic substance into a river at the same time, and the combined exposure causes one discrete fish kill. It may be possible to identify for each discharge a sufficient combination that includes it as a necessary element. The contribution of each discharge can then count as causal even though removing just one would leave enough pollution to kill the fish. Conversely, a factory that held chemicals in storage but released none cannot be deemed a cause merely because a possible spill would also have been sufficient.

The strength of NESS is explanatory discipline: it asks which actual conditions constitute an outcome-producing complex. The weakness lies in the burden placed on the notions of sufficiency, actual contribution, and event individuation. Real events are frequently continuous, probabilistic, and temporally extended. In epidemiology or complex systems, a determinate minimal sufficient set may be unavailable. A highly abstract causal model risks hiding these uncertainties behind logical form. Wright's contribution is therefore best understood as an analytical strategy with demanding empirical prerequisites, not as a mechanical substitute for evidence.

NESS and but-for reasoning may converge in straightforward single-factor cases while yielding different analyses in overdetermination. A legal system need not choose one test as universally exhaustive. It needs to disclose when factual contribution is established, when the causal mechanism is only probabilistically supported, and when a special allocation rule responds to an evidential gap. Treating every difficult case as proof that causation is simply a judicial value judgment would lose information that the theory is capable of preserving.

Preemption, Duplication, and the Difference Between Presence and Production

Overdetermination is often presented through simple symmetrical examples: two fires reach one house together. Preemption has a different structure. One actor initiates a process that produces the harm before a second actor's independently harmful plan can operate. A poisoning causes death before a second poisoned drink is consumed; a shot kills before another assailant can fire. In the latter scenarios the thwarted alternative may explain why the victim was in danger but cannot be treated as actually producing the death in the same way as the effective process.

These cases force distinctions among causal potential, actual causal participation, and counterfactual dependence. When a sufficient process is preempted, absence of the first process might permit the backup process to cause the same type of result. A but-for test focused only on the coarse-grained outcome “death” will then say that the first process was unnecessary. A more specific description of the actual event or its time and manner may expose a difference, but overly fine descriptions can make nearly every antecedent necessary. The theoretical burden is to justify the level of individuation from the explanatory and legal purpose of the inquiry rather than choose it opportunistically.

An adequate account also distinguishes concurrent contributions from serial events. When two sufficient fires merge, it may be meaningless to assign a unique physical share of the final burn. When one fire arrives after the house has already been reduced to ruins, its arrival explains no part of the earlier destruction. In either case the evidentiary question concerns the temporal history and mechanism of the occurrence. Legal doctrine should not confuse uncertainty about causal apportionment with proof that no causal contribution occurred.

The jurisprudential implication is that factual causation is relational and temporally structured. The question is not whether the defendant was dangerous in general or whether an imagined similar act could have caused damage. It is whether the conduct participated in producing the legally described actual harm. That criterion protects defendants against liability for merely possible harms while leaving room for responsibility in cases involving more than one sufficient actual contributor.

Causal Explanation, Intervention, and Structural Models

Interventionist approaches test causal relations by asking how changing one variable through an appropriate intervention would change another. James Woodward's Making Things Happen provides a systematic theory of causal explanation organized around intervention and invariance. A legal inquiry can benefit from that framework when it assesses whether altering a precaution, warning, machine setting, or institutional practice would change an injury-producing process. The framework specifies a disciplined counterfactual question without presuming that every useful intervention is physically available in court (Woodward, 2004).

A causal model might represent a defective guard, a worker's exposure, and a resulting injury as connected variables. It can help identify mediating processes, alternative pathways, and confounding factors. But a model is an artifact of its variables and assumptions. If it represents “breach” as a binary value while ignoring the diverse ways an actor might have complied, the model may obscure the very counterfactual choice a legal duty requires. Similarly, an intervention on the plaintiff's conduct can be causally illuminating but normatively irrelevant if the defendant's duty was to protect against foreseeable plaintiff behavior.

Structural models are particularly useful in distinguishing causal questions from evidential ones. The model states what a proposed system would imply under an intervention; the evidence determines which model, if any, is warranted. Competing models may fit the same observed data and differ in their answers to an unobserved counterfactual. Litigation then faces a question of identification. A mathematically definite answer can still depend on uncertain premises; the expression of a probability does not remove uncertainty about which causal mechanism actually operated.

The lesson for jurisprudence is methodological pluralism disciplined by an explicit target. But-for comparisons, sufficient-set accounts, interventions, and causal-process descriptions are investigative instruments. Their value depends on how well they represent the event and counterfactual the law actually needs to know. In a complex case the appropriate result may be a statement of evidential uncertainty followed by a legally authorized rule allocating that uncertainty, rather than an inflated claim that the mathematics has resolved every metaphysical issue.

Causal Proof, Standards of Evidence, and Statistical Inference

A theory of causal truth and a rule of judicial proof operate at different levels. An event either had a particular causal history or it did not; a court must reach a decision despite limited access to that history. Civil law commonly employs a balance-of-probabilities standard, while criminal conviction generally requires proof beyond reasonable doubt, expressed through the applicable jurisdiction's instructions. Those standards govern warranted legal findings. They do not describe different physical relations between an act and an injury. When two courts applying different standards reach different legal conclusions about the same disputed causal history, the world has not changed because the litigation did.

Statistical evidence creates a recurring ambiguity. A study may establish that exposure to a substance multiplies the population risk of a disease. It does not automatically identify which exposure produced an individual case. An increase in risk may be powerful evidence of causal involvement under additional assumptions, especially where mechanisms, background rates, and alternative explanations are known. Yet the inference from relative risk to individual causation depends on the study population, the individual's exposure, possible confounding, and a defensible reference class. The same numerical association can support very different conclusions when underlying causal structures differ.

This uncertainty is acute in toxic tort litigation and medical negligence. The traditional but-for test asks whether the claimant would have experienced the injury in the absence of the defendant's breach. In a disease with a latency period and several possible exposures, no witness may be able to establish the counterfactual history directly. The law must decide whether to leave the ordinary proof burden in place, recognize a narrowly confined evidential presumption, impose proportionate liability, or articulate a special risk-based rule. Such alternatives allocate the consequences of epistemic limitations; they should be justified as legal choices rather than redescribed as discoveries about what physically occurred (Stapleton, 2003).

An epistemically responsible judgment therefore distinguishes three statements: the defendant's conduct is known to be a cause; the evidence supports a specified probability that it was a cause; and the law assigns responsibility on some other specified ground. The third may be justified by deterrence, corrective justice, the distribution of evidential burdens, or the purpose of a protected duty. Its justification should be argued on those terms. This transparency protects the conceptual integrity of both causal science and legal adjudication.

The Structure of Negligence: Duty, Breach, Cause, and Scope

In negligence law, causation operates within a larger set of conditions. A claimant must ordinarily establish a relevant duty, breach, harm, a connection between breach and harm, and the extent to which that harm falls within liability. The architecture is important because failures of attribution may arise at different places. A manufacturer may have caused an injury without breaching an applicable duty; a professional may have breached a duty without causing the complained-of loss; an injury may have been caused by a breach yet fall outside the risk that made the conduct negligent.

To ask whether negligence caused harm requires a more refined counterfactual than asking whether the defendant's body or machine caused harm. Negligence is an evaluative description of conduct. The relevant question ordinarily compares the actual conduct with a legally appropriate alternative: a careful driver traveling at an appropriate speed, a hospital providing the required treatment, or a business maintaining a safety system. This entails identifying the precaution that the duty required. The answer may differ if the injury would have happened even with that precaution. The causal relation is thus investigated under a legally selected contrast; the evidence for its existence remains answerable to facts.

A distinction between the factual contribution of conduct and the scope of liability avoids circularity. If a negligence rule prohibits leaving a pit unguarded to protect pedestrians from falling, a later injury caused by a completely unrelated mechanism is outside the rationale of that rule even if the defendant's conduct influenced the person's route. Conversely, when the very danger protected against materializes through a complex sequence, its complexity alone does not eliminate responsibility. The law must connect the character of the breach to the kind of harm that occurred, not demand a perfectly foreseen physical script.

Jane Stapleton's analysis of cause-in-fact and scope of liability demonstrates the importance of separating factual involvement from the normative limits of the tort obligation. The categories are connected in reasoning but perform different work. The law becomes more intelligible when it specifies whether the difficulty is proving causal involvement, determining the relevant duty, identifying the protected risk, or limiting the measure of recovery (Stapleton, 2001; Stapleton, 2003).

Palsgraf and the Location of Foreseeability

Palsgraf v. Long Island Railroad Co. (1928) remains a philosophical test case because it makes visible the difference between a causal chain and the reach of a legal duty. Railroad employees assisted a passenger carrying an apparently innocuous package. The package fell, fireworks exploded, and a scale at a distance injured Helen Palsgraf. The parties' dispute could be described as a dispute about remoteness. Cardozo's majority opinion instead grounded the analysis in the absence of a duty of care extending to a plaintiff outside the reasonably apprehended zone of danger. Andrews's dissent supported a more expansive understanding of duty and focused on whether a proximate causal connection should limit liability in the circumstances (Palsgraf, 248 N.Y. 339 (1928)).

The disagreement cannot be reduced to the physical trajectory of the falling package. Both positions can accept an extended causal sequence from the employees' conduct to the injury. Their difference concerns where legal limitations should be located and how they should be justified. A duty-centered theory asks whether the defendant owed the claimant a relevant obligation of care at the time of acting. A proximate-cause theory may recognize a general duty and limit the consequences of breach after factual causation has been established. These approaches can yield similar outcomes while assigning the governing normative work to different doctrinal elements.

The philosophical issue is one of legal architecture. If foreseeability is placed at the duty stage, it helps define the protected claimant and interest. If it is placed at the scope-of-liability stage, it helps define which consequences of a breach are attributable to the defendant. The two placements can matter for questions of adjudicative responsibility, jury instructions, the development of precedent, and the role of policy. The same word causation should not conceal a dispute about the structure of the underlying norm.

Palsgraf also shows why the outcome's level of description matters. One might foresee injury to a person near a moving train but not an injury to a distant bystander through an unexpected explosion and a falling scale. Whether the relevant risk is described broadly as “injury on the platform” or narrowly as “injury to this claimant from the hidden contents of a package” can determine the scope of duty. The description must be governed by legal principle rather than manufactured to make the desired judgment inevitable.

Remoteness, Foreseeability, and the Harm-Within-the-Risk Principle

Remoteness doctrines limit the range of recoverable consequences once some causal connection has been shown. In the common-law negligence tradition, The Wagon Mound (No. 1) ([1961] A.C. 388) marks the rejection, in its context, of directness as a sufficient basis for recovery for an unforeseeable kind of injury (The Wagon Mound (1961)). Jolley v. Sutton London Borough Council later explained that foreseeability does not require precise anticipation of the mechanism by which a foreseeable kind of harm occurs; the relevant description concerns the nature of the risk rather than every detail of its realization (Jolley, [2000] UKHL 31).

The harm-within-the-risk approach asks whether the injury belongs to the class of risks that made the defendant's conduct wrongful. Suppose a contractor negligently leaves an unsecured trench beside a walkway. A passerby falls into it after a sudden distraction. The intervening distraction need not defeat responsibility because the hazard of falling is precisely the concern of the precaution. By contrast, if the trench merely delays a passerby who is later injured by a wholly independent event far away, the counterfactual role of the trench in changing the person's schedule provides a weak reason for liability.

Foreseeability, however, must not become an unexamined substitute for causation. A foreseeable injury can occur for reasons entirely unrelated to the defendant's conduct. A causal contribution can produce an unforeseeable mechanism that nevertheless realizes the risk against which the duty protects. The concepts answer different questions. Causation identifies a historical connection; foreseeability can help define the reasonable precaution and the protected kind of harm; scope of liability determines whether the realized harm is attributable under the norm. Treating foreseeability as a test of physical production would conflate these roles.

An adequate doctrine must also preserve the distinction between the kind of harm and its extent. A claimant with unusual vulnerability may suffer more serious consequences from an injury than an average person. The eggshell-skull principle, as developed in common-law systems, limits the use of unusual susceptibility to defeat responsibility once a relevant injury has been caused. The rationale is tied to taking the victim's existing condition as encountered, not to a metaphysical claim that every later development is part of one undifferentiated cause. The boundary between scope and quantum remains crucial.

Multiple Causes, Divisible Injury, and Damage Allocation

Several agents may jointly contribute to one harm, produce separate harms, or create indivisible injury for which the contribution of each cannot be measured. These possibilities require different answers. Where independent wrongs produce distinct, identifiable losses, allocation may follow the evidence of each contribution. Where forces combine into a single injury, liability and contribution rules may allocate a loss even if no precise causal fraction can be extracted. A legal judgment that each defendant was a cause does not entail that each was a cause of exactly half the injury.

Successive injury adds a temporal problem. In Baker v. Willoughby, reported at [1970] A.C. 467 following a 1969 House of Lords judgment, the claimant's damaged leg was later amputated after a shooting. The case raised whether a subsequent wrongful event reduced the liability of the initial tortfeasor for continuing loss. Its importance lies in the interaction between an initial injury, later independent wrongdoing, and the legal measure of continuing damage rather than in a simple proposition that an earlier physical cause always remains sufficient forever (Baker v. Willoughby, [1969] UKHL 8).

One must distinguish three objects: the bodily state actually produced by the initial event, the economic loss that results over time, and the obligation to compensate for a legally defined loss. A later disease, accident, or intentional assault may affect each differently. A subsequent event can be causally relevant to the claimant's condition and still leave room for a rule allocating continuing financial consequences to the earlier wrongdoer. The descriptive and remedial questions should not be treated as interchangeable.

An account of causation therefore must be accompanied by a theory of apportionment. Joint liability, several liability, contribution among defendants, and limits on recovery are institutional rules for organizing responsibility after some facts of causal contribution are established or presumed. Their philosophical justification may invoke corrective justice, fairness to claimants, the avoidance of windfalls, and the allocation of insolvency risk. The law gains clarity when it states which of these values drives the remedial choice rather than attributing every allocation decision to an alleged property of causation itself.

Exceptional Risk Attribution: Fairchild and Mesothelioma

Fairchild v. Glenhaven Funeral Services Ltd ([2002] UKHL 22) confronts a distinctive evidential obstacle. Workers exposed to asbestos by more than one employer developed mesothelioma, while the state of medical knowledge prevented identification of which exposure initiated the disease. Ordinary individualized but-for proof could not be furnished. The House of Lords formulated an exceptional route to responsibility where wrongful exposure materially increased the risk of the very disease the defendant had a duty to prevent. The judgments expressly recognized the danger of extending the exception without constraint (Fairchild, paras. 8–10, 37–42, 47–61).

The case is philosophically significant because it forces a choice among justice in the distribution of evidential risk, causal ontology, and the protective purpose of a legal duty. The claimant can prove the harmful outcome and multiple wrongful exposures of a disease-producing kind, but cannot prove which specific exposure caused the disease under the ordinary test. If each defendant escapes liability because the physical process is scientifically untraceable, the duty may lose much of its practical protection for the class of harm it addresses. Yet treating exposure to risk as equivalent to proven individualized causation imposes a burden on a defendant whose exposure may not actually have produced the disease.

The strongest interpretation recognizes Fairchild as a specially justified responsibility rule responding to this precise evidential structure. It should not be represented as an empirical finding that every exposure was the actual physiological cause of the disease. Lord Hoffmann's analysis treats causal requirements as conditions of liability shaped by legal purposes and explicitly addresses the relation between factual questions and the norms that determine their content. That approach makes the policy choice visible without pretending that the underlying scientific uncertainty disappeared.

The subsequent litigation culminating in Sienkiewicz v. Greif (UK) Ltd ([2011] UKSC 10) examined the reach of the mesothelioma exception in a single-defendant context involving environmental and occupational exposure. The legal landscape also contains statutory interventions affecting the allocation of mesothelioma loss. These developments show the need for care in generalizing across disease types, evidence patterns, and legal regimes (UK Supreme Court, Sienkiewicz case record).

Intervening Causes and the Idea of a Broken Chain

The Latin term novus actus interveniens describes an event or action treated as interrupting a prior attribution of responsibility. The metaphor of a causal chain can mislead if it suggests that a later event erases an earlier physical contribution. If one person negligently creates a hazard and another person exploits it, the first person's conduct remains part of the history even if the law locates responsibility for the final injury elsewhere. The doctrinal question concerns the legal significance of an intervention, not the physical reversal of the past.

Three factors are relevant. An intervention may introduce a substantially independent process. It may be the deliberate and informed choice of a person capable of accepting responsibility. And it may lie outside the risk against which the original duty was designed to protect. These factors frequently coincide but are not identical. An ordinary third-party action may be precisely the risk that an employer, custodian, or public authority undertook to control. In such a case, treating the action as invariably severing liability would contradict the purpose of the duty.

The relevant case law is context-sensitive. The House of Lords in R v Kennedy (No 2) discussed deliberate and informed self-administration in relation to unlawful-act manslaughter and the statutory offense in issue. It also explained why cases concerning strict statutory protection against pollution cannot be transformed into universal rules for homicide. The court's approach depends on identifying the legally prohibited act before asking whether it caused the death (Kennedy, paras. 14–15, 19–26).

The philosophical lesson is that voluntary agency matters as an independent normative fact and as an event in a causal process. A free choice can be causally influenced by an earlier threat, coercion, or injury even when that influence does not answer the legal attribution question. Conversely, a subsequent intentional choice can bear full responsibility even when a prior actor created an opportunity for it. Autonomy is neither a universally sufficient nor a universally necessary condition for a legal break in attribution. Its relevance must be interpreted in light of the rule governing the case.

Autonomy and Responsibility in R v Kennedy

In R v Kennedy (No 2) the defendant prepared a syringe containing heroin and supplied it to another adult, who voluntarily injected himself and died. The House of Lords allowed the appeal against a manslaughter conviction. Its reasoning centered on the distinction between the defendant's conduct and the fully informed adult's act of self-administration, together with the offense-specific requirement to identify the defendant's unlawful act. The judgment did not hold that earlier conduct could never be causally relevant in an ordinary explanatory sense (Kennedy, [2007] UKHL 38, paras. 13–26).

The court's treatment of autonomy is rooted in criminal responsibility. A person may offer, persuade, or make a course of action available; another may choose it independently. To call the offeror the legal cause of the chooser's act, in the relevant offense, can obscure the legal recognition of the chooser's agency. This is why the judgment places weight on knowledge, deliberation, and choice. It also identifies exceptions involving vulnerability, deception, duress, and a lack of full responsibility. These qualifications matter because an apparent act of choice may not perform the same normative function under coercive or impaired conditions.

The general philosophical claim should be framed carefully. Even when the decision is deliberate, an earlier person's act can be a factual cause of the decision in a broader causal account. A threat may cause someone to flee; a warning may cause someone to seek shelter; an injury may cause someone to request treatment. The legal question is whether the relevant offense attributes the resulting harm to the earlier actor. Recognition of autonomous choice specifies a normative boundary rather than eliminating all possible causal dependence. Treating that boundary as a metaphysical law would make it impossible to describe many ordinary human interactions causally.

Kennedy therefore supplies a powerful test of conceptual discipline. A completed criminal offense has elements defined by law. The prosecution must identify the prohibited act, the required mental state, the resulting harm, and the connection specified by the offense. General statements about causation have limited value until those elements are fixed. What counts as responsibility for self-administered harm in this context cannot automatically determine negligence liability for a defective medication or the obligations of a professional charged with protecting vulnerable persons.

A Statutory Contrast: R v Hughes

In R v Hughes ([2013] UKSC 56) the United Kingdom Supreme Court interpreted an offense involving causing death by driving while uninsured or unlicensed. The mere presence of the defendant's vehicle in a fatal collision, without fault in the manner of driving that contributed to the death, did not suffice under the provision as construed. The court required some act or omission in the defendant's control of the vehicle involving an element of fault and contributing in more than a minimal way to the fatal outcome (Hughes, [2013] UKSC 56, para. 36).

The case is important because an illegal status and a causal event can coincide without the unlawful status explaining the harm. The fact that a driver lacked insurance or a license can be established without showing that the fact of being uninsured or unlicensed caused a collision. If the offense were construed to attach liability for death whenever such a driver happened to be involved, it would approach a fundamentally different rule of strict result liability. The Supreme Court declined that construction in the particular statute. Its reasoning identifies a requirement linking the defendant's driving conduct to death.

Conceptually, a statute may criminalize an act, a status, a dangerous manner of acting, or an act that produces a result. Conflating these objects risks treating any coincidence of unlawfulness and harm as legally causal. A better analysis asks which fact the legislature has made relevant to the harmful outcome. This distinction reaches beyond traffic offenses to environmental crimes, professional licensing violations, workplace safety statutes, and regulated technologies.

The case also shows why normatively meaningful distinctions sometimes determine the description of the causal input. The driver's vehicle may have been physically involved in the collision, yet the relevant statutory question concerned driving in a legally significant manner. The inference from involvement to causing death under the offense is not purely geometric. It requires interpretation of the norm and a finding that the relevant manner of acting contributed to the fatal consequence.

Victim Responses, Medical Treatment, and the Limits of Autonomy

Harm often unfolds through responses: a victim flees an attacker, a physician treats an injury, relatives arrange care, or a patient makes an irrevocable decision under severe suffering. An intervening choice is a new causal event, but its normative significance depends on the relationship between the original wrong and the later event. Ordinary efforts to escape, obtain treatment, or cope with disability may be foreseeable responses to the initial danger. The fact that an additional person participates does not by itself transform the ensuing harm into an unrelated event.

R v Wallace ([2018] EWCA Crim 690) confronted a disturbing sequence in which an acid attack produced profound permanent suffering, followed by the injured person's request for legally administered euthanasia in Belgium. The Court of Appeal rejected the trial ruling that the later choices necessarily terminated causal attribution as a matter of law. It ordered a new trial and examined whether a properly instructed jury could find that the original attack was a significant and operating cause of death, taking account of the gravity of the injuries and the causal proximity of the later decision to the suffering (Wallace, paras. 50–63, 77, 86).

The case should be used for the issue it actually decided. An appellate determination that a causation question could properly be considered by a jury is different from a final finding of the defendant's guilt. The judgment also examines a more nuanced account of volition than the slogan that every fully informed decision breaks a causal chain. Where severe injury radically structures the options available to a person, the autonomy of a subsequent decision and its dependence on the injury must both be examined.

Medical treatment presents similar questions. A treating clinician may contribute causally to a patient's later condition, but the initial tortfeasor's liability may persist when treatment is a normal response to the injury. Exceptionally independent or extraordinary treatment can raise a more difficult question. The explanatory model should examine whether the initial injury remains an operating contributor and whether the resulting harm belongs to the risk addressed by the initial legal rule. A universal formula based on “last event” or “nearest cause” would obscure both inquiries.

Omissions, Duties to Act, and the Counterfactual of Prevention

Omission cases make explicit the normative selection already implicit in many counterfactual inquiries. An omission is usually described relative to an action that might have been performed: a lifeguard did not rescue a swimmer, a doctor did not prescribe indicated treatment, or a factory did not activate a safety mechanism. There is no single physical event corresponding to the negative description “did not act.” Nevertheless, the absence of an intervention can be causally relevant when the counterfactual intervention would have prevented a consequence. The philosophical problem is how to represent that relevance without treating every person who might conceivably have intervened as a cause of every preventable harm.

A distinction between causal relevance and legal duty supplies the first part of an answer. An onlooker might have rescued someone and thereby prevented death, while the law may impose no general obligation to attempt rescue in that jurisdiction. A professional lifeguard under a particular duty occupies a different normative position. The counterfactual causal question—would competent rescue have prevented the death?—may be similar in both situations. The conditions of liability differ because duties, relationships, and legally required conduct differ.

The role of omission also shows why simple physicalist descriptions of causes are insufficient for legal practice. An absence can be embedded in a causal explanation as the missing condition that would have blocked an existing mechanism. It may be more precise to say that the failure allowed an already developing process to continue than that the omission produced the physical process by itself. The distinction becomes especially important when comparing direct harmful action, neglect, and nonprevention. Moore's analysis of omissions asks whether the law's causal vocabulary sometimes expresses a responsibility judgment in place of a productive relation; Sartorio's work on actual causal histories and responsibility offers a different route toward understanding such cases (Moore, 2009; Sartorio, 2016).

A satisfactory legal analysis should identify the actual threatening process, the omitted precaution, the legally relevant alternative, and the evidence that the alternative would have changed the outcome. If the counterfactual outcome remains uncertain, evidential rules must govern how that uncertainty affects liability. Omissions therefore reveal the importance of an articulated baseline: absent a specified duty and a practicable intervention, causal attribution to “failure to prevent” can expand indefinitely.

Criminal Homicide and the Requirement of a Result

Result crimes such as homicide ordinarily require proof of a defendant's prohibited conduct, an appropriate mental element, a victim's death, and a legally sufficient connection between the conduct and death. The structure differs from that of crimes complete upon performing an act without a specified harmful result. An attempted killing and a completed killing can express the same culpable intention while differing in whether death was caused. The difference matters to both legal classification and philosophical theories of responsibility for outcomes.

A person may inflict an injury that is one of several significant causes of death. The law does not usually require the prosecution to show that the defendant's act was the sole or the principal physical cause. Yet it must identify a legally sufficient contribution in the context of the offense. Intervening conduct, treatment, preexisting illness, and the victim's own actions complicate the inquiry. The relevant judicial instructions should preserve the distinction between the existence of some contribution and the legal significance of that contribution. Describing a mere background coincidence as a cause risks converting a result crime into an offense of dangerous proximity to tragedy.

The contrast between Kennedy, Hughes, and Wallace illustrates why one universal maxim about voluntary acts or significant contribution cannot replace interpretation of the particular offense. Kennedy turns on the self-administration of drugs and the identification of an unlawful act; Hughes concerns the legislative meaning of causing death by driving; Wallace concerns the permissible submission of a complex causal sequence to a jury in a homicide prosecution. Each judgment is situated in different statutory or common-law elements (Kennedy, 2007; Hughes, 2013; Wallace, 2018).

The deeper philosophical issue is the role of moral luck. Two equally culpable attempts may differ because one victim survives. A legal system that gives causal consequences weight must explain why the actual resulting harm legitimately affects legal grading or punishment. One response treats actual harm as a distinctive reason for public accountability; another emphasizes culpability and doubts whether fortuitous differences justify large differences in sanction. Causation cannot resolve this dispute alone. It provides an account of what happened; the theory of punishment must justify the significance assigned to it.

Strict Liability, Statutory Purpose, and Regulated Risk

Strict liability changes the role of fault but does not necessarily eliminate the requirement of a relation between an actor and a prohibited result. An environmental statute may impose liability for a discharge without proof that the operator acted negligently. A product-liability rule may allocate responsibility for a defective product without demonstrating carelessness by a particular employee. These regimes are shaped by legislative choices about safety, prevention, and the distribution of losses. Their causal conditions must be interpreted within those purposes.

The contrast between result-based prohibitions and status-based prohibitions is decisive. A rule making it an offense to operate without a license can be violated even if no injury follows. A rule making a person liable for causing a particular contamination requires an interpretation of how the person's activity contributed to the contamination. Where a third party interferes with a defective container, the question may be whether preventing that interference was part of the hazard-control obligation imposed by the statute. An intervening act cannot be evaluated independently of that purpose.

The House of Lords in Kennedy distinguished its criminal law analysis from an earlier pollution case precisely because statutory responsibility depends on the offense being interpreted. The judges explained that causal requirements are sensitive to what the prohibition is intended to regulate. This illustrates a broader jurisprudential principle: the natural language of causation does not supply all the conditions of statutory liability. Legislatures choose the triggering harm, the relevant kind of conduct, and the circumstances in which an intervention alters attribution (Kennedy, para. 15).

The doctrine should nonetheless preserve legal certainty. If a statute imposes serious sanctions, an expansive concept of causation developed after the event may undermine fair notice. Interpreters must seek a principled relation among statutory language, protected interests, foreseeable pathways of harm, and the degree of culpability the regime requires. A theory in which causal language simply means “whatever allocation seems good policy” surrenders a meaningful constraint on public coercion.

Causation in Contract and Economic Loss

Contractual causation is organized around the promise or legally enforceable obligation that was broken. If a supplier delivers equipment late, the buyer may suffer lost sales, additional financing costs, and later strategic losses. The question is not merely whether the breach appears somewhere in a chronological chain. It is which losses were caused by nonperformance under a defensible counterfactual of performance, and which losses fall within the responsibility assumed or allocated by contract law.

The traditional common-law doctrine of remoteness in contract, associated with Hadley v. Baxendale (1854), places emphasis on losses arising naturally or within the parties' reasonable contemplation under the doctrine's specific requirements. This differs in structure from the negligence duty owed to strangers, and from criminal offenses where a result must be linked to a prohibited act. Contractual risk allocation, express limitations, mitigation, foreseeability, and the expectation interest affect the measure of liability even where causation in the ordinary sense is uncontroversial (Hadley v. Baxendale (1854)).

The counterfactual baseline is especially important. If a seller had performed as promised, would the buyer have earned the alleged profits? The inquiry demands attention to market conditions, business decisions, substitutes, and preexisting risks. A claim for lost profit is often a claim about a counterfactual economic trajectory rather than direct physical damage. Causal inference thus interacts with economic models and evidence about expectations. A model that assumes the success of every planned transaction may exaggerate the loss; one that treats every uncertainty as grounds for denying recovery may empty the promise of practical value.

The broader lesson is that a general philosophy of causal involvement cannot deliver a uniform legal test across tort, crime, and contract. Different legal interests generate different relevant counterfactuals and remoteness rules. That difference does not imply incompatible physical universes. It means that legal attribution is a structured application of causal inquiry to institutions with distinct purposes and standards.

Corrective Justice, Responsibility, and the Reason to Repair

A theory of causal connection explains how one event contributed to another. Corrective justice adds a relational question: why should one person restore or compensate another for a wrongfully caused loss? The answer normally refers to the relation between wrongdoer and claimant, the breach of a duty owed, and the loss that constitutes the wrong's relevant consequence. Causation has moral importance because it helps identify whose conduct brought about an injury to another's protected interest. Yet even proof of causal contribution does not by itself establish a duty of repair when no relevant wrong, protected interest, or liability rule exists.

Consider a lawful rescue operation that unavoidably damages property while saving lives. The physical causal history may be clear. Whether compensation is due depends on doctrines of necessity, public obligation, restitution, or special compensation rules. Conversely, negligent conduct may create a serious risk without producing an actual injury. The fact of negligence does not automatically establish compensatory liability for a nonexistent loss. These examples show why causation is important without exhausting the legal theory of redress.

Michael S. Moore offers an especially strong defense of the role of genuine causation in moral and legal responsibility. He argues that law should resist using causal vocabulary as a label for considerations that are properly evaluative. His position creates a challenge for theories that treat remoteness as merely another physical causal property. A conceptually cleaner account can retain a causal relation and then ask what moral reasons support attaching liability to some of its consequences (Moore, 2009).

The opposing danger is reducing every responsibility rule to the actor's metaphysical participation in harm. Responsibility may arise through authorization, enterprise risk, institutional role, or voluntary undertaking, and legal systems may compensate injuries through social insurance without identifying a particular wrongdoer. A comprehensive theory must explain causal responsibility as one significant mode of legal relation within a wider field of normative institutions. Its central value is that it connects claims for redress to the history of the harm while permitting principled limits.

Consequentialist Explanations and the Distribution of Loss

Causal doctrine can influence incentives. If firms expect to compensate for foreseeable injuries produced by inadequate precautions, they have reasons to invest in safer practices. If responsibility follows proof of an actual contribution, actors may internalize part of the costs of risky conduct. These consequences are significant, particularly where the harmful outcome results from many interacting institutional choices. But an incentive analysis must identify whose behavior can be changed, what information actors possess, and how the cost of a rule is distributed.

A risk-based liability exception may protect claimants from systematic evidential impossibility, as the mesothelioma litigation illustrates. It may also impose costs on defendants whose individual contribution to the eventual disease cannot be proved. A sound policy analysis makes both features visible. The correct rule cannot be derived simply by maximizing the number of compensated claims, because law also protects defendants against ungrounded attribution. Nor can it be derived simply by insisting on an epistemically impossible form of proof where the legal system has undertaken to protect against the relevant risk.

The interaction between deterrence and corrective justice is therefore productive. Corrective justice focuses on a relation between particular parties and a particular wrong; deterrence focuses on the general effects of liability rules. A court may use considerations from both, but clarity demands that it identify the kind of argument supporting an exceptional allocation. A policy of encouraging safe conduct does not demonstrate that an exposure was the actual cause of a disease. It may, however, justify a narrowly defined liability rule under the legislature's or court's recognized authority.

The philosophical constraint is institutional legitimacy. Judges operate under law and precedent; legislators can revise rules prospectively; scientific experts estimate causal effects within disciplinary standards. A justification based on broad social welfare should not be smuggled into a factual finding without explanation. A transparent doctrine will state when it moves from a causal proposition to a policy judgment and which institution has authority to make that judgment.

Comparative Legal Traditions and the Limits of Generalization

The concepts factual cause, proximate cause, and scope of liability are products of particular doctrinal histories. Continental private-law systems may organize similar problems through adequacy, protective purpose, legal attribution, or other concepts whose boundaries do not map perfectly onto Anglo-American terminology. Criminal systems likewise vary in their treatment of objective attribution, risk creation, permissible risk, and the realization of legally disapproved danger. Translation should preserve the role a concept actually plays within its legal order rather than equating words across systems on the basis of a superficial resemblance.

A useful comparison asks a structured set of questions. Does the system treat counterfactual necessity as the ordinary point of departure? How does it analyze independently sufficient contributions? Does it limit liability through a theory of foreseeability, the protective purpose of the rule, or a doctrine of objective attribution? Where are acts of victims and third parties considered? Which issues are questions of law, which are factual findings, and which are entrusted to expert evidence? Differences in terminology may disguise similar operational judgments, while similar terminology may conceal substantially different institutional roles.

A philosophical monograph should resist the temptation to invent a single transnational doctrine. Local statutes, precedents, codes, and procedural rules control real outcomes. The comparative significance of the present analysis lies at the level of conceptual questions that legal orders repeatedly face: identifying a historical connection, determining whether it realizes the danger addressed by a norm, and allocating responsibility under justified standards of proof. These questions are portable; particular solutions require jurisdiction-specific source study.

This bounded comparative position also offers a research agenda. Work on civil-law risk attribution and common-law remoteness can clarify which doctrinal distinctions respond to general features of causation and which embody particular choices of legal policy. The limits of this dossier are explicit: its primary case analysis is chiefly Anglo-American, with prominent English decisions and a New York tort decision. It supplies a comparative analytical framework, rather than claiming exhaustive reconstruction of all national systems.

Scientific Causation, Epidemiology, and Law's Evidential Burden

Legal causation frequently depends on external scientific knowledge. Toxicology evaluates dose and mechanism; epidemiology estimates associations and causal effects in populations; engineering reconstructs sequences of failure; medicine examines competing explanations of injury. These methods can make a causal proposition more or less credible. They do not automatically determine the legal duty breached, the level of proof required, or the remedy available. The legally significant question must be translated into empirically tractable variables without replacing it with a different question that happens to be easier to measure.

A frequent mistake is to treat any population-level association as sufficient to identify an individual cause. Another is to demand direct observation of a microscopic process when a robust body of mechanistic and statistical evidence could support an inference. Both mistakes neglect the structure of causal inference. The assessment should consider possible confounding, selection, measurement, alternative mechanisms, and the relevance of the study population to the claimant. The court may then determine whether the resulting evidential state meets the applicable standard or triggers a carefully defined exception.

In settings where harm unfolds through many weak contributors, the concept of individual causal share may itself become difficult. Air pollution, climate-related damage, and large-scale industrial exposure present temporally distributed processes. A single contributor may have a small effect on risk or severity even where aggregate effects are substantial. Scientific modeling may identify contributions at different levels of resolution. The law must decide whether to recognize a claim for an individual share, a collective wrong, a statutory remediation duty, or a public compensation mechanism. These are choices about the subject and form of legal responsibility.

The general principle is methodological fidelity. Causal science should supply evidence about the processes it can investigate. Legal reasoning should state the question for which that evidence is relevant. Where the evidence cannot identify an individual causal history, the law should openly acknowledge the limit and address any justified redistribution of evidential risk. This avoids converting an ethical or institutional decision into an unsupported scientific assertion.

Artificial Systems, Automated Decisions, and Causal Traceability

Automated systems introduce new configurations of familiar attribution problems. A public authority may rely on a predictive model; a company may deploy a decision system that changes access to employment or credit; a fleet of machines may coordinate actions across sensors, operators, and organizational policies. The causal history of a harmful decision may involve training data, model parameters, a deployment choice, user instructions, institutional thresholds, and subsequent human conduct. A legal inquiry must determine which of those factors contributed to the actual result and which are relevant under the applicable duty.

The presence of an artificial system does not make causal responsibility self-evident. If an automated decision changes an applicant's treatment, a factual reconstruction might ask whether the same treatment would have occurred under a lawful alternative model or procedure. The comparison requires more than observing a correlation between model output and adverse outcome. It requires examining the decision pipeline, the authority of the operator, the causal influence of the output, and whether another process would have produced the same result. Logging, provenance records, reproducible configurations, and independent evaluation can improve evidential access to these questions.

Legal attribution remains sensitive to the juridical status of the participants. Developers, deployers, institutional decision makers, and users occupy different roles. A human official's decision to accept a recommendation may be an intervening event, but it may also be an event specifically anticipated by the design and governance of the system. Attribution must consider the purpose of the original duty and the range of risks that proper governance was supposed to prevent. The analysis generalizes the lesson of autonomous interventions without presupposing a universal causal break whenever a human approves a machine-generated recommendation.

For Philosophy by Bogdanova, the Artificial Era opens a further question about the possible historical emergence of distinct nonbiological bearers of action and responsibility. The presently examined law does not establish that philosophical possibility as an existing uniform legal status. This dossier accordingly treats causal traceability and juridical allocation as distinct research questions. The philosophical extension consists in asking how a future legal order would determine action, consequence, responsibility, and provenance among biological and artificial participants, while preserving the evidentiary standards that causal attribution requires.

The Philosophical Dispute: Natural Relation or Legal Construction?

A central disagreement concerns whether legal causation names a relation whose nature is independent of law or a legal construction shaped by the rules that invoke it. The natural-relation position is attractive because it places constraints on adjudication: judges cannot make a defendant cause an event simply by wishing to impose liability. Moore's analysis gives this thought sustained metaphysical and moral expression. Wright similarly insists on separating causal involvement from ultimate liability. Their work responds to the risk that causal vocabulary becomes a vehicle for unarticulated judicial discretion (Moore, 2009; Wright, 1985).

The legal-construction position emphasizes that the law itself defines the relevant harm, counterfactual baseline, duty, evidentiary threshold, and liability consequence. Lord Hoffmann's analysis in Fairchild shows why an apparently factual question can depend on a prior normative decision about which connection matters for a specified rule. This perspective responds to the risk of presenting legal allocations as deductions from metaphysics alone (Fairchild, paras. 47–61).

These positions can be assessed without reducing them to an artificial binary. A court may investigate a real causal history while asking a legally selected question about it. The causal mechanism need not be created by the law; the legal relevance of the mechanism is constituted through law. The court's power to allocate liability without proof of individualized causal contribution is an additional legal fact about the rule, not a retroactive alteration of the event. The distinction supplies a coherent answer to much of the dispute: causation can be real while legal attribution is institutionally structured.

The remaining disagreement concerns borderline cases where causal concepts themselves are plural. Different legitimate descriptions of events and alternative baselines may yield different truths about dependence. Here legal purposes can guide selection without granting unlimited authority over factual conclusions. A defensible position must require publicly intelligible selection rules, fidelity to evidence, and consistency across comparable cases. The aim is a law of causal attribution that is both empirically answerable and normatively accountable.

A Critical Reconstruction of the Main Arguments

The necessity argument states that an act causes a consequence if and only if the consequence would not have occurred without the act. The argument's strength is its clear counterfactual test; its decisive weakness is that necessary dependence is not universal in duplicative causation and preemption. Its defensible form is a useful ordinary criterion for particular kinds of causal history. The sufficient-set argument states that actual causal contribution can be established through necessary membership in a sufficient set. Its strength is its ability to model multiple contributors; its unresolved demands concern what constitutes a genuinely sufficient set and how the actual productive process is identified.

The common-sense selection argument holds that ordinary explanations identify abnormal events and deliberate interventions against normal conditions. Its strength is sensitivity to practical explanation and human agency. Its weakness is that explanatory salience is not equivalent to a reason for legal responsibility. The realist relation argument insists that the causal facts retain their identity independently of liability policy. Its strength is protection against the arbitrary enlargement of responsibility; its challenge is to explain how descriptions of actions, omissions, and complex events are individuated for legal purposes.

The normative-scope argument holds that an actor is responsible only for the consequences that the relevant duty, offense, or protected risk brings within the law's concern. Its strength is that it links legal outcomes to a publicly justified norm rather than to an indefinitely extended physical chain. Its weakness arises if every desired outcome can be rationalized after the event by adjusting the scope. The solution is to insist upon transparent ex ante reasons, fidelity to authoritative sources, and principled comparison across cases.

The evidential-risk argument permits narrowly specified responsibility where ordinary individualized causal proof is unavailable and the legal system has strong reasons to protect against a defined risk. Its strength is that it prevents some duties from becoming practically inoperative. Its difficulty is the possibility of imposing liability on an individual whose exposure was not the actual cause of the disease. The strongest form therefore states its exceptional character and independently justifies the allocation. Together these arguments support a layered architecture rather than one universal causal formula.

Unresolved Questions in Contemporary Legal Causation

Several questions remain genuinely open. First, how should law identify causal contribution when an event has numerous independently sufficient or interacting causes? Philosophical solutions differ in how they individuate the event, specify sufficient sets, or model interventions. Each approach may perform differently when the evidence concerns a deterministic mechanical accident, a probabilistic disease, or an extended economic loss. A unifying test must explain these differences rather than merely redescribe them.

Second, when does risk evidence legitimately replace or modify individualized causal proof? Fairchild shows that ordinary proof rules can fail for structural scientific reasons, but it also shows the need to limit exceptional responsibility. Any expansion into environmental damage, pharmaceuticals, or algorithmic decisions requires a new justification tied to the substantive duty and an independent assessment of evidence. It is not enough that the injured claimant deserves sympathy or that the defendant's conduct was dangerous.

Third, how should omission and human autonomy interact in causal attribution? A deliberate choice may constitute an independent wrong, a response to danger, or the very act an earlier duty existed to prevent. The operative legal purpose can distinguish these cases, but a complete metaphysics of negative causation and agency remains contested. Fourth, how should uncertainty in causal models be expressed in legal findings? Courts need decisive outcomes; science often supplies ranges, sensitivities, and competing mechanisms. Reliable institutions require ways to translate such evidence without overstating its precision.

Finally, emerging distributed systems put pressure on traditional actor-centered concepts. When processes unfold across institutions and technologies, the relevant “act” may be an enduring policy or a configuration maintained over time. Determining whether such a process caused a harm will require careful temporal descriptions, provenance, and institutional responsibility. The question remains the same in principle—what produced the relevant outcome and why should this participant answer for it?—while its empirical and normative complexity grows.

Philosophical Synthesis: Causal Involvement and Norm-Governed Attribution

The analysis yields a systematic account of legal causation as norm-governed causal attribution. Its first proposition is ontological: harms and consequences arise through events, processes, acts, omissions understood against specified alternatives, and interacting conditions. Their historical relations are objects of inquiry. The second proposition is interpretive: the substantive legal rule identifies which conduct, alternative, consequence, and protected risk form the relevant inquiry. The third is epistemic: legal institutions assess the evidence for the selected causal connection under the governing proof regime. The fourth is normative: responsibility and remedy require further legal reasons that cannot be inferred from causal contribution alone.

These propositions establish an order of reasoning. One begins by identifying the legal question without prejudging its answer. One then reconstructs the actual history and relevant counterfactuals, testing dependence, sufficient-set contribution, and mechanisms as appropriate. One determines whether the harm falls within the purpose and scope of the operative rule. One applies the burden and standard of proof, transparently stating whether liability is grounded in proven causation or a special authorized allocation of risk. Finally, one determines the legal consequence—conviction, compensation, contribution, or another remedy—under independent substantive and procedural constraints.

This sequence improves upon a simple division between “factual” and “legal” cause because it reveals that factual inquiry itself requires specification of the act and result, and that legal limits can be distributed among duty, breach, scope of liability, and remedies. It also improves upon undifferentiated policy reasoning because every step has an identifiable object and standard. The approach is compatible with the insight that ordinary causal explanation selects salient conditions while requiring that legal salience be justified rather than assumed.

The synthesis has a further implication. Legal causation is neither identical with moral blame nor exhausted by scientific causal inference. It is a structured relation among causal reality, legal description, evidential warrants, and reasons for public accountability. This relation explains why two legal systems may assign different responsibility for the same physical history without disagreeing about the laws of nature. It also explains why neither system is justified in asserting a factual causal connection unsupported by its evidence, unless it openly applies a different authorized basis of responsibility.

Conclusion

Legal causation asks how the consequences of conduct become attributable under law. Its enduring philosophical difficulty arises from the conjunction of a real causal world with institutions that decide which harms merit public redress, which acts count as legal wrongs, and which uncertainties litigants must bear. The historical trajectory from Hume and Mill through Hart and Honoré, Mackie, Lewis, Wright, Moore, Stapleton, and contemporary causal modeling shows that causal involvement is richer than unconditional counterfactual necessity. Multiple sufficient causes, preemption, omissions, and evidentially inaccessible processes require different forms of analysis.

The case law demonstrates the complementary importance of legal form. Palsgraf locates foreseeability within the architecture of negligence; Fairchild recognizes an exceptional response to inaccessible individualized disease causation; Kennedy gives weight to autonomous intervention within a specific criminal offense; Hughes requires a legally significant contribution by driving; and Wallace shows why subsequent action may remain closely connected to an initial serious attack. None establishes a timeless definition that can be transplanted indiscriminately into every area of law.

The resulting philosophical position is that causal relations must be investigated with conceptual and scientific rigor, while legal attribution must be explained through the norm that makes the consequence significant. Factual involvement, protected risk, standards of proof, and ultimate liability are connected components of one jurisprudential problem. Their explicit separation creates a clearer, fairer, and more publicly accountable practice of reasoning. The unresolved task is to refine this architecture for probabilistic harms, distributed responsibilities, and artificial systems without concealing normative choices behind the language of natural causation.