An evaluation of homicidal ideation is a routine component of a mental status examination and may be evaluated in more depth in forensic evaluations as a dangerousness risk assessment. The evaluation of dangerousness often includes asking about violent fantasies that may have physical or sexual content. The authors examine the circumstances in which the revelation of violent fantasies to a mental health professional may trigger a duty to warn or protect third parties. Legal cases in which violent fantasies were considered in the context of assessing potential dangerousness are reviewed. The research literature on homicidal and sexually violent fantasies in both non-incarcerated and offender populations is examined. No consistent predictive relationship between violent fantasies and criminally dangerous behavior is reported in the available scientific literature. The authors suggest factors that mental health professionals may consider when assessing whether a particular violent fantasy indicates that a patient's thoughts could give rise to a duty.
In 1969, Prosenjit Poddar informed a psychologist at the University of California at Berkeley that he intended to kill a woman when she returned home from abroad. The woman was Tatiana Tarasoff. Concerned about this threat, the psychologist requested the police to detain Poddar. The police, however, released him because he appeared rational. Nothing more was done and two months later Mr. Poddar killed Ms. Tarasoff. The surviving family sued, claiming that the therapist had a duty to warn them of Ms. Tarasoff's danger. The California Supreme Court, rehearing the case Tarasoff v. Regents of the University of California in 1976, held that when a psychotherapist determines or should determine that a patient poses a serious danger of violence to a third party, the psychotherapist has the duty to take reasonable steps to protect that endangered third party.1 In California, this duty was recognized in Civil Code §43.92, which suggests that a psychotherapist may incur a duty to warn and protect a “reasonably identifiable” victim when a patient “has communicated to the psychotherapist a serious threat of physical violence.”2 The statute states that the duty may be discharged when the psychotherapist makes “reasonable efforts to communicate the threat to the victim or the victims and to a law enforcement agency.”2
The California Tarasoff decisions were not well received by psychotherapists and triggered heated discussion and debate. With the possibility of having to breach confidentiality, psychotherapists expressed concern that patients might opt not to discuss thoughts and fantasies of violence and as a result not get appropriate treatment. Furthermore, therapists pointed out that predicting future dangerousness was beyond their expertise and simply was not possible. During the past decade, several courts have limited the extent to which the “Tarasoff Duty” can be applied,3,4 and clinicians have been encouraged to approach the risk of dangerousness as a clinical, rather than legal, assessment.4,5 In the late 1990s and early part of this century, patients have sued mental health providers alleging that confidentiality was breached when the providers made what they believed was a required “Tarasoff warning.”6,7
Medical students and psychiatric residents are commonly taught to assess for “homicidal ideation” as a part of a psychiatric work-up, which may include mental images and ideas ranging from generalized fantasies of violence, to concrete plans and an intention to kill a specific victim, as in the case of Poddar's intent to kill Tarasoff. Information obtained during these evaluations brings up several questions. Is a detailed fantasy of harming others, including specific individuals, enough to trigger a duty to warn potential victims and the police when there is no stated intention? Is the presence of violent or homicidal fantasies an indicator of future violent behavior? If so, which kinds of fantasies indicate future violence, and which do not? This article addresses these questions, reviewing court decisions and other legal literature pertaining to violent fantasies and potential dangerousness; the duty to protect, warn, or control patients; and confidentiality. In addition, selected relevant clinical and scientific literature on the prevalence and hypothesized roles of fantasies in criminal behavior are examined.
Review of Court Decisions
The American Heritage Dictionary defines “fantasy” as “the creative imagination…. An imagined event or sequence of mental images, such as a daydream, usually fulfilling a wish or psychological need.”8 Leitenberg and Henning,9 in their extensive review of sexual fantasies and their relationship to sexual behavior, suggest that fantasy is synonymous with “daydream,” consisting of mental imagery or “an act of the imagination, a thought that is not simply an orienting response to external stimuli or immediately directed at solving a problem…” (Ref. 9, p 470). Only a fraction of the literature on fantasies has included a definition of the term. Some studies have adopted Leitenberg and Henning's definition, though many others do not explicitly define what is meant by “fantasy.”
For the purposes of this article, a “violent fantasy” is a thought in which the subject imagines physically harming another person in some way, such as by murder, sexual assault, or inappropriate sexual activity. It is distinguished from an intention, in that the imagined violence is not immediately aimed at guiding action, and a delusion, in which the distinction between imagination and consensus reality may be lost. An expression of intention to harm another would be considered communicating a “threat” rather than a fantasy. While not immediately aimed at guiding action, fantasy has been described as serving several clinical functions, including containment and relief of anxiety, as well as a substitution, or preparation, for action.9–11 The perceptions of fantasy as a substitution or preparation for violence are illustrated in the following two case discussions.
Although not in the context of a duty to warn or protect, in Pettus v. Cole et al.,12 a case involving the confidentiality of a disability evaluation, the court explored the issue of dangerousness and fantasies. In his disability evaluation, Pettus stated that he regretted not hurting a coworker in Flint, Michigan, and said, “I think if I ever saw him again, I would really try to kill him. I need to get over that.” One of the evaluators, Dr. Unger, reported that Pettus did not have a history of violence and did not plan or intend to seek out the coworker or harm him. In her report, she indicated that while Pettus was angry, his hostile feelings were normal:
Fantasies of performing violent acts are actually quite common in human experience, and are entertained from time to time by even the most gentle of human beings. Rather than being predictive of future violence, such fantasies actually serve as a psychological “safety valve,” permitting the vicarious, but safe and harmless discharge of strong emotions. Experiencing the fantasy of taking violent revenge often reduces the impulse of performing the behavior. There is a very great and very crucial difference between merely thinking about performing some action, and the physical doing of that act [Ref. 12, p 422].
In her testimony, Dr. Unger suggested that violent fantasies do not necessarily signal future dangerousness, but instead may serve a normal function, substituting for violent behavior and potentially preventing future violent acts.
However, courts may also be persuaded that violent fantasies, especially sexually violent fantasies, may be a warning of future danger. Donald Chapman served twelve years at the Adult Diagnostic and Treatment Center at Avenel Prison for the rape and torture of a 23-year-old woman and was released in November of 1992.13 While imprisoned, he revealed continuous fantasies of sexual torture and mutilation of women, with the intention of committing another rape. Upon his release, the psychologist treating Chapman notified police of Chapman's fantasies and intentions, and Chapman was immediately placed under 24-hour surveillance. It was reported that Chapman even told the surveillance team that he was “obsessed with the thought of his ex-girlfriend, who had ′abandoned' him.”14 In response to the warning made by Chapman's psychologist, the New Jersey attorney general's office sought involuntary commitment of Chapman to a psychiatric hospital, which was granted in early 1993.15 As of April 2000, Chapman was on “Conditional Extension Pending Placement” status at Greystone Park Psychiatric Hospital, awaiting transfer to the Sexually Violent Predator facility in Kearny. The Department of Corrections operates this secure treatment facility, but the Division of Mental Health Services directs treatment of its residents.16 This case illustrates the persuasive effects on the court of a patient's history of violent behavior and graphic violent fantasies. The following section reviews legal cases in which violent fantasies are discussed by the court in the context of a therapist's duty to warn or protect a third party from a potentially dangerous patient, or to control the patient to prevent harm to the third party.
Cases Supporting a Therapist's Duty to Warn or Protect
The U.S. District Court for the district of Kansas considered a duty to control in Mahomes-Vinson v. United States et al.17 Nolan Prewett had an extensive history of violent and sexually deviant behavior, and was treated through Veterans Administration (VA) medical centers. Eight days after discharge from inpatient treatment, Prewett raped and killed two young children, and the victims' family sought damages from the VA psychiatrists who treated Prewett. The court denied the defendants' motion for summary judgment regarding a failure to control Prewett, not only because of Prewett's history of violence and sexual deviance, but also because two weeks before the murders Prewett's wife revealed that the patient “had fantasies of little girls.” No details regarding Prewett's fantasies were recorded; however, the plaintiff contended, “that somebody at the VA hospital should have at least contacted Prewett's wife in pursuit of further information.” The court found that issues of material fact existed as to whether the children could be foreseeable victims of Prewett's fantasies.
In Bardoni et al. v. Kim,18 the Court of Appeals of Michigan briefly considered the relevance of violent fantasies in establishing a person as a “readily identifiable target” to whom a duty is owed, although the patient in this case is described as having delusions rather than violent fantasies. Dr. Soon K. Kim, a psychiatrist, had diagnosed paranoid schizophrenia in Richard Bardoni. After failing to show up for several appointments, Bardoni assaulted his wife and murdered his brother and mother. The trial court found that they were not readily identifiable as potential victims and granted summary judgment to Dr. Kim. The Court of Appeals reversed this decision in part, suggesting that determinations of whether a person was a readily identifiable victim could include, in addition to the medical and psychiatric records of the patient, “the patient's violent fantasies with respect to that person….”18 This suggests that some courts may consider violent fantasies, as well as direct threats and violent acts directed toward a person, as potentially establishing a duty toward that person.
In New Jersey's first case regarding the duty to warn, McIntosh v. Milano,19 the Superior Court denied Dr. Milano's request for summary judgment in a wrongful death action and paid special attention to the violent fantasies of his patient, Lee Morgenstein. During two years of therapy, Morgenstein discussed a variety of fantasies involving his next-door neighbor's daughter, Kimberly McIntosh. These fantasies included being a “hero or an important villain,” using a weapon to threaten and ward off people who may intimidate him, sexual experiences, and fantasies of “magical power and violence” and revenge when he became jealous of her other relationships. While Milano denied knowledge of threats to harm McIntosh, Morgenstein had shot BBs at the windows of the McIntosh home and a vehicle. In addition, Morgenstein had purchased and carried a knife. Morgenstein eventually obtained a pistol and shot and killed McIntosh. The court rejected Milano's motion for summary judgment, holding that whether he could or should have found Morgenstein, based on his fantasies of revenge, to be a danger to McIntosh, and whether a duty to warn existed and was breached, was a question for a jury. At trial, however, the jury found for the defendant, ruling that Milano did not breach a duty to warn the victim.
In all three of these cases, the courts considered the expression of violent fantasies to be an indication that the patients were potentially dangerous, particularly in the context of having a history of violent or threatening behavior. These court decisions suggest that under similar circumstances, absent communications of a threat, a psychotherapist may have a duty to warn or protect potential victims.
While academic supervision may seem an unusual milieu for fantasies potentially to trigger a duty to warn, a U.S. District Court found that supervision of therapists-in-training may place the supervising clinician in a “special relationship” with the trainee. In Garamella v. New York Medical College,20 the court determined that the supervising analyst could have had a duty to control a psychiatric resident from acting on his pedophilic impulses or a duty to warn potential victims.
Dr. Ingram was the supervising analyst for Dr. R. DeMasi, a child psychiatric resident and candidate in the Psychoanalytic Division at New York Medical College (NYMC). During supervision, DeMasi admitted having sexual feelings for children. Whether DeMasi had acted on his pedophilic impulses was unclear. When asked about pedophilic behavior, DeMasi had answered “not really.”20 Ingram subsequently informed DeMasi that he was not a psychoanalysis candidate, but did not advise DeMasi to resign, nor did he inform NYMC that DeMasi's psychoanalytic training had ceased. Several months later, DeMasi sexually assaulted a child patient.
In an earlier decision denying Ingram's motion to dismiss the complaint, the U.S. District Court for the District of Connecticut found that as DeMasi's instructor and analyst, Ingram could have known that DeMasi was dangerous: “The court finds that a self-confessed pedophiliac who intends to practice child psychiatry presents a foreseeable threat of harm to future minor patients…” (Ref. 21, p 41). The court also suggested that Ingram had some ability to control DeMasi: “Ingram could have taken steps to redirect DeMasi's professional development without even compromising the confidentiality of DeMasi's disclosures to him” (Ref. 21, p 41). Ingram later asked for summary judgment, arguing that “if a psychiatrist is forced to warn others of every fantasy revealed by patients in the context of therapy, in the absence of any intent to act on the fantasy, then no patient will seek treatment for pedophilia or any other potentially stigmatizing affliction” (Ref. 20, p 175; quotes in original). Summary judgment was denied, and at trial Ingram was found negligent for failure to warn others of DeMasi's pedophilic impulses.22
In the prior cases discussed in this section, revelations of violent fantasies occurred in the context of a previous history of aggression or violence. The Garamella decisions suggested that compared with violent fantasies involving adults, pedophilic fantasies could signal a greater degree of dangerousness. Fantasies involving sexual behavior with children may establish a duty to protect potential future victims, even when there is no expressed intention or a history of acting on the fantasies.
Cases Limiting a Therapist's Duty to Warn or Protect
Some cases that limit a therapist's duty to warn or protect emphasize the importance of confidentiality. Unlike many states that have adopted a duty to protect third parties from patients who make threats, the Court of Appeals of Florida rejected adopting a duty to warn in Boynton v. Burglass.23 While this case was not precipitated by the revelation of violent fantasies of a patient, the court seemed aware of the potential impact of a duty to warn or protect third parties upon psychotherapists' ability to assess such fantasies. In affirming Dr. Burglass's motion to dismiss the complaint, the court commented, “by the very nature of psychotherapy, the patient is encouraged to freely vocalize his fantasies, repressed feelings, and desires…” (Ref. 23, p 451), suggesting that mandated warnings could interfere with treatment by limiting confidentiality. The language used by the defense in this case was very similar to the argument presented by the defense in Garamella,20 but with very different outcomes. In Garamella, the court denied the motion for summary judgment.
However, some courts have rejected the Tarasoff warning without referring to the importance of confidentiality. In Doyle v. United States,24 the U.S. District Court dismissed an action for wrongful death in which the plaintiff claimed that an Army psychiatrist negligently failed to diagnose dangerousness in a patient and failed to warn potential victims. While the patient, Carl Carson, was hospitalized for an evaluation of his remarks regarding homicidal thoughts, the hospital treatment team determined that he “verbalized his aggressive fantasies to shock people… .”24 Two days after his discharge, Carson “acted out a recurring fantasy by shooting and killing” a security guard (Doyle). In the criminal trial, it was revealed that the patient believed the government was “corrupt and evil” and that he had had fantasies of “committing crimes and being involved in gun fights with the police.”24 Significantly, Carson had no other history of violent acts or discipline problems. The court rejected the plaintiff's claim, finding that Carson's victim was not “foreseeable and identifiable,” and that his statements and fantasies were “not enough” to establish his dangerousness, especially considering that he had not committed a previous act of violence.
In an unusual case, White v. United States,25 the U.S. District Court found that St. Elizabeth's Hos-pital did not violate a duty to warn Genoa White of her husband's fantasies of harming her, even though her husband had had a history of violence. This decision was subsequently upheld by the D. C. Court of Appeals. Dwayne White was admitted to St. Elizabeth's Hospital after being found not guilty by reason of insanity of the murder of a police officer. He had an extensive history of “uncontrollable violence” and an “explosive personality.” During weekly individual psychotherapy sessions with a clinical psychologist, White revealed a fantasy of harming his wife, Genoa White, with a gun. This fantasy was discussed several times, but never reported to the hospital administration because the doctor did not consider the fantasy to be a threat, and because White did not express an intent to harm his wife. White was allowed limited privileges to walk the hospital grounds unaccompanied, and he had left the grounds on several occasions. Six months after revealing fantasies of harming his wife, White attacked her with a pair of scissors.
The wife filed suit against the hospital, alleging that the hospital had failed to warn her about her husband's thoughts of violence against her and alleging negligence in allowing her husband to leave the grounds. Quoting Doyle,24 the court rejected the plaintiff's claim and found that the patient's fantasies did not represent an actual threat: “Such statements are commonly expressed to psychiatrists and merely pose but do not answer the difficult question of whether or not danger is actually present” (Ref. 24, p 1289). The court considered the opinion of the patient's psychologist that Mr. White “was able to distinguish between fantasy and acting on his fantasy” and concluded that the hospital did not have a duty to warn: “…although St. Elizabeth's may be obligated to take reasonable steps—including warnings—to protect foreseeable and identifiable victims from a serious danger of violence presented by its patients, Dwayne White's fantasy presented no such danger to Genoa White” (Ref. 25, p 102).
White was unusual in that, despite Dwayne White's history of violence, the U.S. District Court distinguished his violent fantasies from his violent act. This contrasts with the New Jersey Superior Court in McIntosh, in which the court found that a jury could find that a duty to warn existed, based on a patient's fantasies of revenge and in the context of a history of violence. It is also important to note that while the Boynton court emphasized the importance of confidentiality, neither the Doyle nor the White courts discussed issues of confidentiality in regard to the duty to warn. In these cases, the presence of violent fantasies was simply not sufficient to trigger a duty to warn.
In a case involving an alleged breach of confidentiality, a psychologist was found negligent after having given a Tarasoff warning.6 During a fitness-for-duty evaluation, police officer Gordon Garner III, informed clinical psychologist Anthony Stone that he had “experienced a ′homicidal ideation'—or vivid fantasy” of killing his captain, himself, and several other officers and public officials.26 Concerned about Garner's “vivid images of shooting his supervisor at work,” Stone consulted a lawyer with the Georgia Psychological Association, who advised him that he had a duty to warn Garner's supervisor. After waiting two weeks, Stone learned that Garner had ended treatment with his former clinical psychologist, after which Stone informed Garner's superiors of his violent fantasies. Garner was demoted and eventually fired. Garner sued Stone for negligence due to a breach of confidentiality and defamation of character. The plaintiff's lawyers argued that their client's thoughts were “just fantasies” and that Garner had never intended to harm anyone. In addition, Georgia did not have a duty-to-warn requirement, and the plaintiff's attorney argued that even Stone did not believe that any danger was “imminent in nature.” The DeKalb County jury found against the psychologist and awarded Garner monetary damages for the breach of confidentiality.
At the time of this writing, another case alleging malpractice for breach of confidentiality involving disclosure to law enforcement of perceived threats by a patient is pending. In April 2002, in Orange County, California, a teacher who was arrested and charged with making terrorist threats filed a medical malpractice lawsuit against the medical center where he had sought psychological help.7 The lawsuit alleged that the teacher presented himself to South Coast Medical Center, stating that he might go “postal.” Hospital personnel called law enforcement, and the patient was arrested and spent two months in jail. While the charges against the teacher were eventually dismissed, lawsuits against the county, the local sheriff's department, and the medical center remain outstanding.
Summary of Court Decisions
The law regarding a therapist's duty to warn or protect third parties from the therapist's patient varies from state to state. Of the eight cases summarized herein6,17–20,23–25 four suggested that a duty to warn or protect third parties might exist when a patient reveals fantasies with violent or pedophilic content. In these four cases, there was either a known history of violence17,19 or information suggesting some aggressive or sexually inappropriate acts in the past.18,20 One of the cases also discussed the use of evaluations in supervision21 as an alternative to issuing a warning to a potential victim by controlling the potential perpetrator. Of the four cases that limited a mental health professional's duty to warn or protect, three6,24,25 described patients who revealed some form of violent fantasy, but only one noted an established history of violence,25 whereas the other cases did not detail a known history of violence.6,24 Two of the cases contrasted fantasies of violence with threats or intentions to harm a person.6,25 In one case, the imagined victim was not harmed.6 In two cases,18,23 violent fantasies were not documented in the facts of the case, but were commented on by the courts in relation to whether a duty to warn or protect could be found to exist.
Based on this review, a history of violence appears to be not only an important risk factor for future violent acts, but also an important factor in the courts' consideration of whether a duty to warn or protect may exist, regardless of the violent fantasies of the patient. It is noteworthy that none of the cases specifically discussed whether an appropriate risk assessment was performed, although at least one case noted that a defendant physician had failed to review a patient's medical records in addition to not evaluating the patient's fantasies about children.17 Whether violent fantasies per se are a signal for future dangerousness is examined in the next section.
Studies of Violent Fantasies in Non-criminal Populations
Inherent in whether a violent fantasy should trigger a duty to warn or protect is the question of whether these fantasies predict future dangerous behavior. No study of normal (more specifically, non-incarcerated) individuals has adequately addressed this key question. Most studies have examined violent fantasies and imagery in a normal or control population of undergraduate and graduate students.
In a study directly examining the prevalence of homicidal fantasies, Kenrick and Sheets27 reported that 68 percent of undergraduate students endorsed having at least one homicidal fantasy. Approximately 30 percent of men and 15 percent of women endorsed having such fantasies frequently. In general, men endorsed having more frequent and prolonged fantasies, and those longer fantasies tended to include strategies. For both men and women, the frequency of homicidal fantasies correlated positively with reports of physical conflicts. Whether subjects had fantasies inspired by experiences, or sought experiences driven by fantasy, was not examined in their study.
Homicidal fantasies were also specifically studied by Crabb,28 who adopted Kenrick and Sheets'27 procedures, and found that 47 percent of an undergraduate sample endorsed having a recent fantasy of killing someone. Of particular interest is that 3.2 percent reported some element of thrill-seeking as a trigger for a homicidal fantasy: “…characterized by statements such as ′I wanted to know what it would be like to kill someone' ” (Ref. 28 p 230). The author noted that a minority of subjects (2%) indicated the use of some sort of torture device to kill someone in their fantasies. In both of these studies, “fantasy” was not specifically defined, and study participants were asked to describe recent “thoughts about killing” another person.
In two of the legal cases,17,20 sexual fantasies about children were discussed as a potential signal of dangerousness that could require the therapist to take action. Paraphilias, such as pedophilia, are mental disorders defined by the presence of recurrent intense fantasies and behavior in which the sexually arousing object deviates from normal sexually arousing stimuli (e.g., in pedophilia, the deviant sexual elements are children; in sadism, the sexually deviant element is causing suffering). While the prevalence of pedophilic fantasies is unknown, it is likely that such fantasies are more common than pedophilic behavior.29 Fagan and colleagues29 suggest that factors that precipitate pedophilic behavior include the presence of a mood disorder, psychosocial stressors resulting in a loss of relationships or status, and alcohol abuse. Although studies of sexual arousal are beyond the scope of this review, some of the literature on fantasies of sexual violence and sexually inappropriate targets among non-incarcerated individuals is particularly relevant in addressing possible dangerousness.
In a pilot study on the sexual fantasies of men (specifically, French-speaking Canadian men), one-third (33%) endorsed having a fantasy of raping a woman.30 In this study, “fantasy” was defined as organized “mental pictures” that required effort to maintain and were known to be unreal. Other endorsed fantasies included bondage (41%), humiliation (15%), and beating (11%) of a sexual partner. This study dealt only with reported fantasies with no evaluation of the behavior of the subjects. In addition, 64 percent of respondents endorsed fantasies of “initiating a young girl.” While not a report of fantasy per se, Templeman and Stinnett31 also reported sexual thoughts about young persons in a presumed “normal” population. Five percent of their sample of undergraduate men endorsed a desire for sexual contact with girls under the age of 12 and 12 percent for contact with girls aged 13 to 15 years. Over half of the subjects reported engaging in some form of sexual misconduct or potentially criminal sexual offense, the major offenses being voyeurism and frottage.
Possible links between fantasy and behavior have been investigated in studies that included some probe into sexual or aggressive behavior. Greendlinger and Byrne32 reported that 35.7 percent of their sample of undergraduate men endorsed having a fantasy of raping a woman. In this study, “fantasy” was not clearly defined, but was equivalent to “imagery.” Additional sexually violent fantasies included bondage (66.1%), using force to subdue a woman (63.7%), using force for sex (55.9%), and wanting to hurt a partner during sex (44.6%). The phrasing of the questions was apparently important, considering “force to subdue,” “force to have sex,” and “rape” as roughly equivalent acts of sexual aggression, although they were endorsed at different frequencies. While the study did not control for social desirability, one may argue that “using force for sex,” sounds less aggressive and prohibitive than “rape,” which may explain the discrepancy in frequency. The investigators examined the reported fantasies in relationship with the subjects' reports of coercive behavior (ranging from lying and arguments to force and rape) and found a correlation between the total coercion score with the fantasy measure. The authors suggested that sexually aggressive fantasies “serve as models” for aggressive acts (i.e., these fantasies may serve to “prepare for action”).32
Using a sexual inventory, Person and colleagues33 examined the sexual experiences and fantasies of 193 undergraduate men and women from data obtained in the early 1980s. Five percent of women and 31 percent of men reported recent fantasies (within three months) of “forcing a partner to submit.” In addition, one percent of women and seven percent of men endorsed fantasies of “whipping/beating a partner,” while six percent of men had imagined torturing a partner. Regarding cumulative sexual fantasies, 9 percent of women and 24 percent of men had ever fantasized forcing a partner to submit, while 12 percent of men had imagined torturing a sexual partner. This study was repeated 10 years later,34 finding 44.6 percent of men and 22.6 percent of women had fantasies at some time in their life of forcing a partner to submit, 22.2 percent and 3.8 percent, respectively, had fantasies of whipping or beating a partner, and 9.3 percent and 1.9 percent, respectively, had fantasies of torturing a sexual partner. The authors found that the most frequently reported sexual fantasies tend to be also the most frequently reported sexual experiences. While there was a relationship between reports of fantasy and reported behavior, the link was weak, with less than one in five of the subjects who reported fantasies (whether violent or not) actually acting on them.
In addition to paraphilias, patients with other mental disorders may present with fantasies of violence toward others, and most of the legal cases reviewed herein consisted of violent fantasies in mentally ill patients. The MacArthur Study of Mental Disorder and Violence35 is one of the major studies of risk assessment and mental illness, and includes a component in which potential relationships between violent thoughts and fantasies were investigated in patients hospitalized for a mental disorder and later violent behavior.
For this component of the study, the Schedule of Imagined Violence (SIV) was created, consisting of eight questions, starting with “Do you ever have daydreams or thoughts about physically hurting or injuring some other persons?” The other questions assessed recency, frequency, chronicity, type of harm, target or generalized focus, changes in seriousness of harm, and proximity to target. Subjects who endorsed having violent thoughts within two months of their hospitalization were designated SIV+. Approximately one-third of the subjects were designated SIV+ at the time of hospitalization, while 6 in 10 endorsed violent fantasies within a year of discharge. In general, the researchers concluded that the presence of violent fantasies at the time of hospitalization was related to a statistically higher likelihood of violent behavior during the first 20 weeks and one year after discharge. This was especially true in patients who reported continued violent fantasies after discharge and who experienced greater symptom severity during hospitalization, compared with patients who did not endorse recent violent fantasies. However, the relationship is described as “not strong,” and the authors question whether the difference in violent behavior between SIV+ (26% engaged in a violent act), and patients designated SIV− (16%) could be considered clinically significant. Despite this remark, the researchers emphasize that in addition to substance abuse and anger, the presence of violent fantasies in hospitalized mentally ill patients is a potential target for treatment intervention that may ultimately reduce the risk of future violence.35 The nature of the violent fantasies was not described.
Considering the high prevalence of homicidal and sexually aggressive fantasies among non-incarcerated and presumably “non-criminal” individuals, one may wonder to what degree such fantasies could be considered deviant.9 Fantasies of murder were fairly common in “normal” Western European and American populations, while a range of fantasies of aggression and sexual violence were less common, but were by no means rare. Sado-masochistic behavior and fantasies were commonly reported,33,34 as well as fantasies of rape30,32 and sexual activity with very young, if not minor, partners.30,31 While some violent fantasies were relatively common, only a minority of individuals reported fantasies of torture, beating, and degradation of their sexual partners.33,34 It is possible that because such fantasies may be more disturbing to experience, let alone endorse, there would be both conscious and unconscious social pressures inhibiting a subject's self-report. As would be expected, several of the studies of “normal” populations also noted deviant, and in some case criminal, behavior in their subjects, suggesting that fantasy had some correlative relationship with behavior,31,32,34 although this was not exclusive to violent fantasies. In individuals who were hospitalized for mental disorders, one-third endorsed violent fantasies at the time of hospitalization, while over one-half endorsed such thoughts at any time one year after discharge.35 In this population, the presence of violent fantasies at hospitalization predicted a greater chance of violent behavior within one year of discharge, although the relationship was not considered to be strong.
Existing studies have clearly demonstrated that many more people have homicidal and sexually violent fantasies than act on them. The relationship between violent fantasy and behavior in these studies was correlative at best, and no means of identifying the minority of persons with violent fantasies who are at risk of acting on their fantasies was suggested by any of the studies. The next section reviews violent fantasies among incarcerated populations, sexually violent offenders being the most often studied group.
Violent Fantasies in Criminal Populations
Many studies of violent fantasies in criminal populations examined the fantasies of sexually violent and sadistic offenders, the most extreme being sexually sadistic serial murderers. Recent reviews of serial homicide offenders include those by Myers et al.36 and Schlesinger.37
Most studies of sadistic and serial offenders are retrospective descriptive studies based on archival material such as police reports, clinical and legal reports, and book-length biographies, although some studies did include interviews of the subjects. In the FBI studies of 36 sexually motivated murderers (29 of whom had committed multiple murders), “daydreaming” during childhood, adolescence, and adulthood was endorsed by over 80 percent of the offenders.38,39 Sixty-one percent admitted to rape fantasies as youths, of whom half had had the first rape fantasy between the ages of 12 and 17. These studies specified “fantasy” to be an “elaborate thought” or “cognitive preoccupation” with “origins in daydreams.”
MacCulloch et al.40 retrospectively selected a group of 16 male patients who had diagnosed “psychopathic disorder” and who had been convicted of either a sexual offense, or a crime with “sexual connotations.” Of the 16 individuals, 13 described rehearsing their crimes in fantasies and described a sense of pleasure or sexual arousal associated with the fantasies, often accompanied by masturbation. Of these 13 offenders, 9 reported rape fantasies, 8 had fantasies of killing, and 7 reported the use of torture, mutilation, or bondage in their fantasies. Prentky et al.41 compared serial sexual murderers (defined as three or more sexually motivated homicides) with murderers who had committed one sexual homicide. Examining previous interviews and archival materials and coding specific information by questionnaire, the authors found a significantly higher prevalence of violent fantasy, fetishism, and cross-dressing, and organization of the crime scene in the sample of serial sexual murderers. Both MacCulloch et al.40 and Prentky et al.41 have suggested that a progression from fantasy to behavioral try-outs represents a loosening of the restraints that prevent individuals from indulging in their sadistic fantasies. Once this occurs, they suggest, the danger of progression to murder (whether accidental or intended) should be strongly considered. However, as Prentky et al. also point out: “The presence of fantasy alone is a relatively poor harbinger of future conduct” (Ref. 41, p 891).
Langevin and colleagues42 reviewed the literature on deviant sexual fantasy in sex offenders and compared the sexual fantasies of several types of sex offenders, non-sexual offenders, and non-offenders. Using the Clarke Sex History Questionnaire, deviant age/gender fantasies (i.e., fantasies about girls under the age of 16 or boys) were endorsed significantly more by the sex offender groups than by either control group. When fantasies about physically mature females were included, there were no significant differences among the groups. The prevalence of any deviant sexual fantasies with females (e.g., peeping, rape, sado-masochism) ranged from 40 to 82 percent among the sex offender groups, while 64 to 77 percent of the control groups endorsed these fantasies. Discriminant function analysis revealed that no group could be identified from any specific fantasy measure. Langevin et al.42 noted that the vividness and duration of the fantasies was not explored and suggested that while fantasy may not be “central to sexual deviance,” those offenders who do have sexually deviant fantasies may be a distinct group from those who do not. The authors also noted that no causal relationship had been established: sexually deviant experiences may have given rise to sexually deviant fantasies, rather than vice versa. In addition, this study did not distinguish “violent” (e.g., rape) from “non-violent” (e.g., peeping) sexually deviant fantasies.
The roles of fantasies in sexually motivated crimes have also been investigated in juvenile offenders. Myers and his collaborators43–45 collected data from a series of 16 juveniles incarcerated for sexual assault and homicide. Seventy-one percent reported a history of sadistic fantasies, and one-third met criteria for sadistic personality disorder according to the Schedule for Nonadaptive and Adaptive Personality (SNAP). Myers noted that the prevalence of sadistic fantasies in this series was closer to the multiple, rather than the single, sexual murderers of the study by Prentky et al.41
Comparing Violent Fantasies in “Normal” Community Versus Incarcerated Populations
While in this review we did not attempt a complete meta-analysis of studies of violent fantasies, a rough comparison of the prevalence of violent fantasies in offender and non-offender groups is possible. The type or duration of violent fantasies was not consistently examined and in some cases was not considered at all. In addition, few studies defined fantasy, although those that did described elements of mental imagery or imagination; often, the definition of fantasy was to be assumed by the reader. In general, however, some studies used similar enough populations, used similar instruments, or studied similar enough fantasies that their results could be compared.
Sixty-one percent of the group of sexual murderers studied by the FBI endorsed fantasies of rape,39 while specific rape fantasies in non-offender populations were much less common, being endorsed by about one-third of male subjects.30,32 Of interest, the rate of reports of rape fantasies in normal subjects is similar to that in offenders if fantasies of bondage and the use of force to subdue a partner sexually32 are included (approximately two-thirds of non-offenders endorsing such fantasies). In addition, up to half of undergraduate men and a quarter of undergraduate women had endorsed fantasies of sexual coercion.33,34
In the study by Langevin et al.,42 almost half of heterosexual pedophiles had sexual fantasies involving girls aged 13 to 15, while almost 40 percent had sexual fantasies about girls under the age of 12. There was a range of endorsement by “normal” control groups for fantasies of children or young partners, depending on how the question was worded. Up to two-thirds of men in one sample endorsed fantasies about “young girls,”30 while much fewer (5–12%) endorsed a desire for or fantasy of having sexual contact with girls specifically aged 15 and younger.31,42
Sadistic murderers were described in several studies as spending much of their time preoccupied by violent fantasy.40,41 Only Kendrick and Sheets27 reported on the duration of homicidal fantasies in their undergraduate sample. Most men reported that their fantasies lasted at least a “few minutes,” while 18 percent of men endorsed fantasies lasting “a few hours.” The duration of homicidal fantasies in women was significantly shorter than in men.27
Most of the studies of criminal populations examined a rare and extremely disturbed group of sexually sadistic criminals, some of whom were serial murderers. The fantasies observed in these subjects may not represent the fantasies that would be observed in a more broadly defined offender population such as the general population in a prison. It is important to note that some subjects in the “normal” populations in several studies were discovered to have engaged in sexually violent, coercive, or criminal behavior.31–34 While it is often suggested, only a few studies of offender and non-offender populations have demonstrated a relationship between fantasy and behavior, and this relationship has been correlative, not causal.31,32,34,40,41 While not a study of specifically “violent” fantasies, another study was unable to distinguish among types of sexual offenders and non-offenders based on the presence or absence of deviant sexual fantasies.42 In their review of the literature on sexual fantasies, and remarking on violent sexual fantasies in particular, Leitenberg and Henning9 concluded: “there is no evidence that sexual fantasies, by themselves, are either a sufficient or a necessary condition for committing a sexual offense” (Ref. 9, p 508).
Recommendations for Mental Health Professionals Evaluating Individuals Who Report Violent Fantasies
Taken together, the data from available studies suggest that the presence of violent fantasies does not specifically signal potentially violent behavior. Further, the research suggests that violent fantasies are present in a large number of “normal” individuals who presumably have not acted criminally based on these fantasies. There is insufficient scientific evidence that violent fantasies should be considered absolutely predictive of future dangerousness. Even though not communicating a threat, a patient with violent fantasies could still pose a danger.46,47 The nature and quality of violent fantasies and the degree to which a person is preoccupied with them may prove useful in helping the clinician determine whether a patient might be imminently dangerous and whether a duty to warn or protect is warranted. If a patient discloses a violent fantasy, further exploration may be helpful to identify foreseeable victims and to assess the potential of acting on the fantasy.41,42,48
In assessing whether a patient's violent fantasy might trigger a duty to warn or protect, the therapist may find it helpful to consider the following factors. First, elements in the patient's history associated with an increased risk of violence should be considered, such as the presence of a major mental disorder, a history of violence, substance use, and impaired impulse control.5 Although the relationship was not considered to be a strong one, the presence of violent fantasies in mentally disordered hospitalized patients was related to violent behavior within a year of discharge.35 Courts that supported a mental health professional's potential duty to warn or protect potential victims considered a history of past violence to be especially important (see section on Review of Court Decisions).
The nature of the patient's fantasy could also be explored. Brief fantasies of murdering someone in anger are common, as are many milder sado-masochistic sexual fantasies, such as spanking and bondage (see section on Studies of Violent Fantasies in Non-criminal Populations). These common fantasies usually lack actual intentions, even if plans have been considered, while some sado-masochistic fantasies are enacted with consenting adults. In contrast, fantasies of sexual torture are uncommon, and further exploration may be needed to assess the patient's attitude toward the fantasy, such as whether the fantasy is ego syntonic or dystonic.
Finally, the therapist may also consider specific behavior and thoughts related to the fantasy, such as the level of sexual arousal involved and particular individuals included. A consideration of “attack-related behavior” may be helpful, as discussed by Borum and colleagues46,47 in their work regarding “fact-based” threat assessments. In addition to the patient's attitudes related to the fantasy, the clinician may want to assess the capacity of the patient to act out the fantasy, whether the patient has expressed an intention or not. A patient who has begun to act out paraphilic fantasies, whether violent or not, may be especially ominous (see section on Violent Fantasies in Criminal Populations). In the case of pedophilic fantasies, determining the presence of mood disorders, substance abuse, loss of relationships, level of distress, and compulsiveness toward action may be important.29 Loss of relationships or other stressful circumstances and how an individual usually copes with such events could also be assessed. It may be helpful to ask if the patient has discussed the fantasies with other people and if so, whether those people expressed concern and tried to discourage such fantasies.46,47 A thorough assessment of the risk factors for dangerousness, the nature of the fantasy, and the attitudes and behavior related to the fantasy may help distinguish whether the fantasy represents a serious danger. If a mental health professional decides that a warning of some form or some other action is necessary, it has been suggested that the patient be involved in the decision and that the warning use the minimum disclosure required for adequately warning or protecting the possible victim(s).4,49
The inconclusive research findings, the lack of consistent research methodologies and legal standards and the inconsistent court decisions involving violent fantasies make it difficult to provide absolute guidelines for mental health professionals. Future research and the attention of the mental health community to these issues are important, because the risks of failing to protect potential victims or inappropriately breaching a harmless patient's confidentiality are significant. Further research into the relationship between a variety of violent and sexual fantasies and their corresponding behavior may be helpful in delineating relationships between fantasy and future dangerousness. Existing studies report on the prevalence of endorsed violent fantasies, and future studies that examine other qualities of a fantasy, such as vividness, compulsivity, duration, and endurance may reveal distinctions between normative experience and pathology. While cross-sectional and retrospective descriptive studies of offenders are a useful starting point, prospective cohort studies of non-criminal populations have the potential to reveal factors that separate individuals who eventually commit a crime related to fantasy from those who do not. There are numerous challenges that should be addressed by future research. How does one determine which properties of fantasy can be operationalized and measured, and which instruments are reliable and valid in the measure of these properties? Additional key questions involve the determination of appropriate populations for study and appropriate control populations. Finally, investigators must deal with the ethics-related problem of how to address the potential duty to warn and protect others when violent fantasies are revealed in the course of research.
Summary and Concluding Remarks
Violent fantasies are an especially nebulous form of homicidal ideation, compared with direct threats, intentions, and plans to harm someone. These fantasies are most likely to exist on a continuum in their level of detail, intensity, duration, and the emotion involved. A wide variety of psychological functions have been proposed for violent fantasies, including normative means of coping with aggression, anger, and sexual arousal9; pathological replacement of reality and interpersonal connections; and imaginary practices of intended actions, whether harmful or not. Studies of sexual and violent fantasies in “normal” individuals suggest that these fantasies are extremely common (as described in the section on studies of the non-criminal population), and that the majority of people do not act based on these types of fantasies.
Psychiatrists and other mental health professionals are sometimes asked to assess the dangerousness of an individual. A request for this type of evaluation may be instigated by the revelation of an individual's violent or otherwise disturbing fantasy. This complex type of evaluation, requiring the mental health professional to apply an inadequate scientific body of information to attempt to predict future behavior may be further complicated by the need to determine whether a patient's fantasies warrant identifying, warning, or otherwise protecting a third party from the patient. A mental health professional attempting to carry out the Tarasoff duty will not necessarily be protected from charges of breaching confidentiality, especially as more limits are set by the courts.4
Dr. Gellerman would like to thank Charles Scott, MD, and Barbara McDermott, PhD, Department of Psychiatry, Forensic Division, University of California at Davis, for inspiration and helpful suggestions.
- American Academy of Psychiatry and the Law