Neuropsychiatry of frontal lobe dysfunction in violent and criminal behaviour: a critical review
- aSecure Care Program, Department of Psychiatry, Taunton State Hospital, Taunton, MA, USA, and Harvard Medical School, Boston, MA, USA, bDepartments of Neurology, McLean Hospital, Belmont, MA, Massachusetts General Hospital, Boston, MA, and Harvard Medical School, Boston, MA, USA
- Dr B H Price, Department of Neurology, Mclean Hospital, 115 Mill Street, Belmont, MA 02478, USA
- Received 18 July 2000
- Revised 7 June 2001
- Accepted 2 July 2001
OBJECTIVES To establish the link between frontal lobe dysfunction and violent and criminal behaviour, based on a review of relevant literature.
METHODS Articles relating evidence of frontal lobe dysfunction with violence or crime were collected through a MEDLINE search using the keyword “frontal lobe” combined with the terms “aggression,” “violence,” “crime,” “antisocial personality disorder,” “psychopathy,” “impulse control disorders”, and “episodic dyscontrol.” Reference lists were then searched for additional articles.
RESULTS High rates of neuropsychiatric abnormalities reported in persons with violent and criminal behaviour suggest an association between aggressive dyscontrol and brain injury, especially involving the frontal lobes. The studies reviewed support an association between frontal lobe dysfunction and increased aggressive and antisocial behaviour. Focal orbitofrontal injury is specifically associated with increased aggression. Deficits in frontal executive function may increase the likelihood of future aggression, but no study has reliably demonstrated a characteristic pattern of frontal network dysfunction predictive of violent crime.
CONCLUSIONS Clinically significant focal frontal lobe dysfunction is associated with aggressive dyscontrol, but the increased risk of violence seems less than is widely presumed. Evidence is strongest for an association between focal prefrontal damage and an impulsive subtype of aggressive behaviour.
Reports describing high rates of neuropsychiatric abnormalities among death row inmates, forensic psychiatric inpatients, and other persons with histories of violence have led to assertions that evidence of brain-behavioural impairment may mitigate or excuse criminal conduct.1-4 Frontal lobe dysfunction in particular, has been invoked to explain the actions of some persons charged with, or convicted of, violent crimes, who apparently fail to inhibit impulsive, trivially motivated, or habitual aggression.5-7 But whereas clinical observation and current theories of prefrontal network function suggest that frontal lobe disorders may contribute to violent and criminal behaviour, the strength of this hypothesised association has yet to be established.
This paper evaluates the evidence for a causal relationship between abnormal frontal lobe function and violent crime, based on a review of current research literature. We located articles using a MEDLINE search from 1966 through 2000, combining the keyword “frontal lobe” with the terms “aggression,” “violence,” “crime,” “antisocial personality disorder,” “psychopathy,” “impulse control disorders”, and “episodic dyscontrol.” We then conducted a hand search of relevant reference lists. Articles were selected for review if they contained clinical, laboratory, or neuropsychological test data relating frontal lobe function to aggression, crime, or violence. In this review, we adopt the definition of aggression as any threatening or physically assaultive behaviour directed at persons or the environment. “Violence” refers to actions that inflict physical harm in violation of social norms. We divide our findings under the following headings: (1) studies relating clinical focal frontal lobe disorders to violent behaviour; (2) studies reporting neuropsychological measures of frontal lobe function in aggressive and antisocial subjects; (3) studies of clinical neurological findings in violent and criminal populations; and (4) neuroimaging studies of aggressive and violent subjects. We conclude by assessing the magnitude and specificity of the hypothesised link between frontal lobe dysfunction and violence, and discuss implications for future research.
Focal frontal lobe disorders and violent behaviour
Case studies as far back as 1835 have reported the onset of antisocial personality traits after frontal lobe injury.8Such cases typically involve damage to the orbitofrontal cortex, which clinical observation has associated with “poor impulse control, explosive aggressive outbursts, inappropriate verbal lewdness, jocularity, and lack of interpersonal sensitivity.”9Such gross dysregulation of affect and behaviour may occur while cognitive, motor, and sensory functioning remain relatively intact.10 Blumer and Benson dubbed this orbitofrontal syndrome “pseudopsychopathy,”8 based on similarities to psychopathy—a personality type that, as defined by reliable and valid checklist criteria, is strongly associated with violence and criminality.11-13
Case reports have described a similar syndrome of “acquired sociopathy” in persons who had ventromedial prefrontal injury in adulthood.14-16 Although showing minimal impairments on standard neuropsychological tests of intelligence and executive functions, these subjects display marked deficits in real life tasks demanding judgment, awareness of socially appropriate conduct, and the capacity to assess future consequences.17 Persons with frontal network damage acquired before the age of 8 have also been reported to have adult histories of recurrent impulsive, aggressive, and antisocial behaviour, associated with primary deficits in tests of executive function, poor abstract conceptual thinking, inability to envision another person's subjective experience, and immature moral reasoning.18-20 One report, however, has described two cases of improvement in impulsive and antisocial behaviour after frontal traumatic brain injury in adulthood.21
Large systematic studies on cohorts of war veterans with head injury have also tended to find an association between frontal lobe lesions and aggressive or antisocial behaviour, although the prevalence of actual violent crime seems small. Among German researchers who described personality changes in first world war and second world war veterans with frontal lobe injuries, Kleist found a consistent relation between orbitofrontal lesions and subsequent antisocial behaviour.8 Five patients (3%) in a sample of 144 British second world war veterans with penetrating head injury committed “crimes and misdemeanors,” though all five had damage limited to the frontal lobes.22 Less than 5% of all subjects with frontal lobe injury in a similar study of Finnish second world war veterans had a history of criminal conviction, and only one had committed a violent offence.23
The Vietnam Head Injury Study (VHIS) found that subjects with lesions limited to the frontal lobes tended to show more aggressive and violent behaviours compared with patients with non-frontal head injury and controls without head injury.24 About 14% of subjects with frontal lobe injury engaged in fights or damaged property, compared with about 4% of controls without head injury. The study also found a significant association between increased aggression and focal mediofrontal and orbitofrontal injury identified on brain CT. Reports have also found higher rates of antisocial behaviour (including stealing, physical assault, and sexual comments or advances) in patients with frontotemporal dementia, even when compared with equally cognitively impaired patients with Alzheimer's disease.25 26
All of these studies were retrospective, and most did not adequately control for known violence risk factors. The VHIS study, for example, did not report on prior history of aggression, substance misuse, stability of employment, socioeconomic status, the presence of psychiatric symptoms or disorders other than depression, or criminal charges or other legal involvement. Without such data, it remains unclear how much of the increases in aggressive behaviour found can be specifically attributed to focal frontal lobe injury.
Neuropsychological studies of aggressive and antisocial subjects
A previous comprehensive review of neuropsychological studies by Kandel and Freed (1989) found that “evidence for the association between specifically violent criminal behaviour and frontal lobe dysfunction is weak at best.”27 A subsequent review by Pennington and Ozonoff concluded that comorbid attention deficit hyperactivity disorder (ADHD) most likely accounted for deficits in frontal executive function linked with adolescent conduct disorder, but considered that ADHD might worsen aggression in such cases.28 Table 1 summarises results of relevant neuropsychological studies reported since 1989.29-36
One study reported that errors on a single subtest in a battery of executive function measures correlated significantly with a diagnosis of antisocial personality disorder in a male community sample.29 A small study of subjects addicted to cocaine, all of whom met diagnostic criteria for antisocial personality disorder, found that high violence subjects as a group scored significantly better than low violence subjects on a widely accepted measure of frontal executive functioning.30 By contrast, two studies using a laboratory based procedure designed to elicit aggressive behaviour have correlated decreased performance on executive function tests with increased aggression in community samples of male subjects without neurological, psychiatric, or substance misuse histories.31 32
A prospective study found that low scores on executive function tests significantly predicted self reported aggression in 10 to 12 year old boys with paternal histories of substance misuse, but the results did not control for ADHD.33 The same lead authors conducted a subsequent case-control study of aggression in conduct disordered adolescent females, controlling for ADHD: low executive function scores retained a significant independent correlation with physically aggressive antisocial behaviour.34 In a 1 year prospective study of forensic psychiatric inpatients who had committed a violent crime, low scores on three tests of frontal executive function significantly predicted frequency of aggression, accounting for 57% of the variance.35 Although studies of psychopathic subjects have not demonstrated frontal executive dysfunction,27 one report found that, compared with non-psychopathic criminals, psychopathic criminals showed significant deficits on tests specifically selected to assess orbitofrontal and ventromedial functioning.36
Overall, these neuropsychological studies tend to support a significant association between prefrontal executive dysfunction measured by neuropsychological testing and increased antisocial and aggressive behaviour. In populations with prior risk of antisocial behaviour or aggression, the presence of executive function deficits may have value in assessing the future likelihood of aggression. Studies of psychopathic subjects, however, suggest that standard tests of executive function may miss orbital or ventromedial prefrontal dysfunction relevant to aggression. Maturational delay in the development of the prefrontal cortex,37 or deficient education or socialisation related to psychosocial deprivation, or both29 34 38 may also account for deficits found in neuropsychological test performance linked to aggression.
Clinical neurological findings in violent and criminal populations
A previous review concluded that antisocial subjects had more EEG abnormalities, predominantly anterior.39 Subsequent studies have continued to find abnormal frontal EEG activity, as well as diminished frontal event related potentials, correlating with antisocial personality disorder or histories of aggression.40-42 In a study of adult male drug misusers, for example, subjects rated as “high aggressive” on a self report scale showed statistically significant frontal EEG slowing relative to the “low aggressive” group.43 A case series describing recurrent, severe aggression in mentally retarded subjects attributed their behaviour to frontal lobe seizures, based on phenomenological similarities to frontal ictal automatisms; 60% of subjects had abnormal EEGs, with 20% having focal frontal or frontotemporal findings.44 Case reports have also linked orbitofrontal EEG spiking to violent hallucinations and assaultive behaviour.45 A study of 333 prisoners referred for evaluation after being charged with a violent crime specifically related frontal EEG findings to “habitual physical aggression or explosive rages.”46 After exclusion of subjects with clinical evidence of structural brain damage, 56.9% of habitually aggressive subjects had EEG abnormalities (62.2% frontal), compared with 11.8% of other subjects who had committed a single, isolated aggressive act.
Another report of neurological findings in 31 subjects, who were referred by attorneys in connection with claims of mitigation related to murder charges, found that 64.5% showed “some physical evidence of frontal dysfunction.”47 Signs elicited included snout, suck and grasp reflexes, paratonia, abnormal smooth pursuit eye movements, diminished word fluency, and reciprocal hand movement errors. Examination disclosed three or more signs in 32.3%, two signs in 9.7%, and one sign in 22.6%. A retrospective chart review found that a frontal lobe lesion was the best predictor of involvement in a violent episode among inpatients on a neuropsychiatric unit, accounting for 11% of the variance, ahead of number of inpatient days, seizure disorder, history of alcohol misuse, and affective psychosis.48 A prospective study of frontal lobe function and violence in psychiatric inpatients with mood and psychotic disorders found no significant difference in frontal or other neurological findings between violent and non-violent patients.49 Persistently violent patients, however, had significantly more frontal lobe impairment than transiently violent patients and their behaviour seemed less responsive to environmental factors. A related study by the same authors reported that frontal executive dysfunction was significantly associated with a history of community violence, but did not predict inpatient assaults.50
These studies indicate that clinical signs of frontal lobe dysfunction are prevalent in populations of persons prone to violent and antisocial behaviour. Most of the subjects, however, were either referred by attorneys, or had known or suspected neuropsychiatric disorders, and so do not represent violent criminals in general. The mere presence of EEG abnormalities or frontal neurological signs also does not explain whether, or how, such findings contributed to behaviour at the time of an alleged crime. Two parallel prospective studies do not support a retrospective report that frontal lobe findings predict violent behaviour in inpatient settings. Clinical evidence of frontal lobe dysfunction, however, does seem to be associated with recurrent or persistent aggression.46-50
Neuroimaging in aggressive and violent subjects
Morphometric and functional neuroimaging studies of aggressive and violent subjects have consistently found frontal lobe abnormalities (table 2).51-62 A well designed MRI brain volumetric study compared an antisocial personality disorder group with substance dependent, psychiatric and normal control groups.51 The subjects with an antisocial personality disorder showed significant differences on three measures: more violent crimes, more psychopathic traits; and reduced overall prefrontal grey matter volume. An MRI study of interictal aggression in temporal lobe epilepsy found that, compared with both normal controls and non-aggressive patients with temporal lobe epilepsy, patients with temporal lobe epilepsy with recurrent episodic aggression had statistically significant frontal grey matter reductions; the area of maximum difference involved the left anterior frontolateral cortex.52
Three studies have used single photon emission computed tomography (SPECT) brain scanning to evaluate antisocial and aggressive behaviour.53-55 A comparison of alcohol dependent subjects with healthy controls found that alcoholic subjects with an antisocial personality disorder had significantly greater frontal hypoperfusion than other alcoholic subjects.53 As a group, adolescent and adult psychiatric patients who had physically attacked another person or destroyed property showed significantly decreased prefrontal activity compared with matched, non-aggressive psychiatric patient controls.54 Compared with patients with non-aggressive dementia, patients with aggressive dementia with the same degree of cognitive and psychiatric impairments had significant left anterior temporal and bilateral superior frontal hypoperfusion, but no significant differences in orbitofrontal regions.55
Two positron emission tomography (PET) brain scan studies, which compared forensic psychiatric patients with normal controls, documented decreased frontal cortical blood flow or metabolism associated with “repetitive” and “purposeless” violent behaviour.56 57 Another PET study of “impulsive aggression” found that, compared with non-psychiatric controls, patients with personality disorders (chiefly antisocial, borderline, and narcissistic) showed decreased anterior medial and left anterior orbitofrontal metabolism, which correlated with increased scores on a self reported aggression scale.58 Frontal cortex metabolism did not distinguish patients with antisocial personality disorder from controls.
Another study examined 41 persons charged with murder or manslaughter, who were referred for PET in connection with psychiatric evaluations for criminal responsibility, competence to stand trial, or claims of mitigation.59 Compared with controls (matched for age, sex, and diagnosis of schizophrenia, if present), “murderers” as a group showed statistically significant bilateral prefrontal metabolic decreases during a frontal lobe activation task. A follow up report on the same subjects found that only those subjects blindly rated as lacking histories of psychosocial deprivation had significantly lower overall prefrontal metabolic rates.60 A further study separated these same subjects into “predatory” versus “affective” murderers, based on a forensic typology distinguishing controlled, purposeful aggression to achieve a desired goal from impulsive, emotionally charged aggression.61 Affective murderers had significantly lower prefrontal metabolic activity compared with controls, whereas frontal metabolism in predatory murderers resembled controls. In a PET study of healthy volunteer subjects who were instructed to imagine a scenario involving their own aggressive behaviour, visual evocation of unrestrained aggression correlated with significant focal reductions in ventromedial frontal blood flow, compared with an emotionally neutral scenario.62
The cumulative evidence from these neuroimaging studies points to a strong association between increased aggression and reduced prefrontal cortical size or activity. Although most studies cite bilateral prefrontal abnormalities, others specifically cite left anterior frontal or orbitofrontal findings, as well as non-frontal brain regions. These inconsistencies may reflect variation related to experimental conditions, limitations of imaging technology, or subject selection. Most of the subjects in these studies had known or suspected psychiatric disorders potentially contributing to alterations in prefrontal function. Studies using PET have documented focal decreases in frontal cortical activity associated with various neuropsychiatric disorders, as well as transient mental states, such as induced sadness, and episodes of mood disorder.63-65 The reported reductions in prefrontal size or activity may, therefore, represent a predisposition to affective states relevant to aggressive behaviour, without necessarily signifying an incapacity to avoid actual violent acts.66 The trend in neuroimaging findings, which associates prefrontal abnormalities with “purposeless” or affective aggression, as opposed to premeditated or predatory behaviour, supports this interpretation.
The studies surveyed in this review indicate that clinically significant frontal lobe dysfunction is associated with aggressive dyscontrol. Subjects with both traumatic and neurodegenerative disorders primarily involving the prefrontal cortex display increased rates of aggressive and antisocial behaviour compared with subjects who have no, or non-frontal brain injury. Studies employing neuropsychological testing, neurological examination, EEG, and neuroimaging have also tended to find evidence for increased rates of prefrontal network dysfunction among aggressive and antisocial subjects. Prefrontal network dysfunction seems to be most specifically associated with a recurrent, impulsive subtype of aggression that may contribute to some violent behaviour. Two prospective studies suggest that in populations at risk for antisocial or aggressive behaviour, performance on neuropsychological tests of executive function may have value in assessing future likelihood of aggression.33 35No study, however, shows that disorders of prefrontal cortex predict violent crime.
Methodological problems in this literature include a lack of prospective data, small subject numbers and lack of adequate controls for known violence risk factors. Study samples often draw from groups (prisoners, attorney referrals, or those with severe neurological or psychiatric illness) that do not mirror the general population or even the larger criminal population. Reports describing persons charged with violent crimes tend to cite gross measures of brain function with low specificity and questionable clinical significance, while failing sufficiently to relate the clinical data to the specific aggressive behaviours in question. Standard neuropsychological tests of executive function typically employed in studies of antisocial subjects also may not detect orbitofrontal or ventromedial dysfunction relevant to aggressive behaviour. Although the bulk of research on violent and criminal behaviour points to multiple, probably interacting, causal factors, few studies attributing violent crime to frontal lobe dysfunction adequately address concurrent psychosocial variables such as emotional stress, drug and alcohol misuse, physical and sexual abuse, family breakdown, and poverty.4
Studies of subjects with acquired frontal lobe injury support the expected association of increased aggression with focal orbitofrontal, or ventromedial frontal injury, or both. The neuropsychological literature, however, tends to find increased aggressive behaviour associated with deficits in executive function, which correlate with dorsolateral prefrontal dysfunction.31 32 One hypothesis to account for discrepant localisation data is that orbitofrontal and dorsolateral prefrontal dysfunction contribute to aggressive dyscontrol in different ways. Dorsolateral dysfunction may predominate in persons with comorbid features of fetal or birth related brain injury, developmental learning disorders, attention deficit hyperactivity disorder, substance misuse, and antisocial personality disorder.28 67 Elliot characterised this group as having episodic aggressive dyscontrol rooted in “developmental deviance” manifested by “attention deficit disorder and minimal brain dysfunction,” and associated with neurological soft signs and executive function deficits.68 Resulting educational and social failure likely contribute to aggressive and antisocial life adaptation, as well as to associated poor neuropsychological test performance. Executive function deficits, therefore, may increase the risk of violence via direct effects on impulse control or through associated psychosocial effects, or both, either interactively or independently.
Persons who have clinically evident neuropsychiatric disorders involving focal injury to structural-functional components of the frontal network, particularly the orbital and ventromedial prefrontal cortex, comprise a different group. Retrospective data strongly support a link between the disinhibited type of frontal network syndrome and aggressive dyscontrol. Case descriptions suggest that focal orbitofrontal injury specifically impairs capacities for social judgment, risk avoidance, and empathy that inhibit inappropriate or reflexive aggression. The actual frequency of violent behaviour, however, seems relatively low. Based on results reviewed here,24 48 a reasonable conjecture for the increased risk of violence associated with clinically significant focal frontal lobe injury might be 10% over the base rate for a given population. Confirmation of this estimate must await prospective studies.
In addition to using prospective design, future studies testing the relation between frontal lobe dysfunction and aggression should incorporate controls for known risk factors contributing to violent behaviour. Clinical description of the nature and extent of frontal lobe impairments, coupled with attention to the type (premeditated versus impulsive), frequency, and severity of aggressive behaviour, should help to clarify the brain-behaviour relations involved. Accurate measurement of the increased risk of violence in subjects with prefrontal dysfunction also requires comparison with rates of aggression in appropriate controls. The neuropsychiatric evaluation of violent patients should include clinical assessment for frontal lobe impairment and neuropsychological evaluation of executive functions, particularly in cases involving recurrent, impulsive aggression. Further progress in the study of aggression and frontal lobe dysfunction will require a forensically informed, interdisciplinary approach that integrates neuropsychiatric, neuropsychological, and psychophysiological methods for the study of brain localisation, social cognition, and emotional processing. Better understanding of brain injury and aggression can then inform medical, public health, and social policy interventions to prevent violence.69
We acknowledge Drs Shervert Frazier, Paul S Appelbaum, and Kenneth L Appelbaum, for their encouragement and support. We also thank Dr James Ellison for his helpful review of the manuscript.