Remote or Accidental Results

Remote or Accidental Results

The same-kind requirement covered above is only half of § 2.03(2)(b). The provision also demands that the actual result be "not too remote or accidental in its occurrence to have a [just] bearing on the actor's liability or on the gravity of his offense," so a harm of exactly the right kind can still fall outside the element if the route to it is too stretched or too coincidental.¹ Paragraph (3)(b), which governs offenses requiring a reckless or negligent result, repeats the clause with the probable result as its baseline in place of the designed or contemplated one.² One inquiry therefore serves both the purposeful-or-knowing track and the reckless-or-negligent track, so this chapter develops it once here and applies it again when it reaches paragraph (3).

The Code supplies no mechanical definition of remote or accidental, and the omission is deliberate. In Commonwealth v. Rementer the Pennsylvania court examined that state's version of the clause, which omits the word "just," and observed that neither the Pennsylvania statute nor the Model Penal Code tries to catalogue the divergent events that can separate an actual result from the one intended or hazarded.³ The provision is written in general terms so that the trier of fact can use common sense in deciding whether the result has a bearing on the actor's culpability or is too remote to have one. That approach follows from the theory that but-for causation is the only strictly causal requirement imposed generally, and that what remains is how far liability should extend in light of the actor's culpability.⁴ In practice the fact-finder asks whether the connection between act and harm is so stretched that holding the actor liable would be unfair, and unusual or unexpected causal mechanisms are the usual reason the answer is yes.⁵

The reason for leaving the question open lies in what the clause protects. An actor is blamed for what the actor was prepared to bring about, or risked bringing about, and a chain of events the actor neither designed nor risked adds nothing to that blame. If the result were attributed anyway, the actor would be punished for the luck of how events unfolded. The words "remote" and "accidental" can be read as pointing at two different ways a result comes loose from the act: through a long or attenuated sequence of events, or through a coincidence in which the act did not operate as acts of that kind ordinarily do. The phrase "just bearing" then supplies the fairness question that decides how much looseness is too much, and the reference to the gravity of the offense shows that the answer can affect how serious the offense is as well as whether the actor answers for the result at all.

The clearest illustration of the extreme is a shooter who fires at a victim with intent to kill and misses, but the shot makes the victim change direction, and while walking the new way the victim is struck by lightning and dies. The shot is a but-for cause of that death, since the victim would not have been standing there otherwise, and death is the very harm the shooter designed. Even so, most people would refuse to say the shooter caused the death and would recoil from punishing the shooter as severely as a murderer, so the sound outcome is liability for attempting the killing and not for the death itself.⁶ The Code reaches that outcome by putting the question to the jury's sense of justice rather than by announcing a rule about routes.⁷

Courts are somewhat more willing to attribute a result to an actor who intended it, even when it arrived by an unplanned route, and the lightning case lies at the far edge of that willingness.⁸ A result that follows an unplanned path can therefore still be fairly laid at the actor's door. The Rementer opinion points to an analogous New York prosecution arising from the Howard Beach incident, People v. Kern, in which several youths relentlessly chased Michael Griffith during an assault, forced him to jump a barricade and try to cross a well-travelled six-lane highway, and saw a passing motorist strike and kill him.⁹ The defendants argued that their chase had ended and that the motorist's intervention made their conduct too remote a cause of death, and the court emphatically rejected both arguments.¹⁰ A victim driven by the attack itself into the path of traffic is still inside the sequence the assault started, so the death is no coincidence that the assault merely preceded.

The two scenes differ in what stands between the assault and the harm. In the Howard Beach analogue the victim's flight was a reaction to the pursuit, so the attack kept operating until the collision. In the train scene the effects of the attack had ended before the victim made an independent trip, so the collision connects to the assault only in the but-for sense. That distinction is one of degree, and courts reach different outcomes on closely similar facts because the decided cases rest on fine distinctions among intervening, concurrent, natural, and human causes.¹¹ An exam answer should therefore identify the fact that stretches the chain, state the argument on each side, and commit to a conclusion.

Courts and commentators often describe the same judgment in common-law terms, and those terms help a fact-finder reach an answer. One familiar way to organize the inquiry asks whether the result was an objectively foreseeable consequence of the conduct, judged by a reasonable person; whether the conduct is an efficient cause of the result, usually pictured as the number of links in the chain between them; and whether an independent, free-willed agent stepped in and broke the chain.¹² Decisions speaking of natural and probable consequences add that liability requires a cause that is direct and also not so remote as to fall outside the natural and probable consequence of the act, and they reason that punishing a remote consequence as severely as a natural and probable one serves little purpose when the aim is to punish in proportion to moral culpability.¹³ Those formulations illustrate how a fact-finder reasons toward an answer under the Code's language. Students go wrong when they treat foreseeability or superseding cause as elements the Code adds, because the statutory question remains whether the result is too remote or accidental to bear justly on the actor's liability.

The next three subsections apply the standard to settings that recur in decided cases. The first takes up harm that flows directly from the defendant's act by a path different from the one expected. The second concerns a victim whose pre-existing weakness the defendant did not know of, and the third an act or event from someone or something else that arrives between the defendant's conduct and the harm. The question stays the one stated here in each setting, and the settings differ in which fact threatens to stretch the connection.

Unexpected Routes to Harm

The first setting for the remote-or-accidental inquiry is the simplest one. The harm comes from the defendant's own act, no other person steps in between, and yet the harm arrives by a path different from the one the defendant pictured. The Code offers no formula for that situation and leaves the fact-finder to ask, using common sense, whether the connection between act and harm is so stretched that attributing the result would be unfair.¹⁵ In this setting the answer is usually that the connection holds, and the reason shows most plainly at the mild end of the range of departures. A defendant who aims at the victim's heart and hits the head, or who aims a rock at the victim's right eye and strikes the left, has produced the intended kind of injury through a slightly different point of impact, and courts have no difficulty holding that defendant liable for the result.¹⁶

Departures from the expected path run along a spectrum, and this subsection compares the two ends. At one end is the shot that lands a few inches from where it was aimed. At the other is the shot that misses altogether, makes the victim turn, and leaves the victim to be struck by lightning on the new route. Between the two sit cases such as a wounding followed by negligent medical treatment, which depend on an outside actor and are taken up in the subsection on intervening acts and events.¹⁷

Direct sequences near the mild end rarely escape liability, and two reasons follow from the structure of the inquiry. The functional reason is that when nothing but the defendant's own force and the victim's body or surroundings reacting to it lies between act and harm, the chain has few links, and the defendant seldom has anything to point to that makes it look bizarre. The incentive reason is that courts are somewhat more willing to attribute a result to a defendant who intended it, even when it arrives by an unplanned route.¹⁸ That willingness is sensible, because a rule excusing the defendant whenever violence unfolded in a surprising way would reward the defendant for the unpredictability of the very violence chosen. Blame would then shrink as the harm grew more erratic, which inverts the idea that responsibility tracks what the actor set out to do.

The poisoning scene in the discussion of same-kind harm shows the ordinary variation. The wife who put poison in her husband's breakfast milk designed his death, and the dose, though too small to kill, left him dizzy enough to fall and fatally strike his head. The route differs from the one she expected, yet every link was supplied by the poison itself: it entered his body as intended, it produced dizziness, which is an ordinary effect of poison, and the dizziness produced the fall. No independent person or happening supplied the fatal blow, and the floor did no more than a floor does. A fact-finder asking whether the death is too remote or accidental to bear justly on her liability would very likely say it is not, because the result is the one she designed and the route is one an ordinary effect of her act could take.

The extraordinary variation looks different, and a pair of versions that differ in one fact shows what changes.

The two versions differ in a single fact, and that fact is the one that matters. In the first, the effects of the beating had run their course before the victim set out, so the train connects to the beating only because the beating started the day's events. In the second, the impaired judgment that put the victim on the tracks is itself an effect of the blows, so the beating is still operating when the train arrives. The first path resembles the lightning scene, where the shot did no more than place the victim where an unrelated event happened to strike. The second resembles the poisoning, where the act's own continuing effect carries the harm to its end. As in the lightning scene, the sound outcome in the first version is liability for the attempt and for the assault actually inflicted, and not for the death.¹⁹ The second version is the closer call, and a fact-finder could readily attribute the death to the defendant.

Both versions contain an event the defendant did not supply, since a train arrives in each, so asking whether an outside force appeared cannot separate them. Students go wrong when they treat the direct-or-intervening label as the whole test, because the label only shifts the odds. Tort law's older direct-cause test shows the limit. It explained neither the decisions that imposed liability nor those that refused it: a later independent act, such as a second driver's negligent failure to avoid a collision the first driver began, does not ordinarily relieve the first driver, and a sequence that looks direct can still strike courts as too unexpected to charge to the defendant.²⁰ Proximate cause in general resists absolute rules. It is decided on the facts through logic, common sense, justice, policy, and precedent, and the decided cases are best read as illustrations of where careful judges have drawn the line.²¹

Negligence cases reach the same pattern from a different direction. In United Novelty Co. v. Daniels an employer let a worker clean machines with gasoline in a small room heated by an open flame, and a gasoline-soaked rat ran to the heater, caught fire, and ran back to the machines, causing an explosion that killed the worker; the court held the employer liable because the manner was unusual but an explosion was exactly the kind of accident that gasoline near an open flame invites.²² That decision measures the result against the risk that made the conduct wrongful, while the Code measures it against the result the actor designed, contemplated, or hazarded.²³ The lesson carries over anyway, since an odd route to the expected kind of harm leaves the harm inside what the defendant's conduct invited.

Every route in this subsection began and ended with the defendant's act and the victim's physical condition. The next subsection takes up one feature of that condition, a pre-existing weakness the defendant did not know about.

Vulnerable Victims

The previous subsection traced harm along unexpected routes, and each route ran through the victim's body or surroundings reacting to the defendant's act. The body can surprise the defendant in another way, through a weakness the defendant never knew about. When a defendant attacks with intent to kill and inflicts what would be a non-fatal wound on a person of ordinary health, but the victim is weakened by disease, earlier wounds, exposure, or intoxication and dies, the defendant is guilty of murder, because the attack is a direct cause of the death.²⁵ That holds even when the victim would eventually have died of the underlying condition, so long as the attack carried the victim off sooner than otherwise would have happened.²⁶ Weakness from sickness or prior injury appears among the unusual patterns that give causation its difficulty, alongside concurrent wounds, negligent medical treatment, and misdirected shots.²⁷ In People v. Wilks the court rejected the argument that a person with a different medical history might have survived the trauma the defendant inflicted, treating that argument as no defense.²⁸

The rule rests on two separate reasons, one structural and one about who should carry a risk nobody could know. The structural reason comes from the Code's own language, which asks whether the actual result involves the same kind of harm as the one designed and whether it is too remote or accidental in its occurrence to bear justly on liability.²⁹ A weakness is a standing condition rather than an event. Nothing new enters the sequence after the defendant acts, so the attack remains the force that produces the death, and the weakness only determines how little force the attack needed. When the defendant meant to kill, death is the very harm designed, which satisfies the same-kind requirement, and no coincidence separates act from harm. The Code contains no separate clause for frail victims, so the result follows entirely from those two requirements.

The second reason concerns incentives and fairness. A person who chooses to injure another creates the occasion on which the victim's fragility matters, and it would be arbitrary to let that person's liability shrink because the victim happened to be more susceptible than average. If frailty worked as a defense, protection would be thinnest for the sick and the intoxicated, who are the people least able to absorb a blow. The prosecution would also have to prove how some hypothetical healthy person would have fared, which is a question about facts the defendant never considered. Refusing the defense places the risk of an unknown frailty on the party who chose to attack.

Tort law reaches the same place on a different measure. Under the thin-skull or eggshell-plaintiff rule, a defendant who could foresee some injury to the victim answers for the full injury actually suffered, as when a hemophiliac dies of blood loss after being knocked down and sustaining small lacerations, because the defendant takes the plaintiff as found.³⁰ Courts apply that rule universally even though it sits awkwardly beside foreseeability limits on tort liability.³¹ The tort measure is the foreseeable type of harm, while the Code's measure is the designed or contemplated result, so the tort cases support the reasoning without supplying the criminal rule.

The weakness also need not be a wound aggravated by a later blow. In People v. Stamp a robbery frightened a sixty-year-old man who had advanced heart disease, and he collapsed 15 to 20 minutes later and died of a heart attack. Physicians testified that his atherosclerosis was progressive and ultimately fatal, yet but for the robbery there would have been no fatal seizure at that time.³² The court affirmed first-degree murder convictions and approved an instruction that an unlawful act counts as a proximate cause of death even though the person killed was already enfeebled by disease, injury, or other physical condition.³³ Two cautions attach to that decision. The convictions rested on the felony-murder rule, which holds a felon strictly liable for killings in the course of the felony and so dispenses with any intent to kill, which means the decision teaches the causation point while the mental state came from the felony-murder rule. California's current causation instruction also asks whether death was the direct, natural, and probable consequence of the act and whether a reasonable person would know death was likely if nothing unusual intervened, which leaves open whether a jury instructed that way would return the same verdict on those facts.³⁴

Satisfied causation answers one question only, whether the defendant's conduct is the cause of the death. A second question remains about the mental state the offense demands toward that result. The Code's causation rules apply when an offense makes purposely, knowingly, recklessly, or negligently causing a result an element, which presupposes that the offense has already fixed the mental state the prosecution must prove.³⁵ A vulnerable victim enlarges the harm the defendant's act produces, and the frailty does nothing to enlarge what the defendant meant or recognized. Causation does not supply intent, so any shortfall here is a limit on mental state and should be labeled that way on an exam, since calling it a failure of causation misstates the layer at which the defendant escapes the greater offense.

The two scenes differ in a single fact, the defendant's purpose, and that fact decides the grade of offense while causation stays constant. Which lesser offense fits turns on the recklessness and negligence rules taken up later in this chapter. Because the two layers are easy to blur, the problem is best worked in two passes.

The weakness in this subsection was a condition the victim already carried when the attack began. The next subsection takes up a different disturbance, an act or event from someone or something else that arrives between the defendant's conduct and the harm.

Intervening Acts and Events

The previous subsection dealt with a weakness the victim already carried when the attack began. A different disturbance arrives after the defendant has acted, when some other act or event enters the sequence before the harm occurs. The Code has no separate clause for these outside forces. They are measured by the question that governs every divergence case, which is whether the actual result is too remote or accidental in its occurrence to have a just bearing on the actor's liability or on the gravity of the offense.³⁸ That language is deliberately general, and it leaves the fact-finder to apply common sense to whatever unusual event the facts supply instead of cataloguing every possible divergence in advance.³⁹ What the fact-finder needs, then, is a way of seeing which outside forces leave the defendant's responsibility intact and which remove it.

An intervening cause is an event that comes between the defendant's conduct and the end result, and it is called a superseding cause when it is strong enough to relieve the defendant of liability.⁴⁰ It may come from the victim, from a third person, from the defendant, or from a non-human source such as weather.⁴¹ Its arrival never disturbs but-for causation, because the defendant's act remains an antecedent without which the result would not have occurred. What it raises is the legal question of which but-for causes should answer for the harm.⁴² Without that second screen, responsibility would run back through every link of a long chain, and the criminal law, which punishes the actor for the result, draws the screen more narrowly than tort law does.⁴³

Courts and commentators sort outside forces into two groups. The sorting is a convenient way to organize the decisions, though neither label appears in the Code's text. A dependent force is one the defendant intended, could reasonably have foreseen, or that is otherwise sufficiently related to the defendant's conduct, and when the force is of this kind the fact that another agent contributed does not relieve the defendant.⁴⁴ An independent force produces the harm in such an unexpected or unusual manner, so fortuitously and unconnected to the defendant's conduct, that punishing the defendant for the result would be unjust.⁴⁵ The sorting turns on where the explanation for the harm lies. When the outside force responds to what the defendant did, the defendant's act accounts for why the force appeared at all, so the result remains an outgrowth of the choice to act. When the force is a coincidence, the defendant's act accounts only for where the victim happened to be when it struck, and the real explanation of the harm lies elsewhere. If every intervening human choice broke the chain, a defendant would escape responsibility for the predictable reactions of victims, pursuers, and doctors to the very injury the defendant started, and the graver the injury, the more reactions there would be to point to. Tort vocabulary also uses the two labels in a different sense, so an answer that describes the substance of the force is safer than one resting on the label alone.⁴⁶

Responsive acts by other people are the clearest dependent forces. In Commonwealth v. Rementer the court pointed to an analogous New York prosecution arising from the Howard Beach incident, People v. Kern, in which youths relentlessly chased a man who was forced to jump a barricade and cross a six-lane highway, where a passing motorist struck and killed him.⁴⁷ The defendants argued that the chase had ended and that the motorist's intervention made their conduct too remote, and the New York court emphatically rejected both arguments.⁴⁸ The victim was on the highway because the chase put him there, so the motorist's act arrived in a situation the defendants had built. Even a later actor's carelessness does not necessarily detach the result. In People v. Ryan an intoxicated driver caused two collisions and left the car stopped in a high-occupancy lane, and a police officer who came to the scene was struck and killed by another motorist who drove past while looking at the parked cars. The appellate court upheld the finding that the death was reasonably foreseeable, because the driver's conduct had created a roadway danger to first responders in the aftermath of the crashes.⁴⁹

Medical treatment is the most common responsive force, and it rarely breaks the chain. A defendant who wounds with intent to kill creates a situation in which doctors must act, and imperfect treatment is one of the ways a serious injury plays out. The sequence in which the wound is not fatal but a doctor's negligent care lets a preventable infection kill the victim is a standard intermediate case between the slight variation and the lightning strike.⁵⁰ The annotated decisions resolve it against the defendant: in Gibson v. State a fatal staph infection that developed after surgery made necessary by a beating was not an extraordinary intervening cause, and a rape victim who died weeks later while being fed at a nursing home was still the defendant's victim because the need for feeding traced to the crime.⁵¹ The same reasoning covers the physician who withdraws life support from a victim whose condition has become irreversible. Courts treat that decision as a foreseeable consequence of medical treatment and not as an independent intentional killing, so the original attacker remains the cause of death.⁵² In Quintanilla v. State the court likewise refused to treat the family's decision to discontinue life support as a concurrent cause of death, because without the drunk-driving crash the passenger would never have been on life support.⁵³

The two hospital scenes differ in one fact, whether the intervening act was a flawed response to the wound or an unrelated deliberate attack, and that fact decides which side of the line the result falls on. A pure coincidence sits at the far end of the line. In one standard illustration, a defendant inflicts a minor wound, and the victim, driven to a doctor's office for stitches, is killed in the waiting room by a disgruntled patient who opens fire. The assault is a but-for cause, since without it the victim would not have been in that room at that moment. Yet the killing is bad luck. It was neither foreseeable nor made more likely by the assault, so the gunman is an independent intervening cause and the defendant is not the proximate cause of the death.⁵⁴ The lightning strike discussed earlier is the same pattern with a non-human actor, and it produces the same sound result of liability for the attempt and the assault but not for the death.⁵⁵

The highway chase and the waiting-room shooting both contain a third party's act that the defendant did not supply, so the mere presence of such an act cannot separate them. Students lose this issue when they assume that any free human act breaks the chain. In the chase, the defendants' conduct explains why the victim was exposed to traffic, while in the waiting room the assault explains only why the victim was in the building. Tort law reaches a similar place: a dealer who negligently failed to tighten a customer's lug bolts remained liable when the wheel came off, the car stopped on a bridge, and a flatbed truck's crash into a helper's van killed the customer, because the truck driver's culpable driving was not so highly improbable and extraordinary as to bear no reasonable connection to the risk the dealer created.⁵⁶ The inquiry always returns to whether the outside act is a response within the risk of the defendant's conduct.

The victim's own conduct is an intervening force from the first source, and it receives the same treatment with one added variable, which is how deliberate the victim's choice was. A kidnapper forces a man into a car, and the man leaps from the moving vehicle and is badly hurt. He made a choice, but experience shows that captives take serious risks to escape, so the leap is a dependent force and the kidnapper is the proximate cause of the injuries.⁵⁷ The standard list of problem cases includes the victim who commits suicide to escape the pain of a wound and the victim who dives into a river to dodge a bullet and drowns.⁵⁸ Those reactions, like the leap, respond to a danger the defendant created, and courts in most jurisdictions treat a wounded victim's refusal of treatment, or a suicide prompted by the injury, as a foreseeable response rather than a break in the chain, though the answer remains a question of fact.⁵⁹ The contrast is the victim whose act is a free and deliberate choice instead of a reaction. In J.A.C. v. State, as recounted in Velazquez v. State, a passenger in an illegal drag race grabbed the steering wheel instead of the gear shift he was assigned to operate, and the car crashed and killed him. The court held the driver improperly convicted of vehicular homicide, treating the passenger's reckless act as an independent intervening act, since the passenger in effect killed himself.⁶⁰ Dr. Kevorkian's prosecutions show the same boundary: the charges were dismissed because patients of sound mind pressed the button of his device as an act of independent free will, and many states answered with assisted-suicide statutes that avoid the problem.⁶¹

New Jersey's counterpart to the Code adds words that bear directly on this subject. For a purposely or knowingly caused result, the actual result must not be too remote, accidental in its occurrence, or dependent on another's volitional act to have a just bearing on the actor's liability or on the gravity of the offense.⁶² The premise is the Code's own, that variations between the designed result and the actual one are problems of culpability and not problems of metaphysical causation.⁶³ The added phrase might look as though it makes every third-party choice a break in the chain, but a New Jersey decision applying it to felony murder reads it otherwise. Deaths from self-defense or retaliation are not too dependent on another's volition, because they are the foreseeable result of the risk the felon created. The decision marks the outer limits with two examples: a bank robber is not liable when an employee pushing the alarm button is electrocuted as the robber steps inside, but the robber is liable when a shopkeeper's shot at the robber kills a bystander.⁶⁴ The volitional-act clause therefore points the same remoteness inquiry at another person's choice without changing its logic.

Step three does the real work, because the responsive-or-coincidental question explains why flights, rescues, and treatment stay attributed to the defendant while lightning and unrelated shootings do not. The three subsections on remote or accidental results have now covered unexpected routes, vulnerable victims, and outside forces, and all of them apply one clause. The same clause reappears for offenses that require recklessly or negligently causing a result, where it is measured against a different baseline, the risk the actor was or should have been aware of.⁶⁵