Concept
Relationship to the Dangerous Person
Duty to control the conduct of third persons
Relationship to the Dangerous Person
The first exception to the no-duty default looks at the defendant's tie to the person who will cause the harm. Under Restatement (Second) of Torts § 315(a), a defendant must control a third person's conduct when a special relation between the defendant and that third person imposes the duty.1 The inquiry concerns authority over the wrongdoer. The victim's own connection to the defendant belongs to the second exception, taken up after this one, so a plaintiff who relies on § 315(a) builds the case by showing what the defendant could do about the dangerous person.
The qualifying relations share one feature: each gives the defendant a means of restraining the dangerous person. Courts explain the exception as resting on the defendant's acceptance of responsibility by taking charge of a person who poses a risk, and on the fact that the defendant is often uniquely positioned to prevent the harm. An employer's duty adds a further reason, the benefit the employer draws from the employee's work.2 A duty to restrain someone means little unless the defendant holds the power to do it, whether through physical custody, parental or workplace authority, or the trust a patient places in a therapist. A relationship that supplies the power also singles out one defendant from everyone who might have intervened.
Three sources of authority organize what follows. Custody comes first because it is the paradigm: a defendant who takes charge of a person likely to cause bodily harm if not controlled must use reasonable care to control him. Family and employment come second, as with a parent and a minor child or an employer and an employee. The professional relationship between therapist and patient comes last, in the setting of the best-known application of § 315(a). Two limits run through all three and return later in the chapter: the harm must have been foreseeable to the defendant, and the duty requires only reasonable care.
Custody and Section 319
Section 319 states the paradigm: “[o]ne who takes charge of a third person whom he knows or should know to be likely to cause bodily harm to others if not controlled is under a duty to exercise reasonable care to control the third person to prevent him from doing such harm.” Maryland's highest court adopted that rule as state common law in Lamb and restated it in Remsburg v. Montgomery, where it set § 319, the rule for “those in charge of persons having dangerous propensities,” beside the parent, employer, and possessor-of-land rules of §§ 316 through 318.3
Taking charge triggers the duty because of what custody does to the defendant's position. A defendant who takes charge of a person who poses a risk of injury accepts responsibility for that person and is often uniquely positioned to prevent the harm, which is why the rule applies primarily to those in charge of mental patients or prisoners.4 An institution that confines a dangerous person has taken over the work of restraining him, so the means of preventing injury sit with the institution and no one else. The duty is one of reasonable care, not absolute protection: both the Second and Third Restatements describe it as a duty to exercise reasonable care to prevent injury from the dangerous person.5 A custodian who acts reasonably and still loses a patient has not breached.
The operative words come in degrees. Remsburg quotes Lamb: because there are degrees of being “in charge” and having “control,” the illustrations that accompany § 319 are “obviously not by way of limitation.” The same illustrations nonetheless suggest that § 319 has “peculiar application to custodial situations,” a reading that draws support from Prosser and Keeton's remark that the relationships § 319 covers “are custodial by nature.”6
The rule reaches two kinds of defendants. In the first, the defendant has charge of one or more persons of a class for whom a tendency to act injuriously is normal. In the second, the defendant has charge of someone outside any such class who has a peculiar tendency to act injuriously, one the defendant knows or should know of from personal experience or otherwise.7
Courts disagree about how much control the first step demands. The Supreme Court of Virginia treats taking charge as a threshold fact that the plaintiff must prove. In Nasser v. Parker, hospital defendants had released the victim's former boyfriend without warning her, and the court required a showing that they had “taken charge” of him within the meaning of § 319; that showing was missing. The court described the competing approach as “at odds with [its] interpretation of the Restatement's provisions relating to one's duty to control the conduct of a third person.”8
Other courts accept less than literal custody. In Estates of Morgan v. Fairfield Family Counseling Center, the Ohio Supreme Court faced an outpatient over whom the defendants' control was attenuated. It held that a want of close control does not by itself defeat a duty, and it directed a balancing of four considerations: the therapist's ability to control the patient, the public's interest in safety from violent assault, the difficulty of forecasting whether a patient poses a substantial risk of physical harm to others, and the goals of placing patients in the least restrictive environment and of preserving the confidentiality of therapy.9 Hertog v. City of Seattle took a similar position for parole supervision: custodial control is not required, and the question is the officer's relationship with the parolee.10
The two approaches differ in where they put the weight. A court following Nasser makes taking charge a precondition and ends the inquiry when the plaintiff cannot prove it. A court following Morgan treats the extent of control as one factor in a larger weighing. A sound answer states the rule, applies the custody facts to each of its parts, and flags that jurisdictions divide over whether attenuated control suffices.
Parents, Employers, and Others
Three further special relations, none of which depends on physical custody, appear in §§ 316 through 318. Maryland's highest court summarized them in Remsburg: § 316 gives a parent a duty to control a minor child, § 317 gives a master (today, an employer) a duty to control a servant (an employee), and § 318 gives a possessor of land or chattels (movable property such as vehicles and tools) a duty to control a licensee, someone on the land or using the property by the possessor's permission.12 Each is a relation between the defendant and the dangerous person, so the inquiry stays on what the defendant could do about that person.
What the three relations share is authority. A parent directs a child's daily life, an employer chooses where employees work and what equipment they use, and a possessor decides who may enter the land or use the property, so each holds a right and an ability to control that an ordinary bystander lacks, and the law asks the defendant to use that authority with reasonable care.13 The relations draw added support from different sources. The employer profits from the employee's work and is often uniquely positioned to prevent the harm.14 The parent's authority carries a counterweight absent from the other two, and it governs most of what follows.
Parents hold the clearest authority of the three, yet courts hesitate to hold a parent liable for a child's torts, even when the parent knows the child is dangerous and could take steps to prevent harm.15 The hesitation reflects child-rearing values. Teenagers need experience with freedom as well as control, and courts prefer to interfere with parental decisions only in clear cases, so a parent who gives a teenager reasonable latitude may have chosen soundly even if the choice turned out badly.16 The standard remains reasonable supervision, and it does not vanish as the child grows. Camp v. Gregory used the parental relationship as its yardstick. We expect a parent to have the child in hand, protecting him from dangers he is not equipped to avoid; the expected supervision lessens as the child matures but never ceases altogether; and we still expect parents to see that their children attend school, obey curfews, and stay out of trouble. Parents are not insurers of their children's conduct, but they can be held liable when they fail to exercise a reasonable degree of supervision.17
Courts reconcile a parent's authority with their hesitation by demanding specific foreseeability. A parent who can foresee that a child will cause harm of some kind, but not the specific harm or the specific occasion for it, is not liable.18 In Dinsmore-Poff v. Alvord, the Alaska Supreme Court required the plaintiff to show that the parent had reason to know, with some specificity, of a present opportunity and need to restrain the child in order to prevent some imminently foreseeable harm. It added that “[g]eneral knowledge of past misconduct is, in other words, necessary but not sufficient for liability.”19 The reason is practical. A parent who knows only that the child has misbehaved before has no instruction about what to do tonight, and a duty triggered by past misconduct alone would leave confinement as the only safe course, the outcome the freedom rationale warns against. Knowledge of a particular target and a particular time gives the parent something concrete to prevent, by keeping the child home or warning the likely victim. Decisions such as Wood v. Groh fall on the liability side of the line, where parents who knew of a specific propensity and an imminent danger or occasion for it negligently failed to control the child or to warn potential victims.20
Age pushes the analysis further toward no duty. Older children are more dangerous but also harder to control, and courts have shown a strong disinclination to impose liability in those cases.21 Section 316 speaks of a minor child, so once the child is an adult the parental authority that justified the duty has fallen away.22 A parent who knows an adult son is violent, and who could stop paying his rent, has not acquired a right to direct his conduct. A parent is therefore unlikely to be held responsible for an adult child's foreseeable misconduct absent some unusual element of control, and financial support alone is probably insufficient.23 A West Virginia decision shows what an unusual element can look like. In Courtney v. Courtney, a mother who supplied her grown, married son with drugs and alcohol, knowing that they made him violent, was subject to liability to his abused wife.24 Her own conduct fed the danger, which is why treating a family tie plus knowledge of the danger as enough fails: neither supplies authority. A parent who negligently leaves firearms available to a child presents a different issue, because the parent's own affirmative conduct creates the risk, a theory this chapter takes up under risk created through third parties.25
The employer's duty depends on a place or an object. The Restatement recognizes a duty to control an employee who is using the employer's premises or chattels, and that condition marks the points at which the employer's authority is real.26 On its own property or with its own equipment, an employer can send a worker home, take away the keys, or close the gate. An employee at a weekend gathering across town, in a private car, is beyond that reach. A shop owner who sees a visibly drunk clerk reach for the keys to the company van fits the relation; the same clerk driving his own car home from a party does not. Even inside the relation the duty has a foreseeability limit: where the employer could not foresee a risk of harm, no liability results.27 Two take-home asbestos decisions show the limit at work. The Tennessee Supreme Court found that an employer owed a duty to family members exposed to asbestos that employees carried home on their work clothes in Satterfield v. Breeding Insulation Co., while the Sixth Circuit in Martin v. Cincinnati Gas & Electric Co. found no liability to a family member in the same position because it was not reasonably foreseeable that the asbestos would endanger that person.28
Section 318 extends the same logic to a possessor who permits another to use land or chattels, because the permission is the source of the possessor's power. The Restatement recognizes the duty toward those who are on the land or using the chattels with the possessor's consent.29 In Volpe v. Gallagher, the Rhode Island Supreme Court recognized that a landowner could owe a neighbor a duty to control a dangerous person who lived in her home and stored weapons there.30 A possessor who allows the use can withdraw consent, order the person off, or take back the chattel, and that leverage is what makes a duty to reduce the risk sensible. For the same reason the duty does not reach a person who enters without the possessor's consent, since the possessor then has no permission to withdraw.
A spouse or other relative falls outside §§ 316 through 318, so a claim against one depends on a court's willingness to recognize a further duty, and courts divide. In J.S. v. R.T.H., the New Jersey Supreme Court held that a spouse must take reasonable steps to prevent child abuse, or warn of the risk, if the spouse “has actual knowledge or special reason to know of the likelihood of his or her spouse engaging in sexually abusive behavior against a particular person or persons.” The court relied on a strong public policy of protecting children from sexual abuse and on a New Jersey statute requiring reporting of child abuse even by spouses.31 Those supports were specific to child abuse. A later New Jersey appellate decision, Sacci v. Metaxas, refused to require a wife to warn a shooting victim that her violent husband might kill him, and a California appellate court in Eric J. v. Betty M. denied the claim of a mother whose young son was abused by her boyfriend, a man previously convicted of child abuse, against his family members who had not warned her of the conviction.32 Even the court that recognized a spousal duty tied it to actual knowledge of a particular risk to particular people. An answer on such facts should say that no § 316 through § 318 relation exists, present J.S. as one court's extension, and note the contrary authority.
Therapists and Tarasoff
The most famous application of § 315(a) involves a therapist and a patient. In Tarasoff v. Regents of the University of California, the California Supreme Court held that when a therapist determines, or under the standards of the profession should determine, that a patient presents a serious danger of violence to another, the therapist incurs an obligation to use reasonable care to protect the intended victim.34 The opinion is easiest to learn in the order the court reasoned: it starts from the no-duty default, states the special-relationship requirement, identifies the relationship that meets it, defines what the duty demands, and leaves open questions on which later courts have divided.
Poddar, a therapist's patient, killed Tatiana Tarasoff. Her parents sued the therapists, the police who had been involved, and the Regents of the University of California as the therapists' employer. The superior court sustained the defendants' demurrers, which are motions asserting that the complaint states no claim even if its allegations are true, and the parents appealed.35 Because the case reached the California Supreme Court on demurrer, the question was whether the complaints, or an amended version, could support a duty.
The parents proposed to amend the second cause of action to allege that Tatiana's death resulted from the defendants' negligent failure to warn her or others likely to tell her of the danger. The defendants answered that they owed her and her parents no duty of care and were therefore free to disregard her safety.36 The court began with a point about duty itself: legal duties are not discoverable facts of nature but conclusory expressions that, in cases of a particular type, liability should be imposed for damage done.37 It then recognized the limit that governs this whole area. When avoiding a foreseeable harm would require the defendant to control another person's conduct or to warn of it, the common law has traditionally imposed liability only if the defendant bears a special relationship to the dangerous person or to the potential victim. The limit exists because foreseeability measures how likely the harm is without saying which of the many people who might have responded must do so; a relationship names the one. Because the therapist-patient relationship met the requirement, the court did not decide whether foreseeability alone creates a duty to protect a potential victim of another's conduct.38 Tarasoff is therefore no authority for the proposition that foreseeability by itself imposes a duty to protect.
The court reached its holding through the relation to the dangerous person. It did not need the kind of relation § 315(b) describes, one that gives the endangered person a right to protection, because the relation between therapist and patient satisfied the requirement.39 California decisions recognizing such a duty had involved defendants in a special relationship both to the victim and to the person whose conduct created the danger, and the court refused to confine the duty to those cases. Decisions elsewhere hold that the single relationship of doctor to patient supports a duty to use reasonable care to protect others against dangers emanating from the patient's illness; a doctor, for example, is liable to persons infected by a patient if the doctor negligently fails to diagnose a contagious disease.40
The relationship carries the duty because of what it puts in the therapist's hands. Confidences shared in treatment mean the therapist may be the only person besides the patient who knows of the threat, and the therapist can act on that knowledge by warning the victim or calling the police. That is the unique position that justifies the control relations generally.41 It also explains why the duty does not depend on literal custody. The court described the discharge of the duty as requiring one or more of various steps, depending on the case, from warning the intended victim or others likely to apprise the victim, to notifying the police, to taking whatever other steps are reasonably necessary.42 Reasonable care can be satisfied by the most limited effort when control itself is not available.
The police defendants stand on the other side of the line. The plaintiffs pleaded no relationship between Poddar and the police that would impose a duty to Tatiana and suggested no other basis for one, so the court upheld the dismissal of the claims against them.43 The contrast shows that the duty comes from the relationship to the dangerous person. The therapists had a treatment relationship with Poddar; on the pleadings, the police had none that imposed a duty.
Predicting violence is hard, and the courts that balance attenuated control list that difficulty among their factors.44 Tarasoff meets the concern in how it defines the trigger. The duty arises when the therapist determines, “or pursuant to the standards of his profession should determine,” that the patient presents a serious danger of violence.45 A trigger keyed to professional standards asks what a careful therapist would have recognized, so a therapist who reaches a reasoned judgment that accepted practice supports does not breach merely because the judgment proves wrong.
The reach of the duty has limits, and courts have drawn them in different places. Virginia and Ohio, discussed above, disagree about the control a relationship must carry. As a casebook observation puts it, taking charge was not the case in Tarasoff itself, so a court that demands that showing will not find a duty on the treatment relationship alone.46 The Ohio approach weighs control as one factor among several.47 A second limit concerns the victim. Some courts decline to stretch the duty to the public at large: in Leonard v. State, an Iowa court held that a psychiatrist owes no duty to members of the general public in discharging a patient.48 That result fits Tarasoff's own wording, which speaks of an intended victim.49 A duty to a named person gives the therapist something definite to do, such as warning her, while a duty to everyone a discharged patient might later meet has no stopping point. Finally, courts treat the duty to protect as subsuming a duty to warn, and whether it reaches further, to committing the patient or retaining custody, remains a separate question.50