Abstract
One decade ago, important reviews were published regarding the biological basis of personality disorder in general [1] and psychopathy in particular [2]. At that point, there had been few reports of relevant neuroimaging studies. This paper will primarily examine recent studies of neuroimaging in psychopathy.
Early descriptions of psychopathy may include Pinel's [3] ‘manie sans delire’, which was characterized by impulsive and violent acts, and Rush's [4] ‘derangement of themoral faculties’, whichwas characterized by irresponsibility without shame. Pritchard [5] described ‘moral insanity’ in which the intellect was intact, but damage was assumed to the moral qualities. A century later, Partridge [6] introduced the term, ‘sociopathy’.
A most influential contribution was made by Cleckley [7] in his book The Mask of Sanity. He used the terms sociopathy, psychopathy and antisocial personality interchangeably. His description of a group of individuals with average or superior intelligence, free of psychosis, cold and callous, abnormally aggressive and irresponsible, and unable to make enduring relationships or to learn from experience became the stereotype incorporated into the Diagnostic and Statistical Manual of Mental Disorders (DSM) and the International Classification of Diseases (ICD) systems as antisocial personality disorder (APD)/dissocial personality disorder, respectively.
Cleckley's [7] contribution also formed the basis of the Psychopathy Checklist – Revised (PCL-R) [8]. This widely used instrument is categorical, and has been used as both an additional and an alternative diagnostic tool. It also provides a dimensional score that is useful in research and clinical practice. It is a 20-item clinical and actuarial rating scale, which yields a two-factor structure. Subscore factor one (F1) is said to reflect ‘core’ psychopathology, including interpersonal callousness, domination-seeking and emotional detachment. Subscore factor two (F2) represents a history of antisocial behaviour, impulsiveness and violence.
Dolan [2] reached tentative conclusions about neuropsychological, genetic, biochemical and electrophysiological correlates of psychopathy. The neuropsychology of psychopathy is a highly specialized field, and recent advances have been reviewed [9–11].
The neurophysiologic investigation of psychopathy, where this refers to electroencephalogram (EEG) and autonomic system studies, has not been studied extensively over the last decade. However, recent EEG studies with [12] and without [13] event-related electroencephalographic potential (ERP), combined with neuropsychological testing, identified disturbances in frontal functioning. Important work [14] using neutral and emotional words and ERP suggested that psychopaths process and respond to emotional words as if they were neutral words, devoid of affective significance. More recent work with ERP [15] has supported the abnormal processing of semantic and affective verbal information. The less common use of EEG, ERP and autonomic study techniques in recent years, at least in part, reflects their limited spatial ability, and the advances in neuroimaging methods.
Genetic studies of the last decade have substantiated and extended previous findings. The new studies have included adoption [16], and twin studies of adult females [17], 7–16 year olds of both sexes [18] and 16–18-yearold males [19]. All found a significant contribution to antisocial behaviour from both genes and environment. In the twin study of males [19], two dimensions, impulsivity/ aggressivity and detachment/callousness were considered separately, and both were associated with additive genetic factors. However, there was some evidence that these dimensions were etiologically independent. Rhee andWaldman [20] conducted a meta-analysis of 51 twin and adoption studies dealing with antisocial behaviour, and found additive genetic influences (0.32), non-additive genetic influences (0.09), shared environmental influences (0.16) and non-shared environmental influences (0.43).
Dolan [2] reviewed 13 studies of serotonin (5-HT) function, and found reduced activity appeared to correlate with the behavioural and affective states of psychopathy. The studies of the last decade have been built on this foundation. Two come from Dolan's own department. In the first study [21], offenders with aggressive personality disorder and controls underwent assessment of 5-HT function (prolactin response to d-fenfluramine), neuropsychological testing and magnetic resonance imaging (MRI) estimation of frontal and temporal lobe volumes. There were no significant group differences in regional brain volumes. Impulsivity appeared to be contributed by both impaired neuropsychological function and 5-HT function, while aggression appeared to correlate only to impaired neuropsychological function. In the second study [22], personality disordered offenders were again examined using the d-fenfluramine challenge, and the screening version of the PCL was used to quantify the components of psychopathy. Serotonin function did not correlate with psychopathy as a single measure. However, arrogant-deceitful scores correlated positively with 5-HT function, and impulsive-antisocial scores correlated negatively with 5-HT function. Two studies of serotonin activity have come from a Swedish group [23–25], the second being a replication of the first. Violent sexual offenders were examined using the PCL-R and estimation of the cerebrospinal fluid (CSF) metabolites of dopamine (HVA) and serotonin (5-HIAA). The results linked outwardly directed aggression in psychopathy to lowserotonin and high dopamine function. Thus, with the exception of the Dolan et al. [21] study in which impulsivity, but not aggression was associated with serotonin dysfunction, the association of aggression and serotonin dysfunction is a frequent finding. It must be recognized that impulsivity and aggression are interlinked, and the d-fenfluramine challenge test may not be the optimal measure of serotonin function.
The evidence, therefore, suggests the possibility of some pathophysiology underpinning psychopathy. The aim of this paper is to examine, in greater detail, the contribution of neuroimaging to our understanding of this disorder.
Method
Literature searches were performed using PubMed and the search words psychopathy, antisocial personality, dissocial personality, image and imaging. Reference lists in recent papers and textbooks were examined for additional studies.
Results
Structural
Five studies of brain structure in psychopathy and related conditions were located and have been summarized in Table 1.
Structural neuroimaging in psychopathy
Raine et al. [26] compared a group of community-based volunteers with APD with three control groups (healthy patients, substancedependent individuals and psychiatric patients without APD) usingMRI and focusing on the grey and white matter volume of the frontal lobes. In addition, they studied autonomic activity (skin conductance and heart rate) during a social stressor. Compared to control groups, those with APD had a significant reduction (11%) in prefrontal grey, but not white, matter of the frontal lobes. In addition, the antisocial group also showed reduced autonomic activity (both skin conductance and heart rate) in the social stressor condition. Analyses of the data indicated that the grey matter and autonomic deficits could not be attributed to psychosocial risk factors. The authors related these deficits to the inability to make good choices and learn from experience, which characterize APD.
Laakso et al. [27] studied a group of habitually violent male prisoners diagnosed with APD and type 2 alcoholism. They correlated MRI-derived regional hippocampus volumes with the degree of psychopathology determined using the PCL-R. They found a strong negative correlation between the psychopathy scores and the volume of the posterior half of the hippocampus on both sides of the brain. The authors pointed to the studies that have established that the hippocampus is important in classical conditioning and social learning, and opined their findings may be a first step in illuminating part of a neural circuit predisposing to psychopathic behaviour. They cautioned that their observations may not be generalizable to non-alcoholic, non-violent psychopaths.
Laakso et al. [28] compared the group of violent male prisoners with APD and type 2 alcoholism who were mentioned above [27], and a group of hospital staff, students and their relatives. The degree of psychopathy was determined using the PCL-R. Magnetic resonance imaging was used to quantify the total prefrontal, and prefrontal white and cortical volumes. When differences in education and duration of alcoholism were controlled, there were no differences between the groups in any region. Nor were there significant correlations between any of the measured volumes and degree of psychopathy.
Dolan et al. [21] studied two groups drawn from maximum security psychiatric hospitals: personality disordered offenders (18 with psychopathy and six without) and controls, who were staff members. Patients underwent assessment of 5-HT function, and MRI quantitative measurement of frontal and temporal lobe volumes. The Special Hospital Assessment of Personality and Socialization (SHAPS) inventory was used to determine a belligerence score (impulsivity/hostility). The study examined the separate and potentially interacting roles of 5-HT, neuropsychological deficits and brain morphology in the pathogenesis of impulsive aggression in personality disordered patients. For the purposes of this review, the important finding was that no significant differences were found between the groups in frontal or temporal lobe volumes, after adjusting for head size. However, one-quarter of the patient population did not have psychopathic personality, and the measurements were of whole lobes and did not address the grey or white matter separately.
Raine et al. [29] studied groups of psychopathic and nonpsychopathic individuals from the community measuring the corpus callosum (volume of white matter, thickness and length, using MRI), the function of the corpus callosum (using divided field tasks), autonomic activity during social stress, personality measures of affective and interpersonal deficits and verbal and spatial ability. They found psychopathic individuals, compared to controls, had statistically significant (23%) larger callosal white matter volume. There was a reduction (15%) in thickness, but an increase in length (7%). There was a statistically significant increase in functional interhemispheric functioning. Larger callosal volumes were associated with antisocial scores (affect and impulsive/irresponsible, but not arrogant/deceptive factors), low autonomic response to stress and low spatial ability. The main group effect remained significant when alcohol and cannabis use, schizophrenia-spectrum disorders, head circumference and intelligence quotient (IQ) were entered as covariates into their calculations. The authors argued that these results were consistent with a neurodevelopmental perspective of adult psychopathic antisocial personality, but not with a discrete trauma or a degenerative disease process. They cautioned that an association had been shown between callosal deficits and psychopathic antisocial personality, but that causality could not be assumed.
Functional
Fifteen functional imaging studies in psychopathy and related conditions have been summarized in Table 2.
Functional imaging in psychopathy
Goyer et al. [30] used positron emission tomography (PET) to study regional cerebral metabolic rate of glucose (rCMRG) in a group of individuals with personality disorder. They found an inverse correlation between life history of aggression and rCMRG in the frontal cortex during an auditory activating procedure. They also compared subgroups (6=borderline and 6=antisocial personality) with a control group, finding a significant decrease in frontal metabolism in the borderline, but not the antisocial subgroup. The majority of the total patient group (13/17) also had an axis I diagnosis, and the majority of the antisocial group (4/6) had a history of alcohol abuse. Thus, these were of heterogeneous groups, and the results could at most suggest a relationship between impulsive aggression and frontal cortex function.
Raine et al. [31] compared a group of people accused of murder, who were claiming to be not guilty by reason of insanity, or who were considered incompetent to stand trial, with a group of age- and gendermatched controls. Three of 22 accused people had schizophrenia, and were matched with individuals suffering this condition. However, many others had a history of head injury, substance abuse, mood disorder, epilepsy, learning difficulties or personality disorder, and these could not be accurately matched. Patients performed a continuous performance task before PET evaluation of rCMRG. The accused individuals had significantly reduced rCMRG in the lateral and medical prefrontal cortex. The groups performed similarly on the continuous performance task. The authors conceded that the groups could not be perfectly matched. However, differences were not found to be a function of raised levels of left-handedness, schizophrenia, ethnic minority status, head injury or motivation deficits in the murder group. The authors cautioned that their results related to this select group and the findings could not be generalized to violent offenders in general.
Volkow et al. [32] used PET to study rCMRG and compared a group of psychiatric patients with a history of repetitive violence and a group of normal control patients. The patients were incarcerated in a state psychiatric hospital and were diagnosed with intermittent explosive behaviour or APD. Their violent behaviour had been characterized a purposeless meaning ‘without a desire to achieve a specific goal’, and had led to arrest. There was variation between the scans of individual patients, but as a group they showed significantly lower rCMRG in the medial temporal and prefrontal cortices, compared to the controls. Most patients had abused drugs in the past and 3/8 suffered schizophrenia or schizoaffective disorder andwere taking antipsychotic medication at the time of the study. The authors considered that the results implicated medial temporal and prefrontal cortex dysfunction in the violent behaviour of this patient group, but bias cannot be excluded.
Kuruoglu et al. [33] used single photo emission computerized tomography (SPECT) to study regional cerebral blood flow (rCBF) in 40 individuals with alcohol dependency, of whom 15 could be diagnosed with APD. They concluded that alcohol consumption was associated with significant reduction in rCBF in the frontal regions. A secondary finding was that the patients with APD exhibited a greater degree of hypofrontality than the remaining patients. The authors speculated that those with APD may be more vulnerable to the toxic effects of alcohol. However, the study design meant it was impossible to distinguish between primary and secondary APD.
Intrator et al. [34] examined substance abuse patients who were psychopathic (who scored more than 25 on the revised PCL-R), nonpsychopathic substance abuse patients and healthy volunteers, using SPECT and a lexical decision task (LDT). The LDT involved neutral words interspersed with pronounceable non-words (meder, tisan), and emotional words interspersed with non-words; the patients were asked to indicate whenever the stimulus was a real word. The focus of statistical analysiswas amidventricular slice, and the regions of interest (ROIs) were the left and right frontotemporal regions and the left and right contiguous deeper regions, including the basal ganglia. The psychopathic group showed greater relative activation in the emotional condition in all four ROIs, whereas the other groups showed greater activation in the neutral condition. The psychopathic group was significantly different to the other two groups in the left and right frontotemporal regions, and was significantly different to the normal controls in the two contiguous subcortical regions. The increase in activity in the psychopathic group in the emotional condition was described by the authors as ‘counterintuitive’. They speculated that as psychopathic individuals have a primary deficit in emotional processing, they may require additional resources to perform emotional tasks. The study showed that psychopathy is associated with functional abnormalities in the brain mechanism involved in linguistic and affective processing.
Raine et al. [35] extended their earlier study [31] of murderers who were claiming to be not guilty by virtue of insanity, by almost doubling the number of patients and addressing activity in subcortical structures. Murderers were again matched with healthy controls, except in the case of the 6/41 who suffered schizophrenia, who were matched with nonmurderer controls who suffered this condition. Murderers with other disorders could not be matched. Groups were matched on age, gender and ethnicity. The murderers showed reduced glucosemetabolism in the prefrontal cortex, superior parietal gyrus, left angular gyrus and corpus callosum, and they showed abnormal asymmetries (left hemisphere lower than right) in the amygdala, thalamus and medial temporal lobe. The authors sound many cautions, pointing to the importance of social, psychological, cultural and situational factors underpinning violence. However, they conclude their results suggesting that the reduced activity in the prefrontal, parietal and callosal regions along with the observed asymmetries of activity may predispose this specific group to violence.
Raine et al. [36] re-examined earlier data [35], dividing murderers into two groups: those with and without early psychosocial deprivation. They found murderers without early psychosocial deprivation showed significantly lower prefrontal glucose metabolism. They argued that for these offenders, the ‘social push’ was less, and their brain abnormalities may have provided a greater predisposition to violence.
Schneider et al. [37] compared psychopathic individuals and healthy volunteers using classical conditioning and an aversive stimulus in functional (f)MRI. The aversive stimulus was the odour of rotting yeast as the unconditioned stimulus and the conditioned stimulus was a picture of a face with a neutral expression. FunctionalMRI was performed during habituation, acquisition and extinction. With the pairing of stimuli, subjective ratings of the face changed in the negative direction for both groups, indicating that behavioural conditioning had occurred in both. During the acquisition phase, the bilateral activity in the amygdala and dorsolateral prefrontal cortex (DLPFC) differed between the groups. The controls showed decreases in these structures while the psychopathic group showed increases. Thus, the psychopathic group showed 862 NEUROIMAGING IN PSYCHOPATHY different brain activity (amygdala and DLPFC) to the control group during the processing of negative affect. The authors speculated that the increased activity in the psychopathic group may reflect a need for an additional effort during this process.
Smith [38] compared psychopathic inmates, non-psychopathic inmates and a healthy control group using fMRI during a response inhibition (Go/NoGo) task; PCL-R was scored. There was significantly less activity in the DLPFC of the psychopathic inmate group compared to the normal controls. Also, in the psychopathic inmate group, there was a negative relationship between activity in the DLPFC and the PCL-R score. The author suggested that the disinhibition observed in psychopathy may be related to a deficit in the processing of information in the DLPFC.
Kiehl et al. [39] studied psychopathic criminals, criminal nonpsychopaths and control non-psychopaths (all male) using whole brain fMRI while patients were responding to affectively ladenwords. In three phases, participants were first asked to memorize a list of 12 words; during the second phase, they were asked to rehearse the learned list; and during the third, to identify learned words from a list that included new words. This processwas repeated with affectively neutral and affectively negative words. There were no group differences during processing of neutral stimuli. However, for the criminal psychopaths, compared to the other two groups, the processing of affective stimuli was associated with less limbic (amygdala/hippocampal formation, parahippocampal gyrus, ventral striatum and anterior and posterior cingulate) activation. Criminal psychopaths also showed overactivity in the bilateral frontotemporal cortex. This was considered to be a visualization of the neural processes that support the frequently described abnormal affect of psychopathy.
Soderstrom et al. [40] studied a group of incarcerated violent offenders (subsequently found guilty of homicide, attempted homicide, aggravated assault, arson, rape or sexual abuse of minors) who had a range of psychiatric disorders. They compared scores on the PCL-R with rCBF determined using SPECT. The PCL-R is composed of Factor 1: disturbed interpersonal attitudes, and Factor 2: impulsive antisocial behaviour. The authors used a newapproach in which Factor 1 is divided into deceitful interpersonal style (New Factor 1) and affective unresponsiveness (New Factor 2). They found a significant negative correlation between Factor 1 (and particularly New Factor 1) results with frontal and temporal perfusion. This relationship was most clearly observed in the head of the caudate nuclei and the hippocampi. The authors defended the study design stating that it was ‘next to impossible’ to recruit a control group that wasmatched for all confounders and concluded that their results added evidence indicating aberrant frontotemporal activity may be a factor in violent behaviour.
Muller et al. [41] studied criminal psychopaths and healthy volunteers using whole brain fMRI to illuminate the response to affective stimuli. Patients viewed images with neutral, positive and negative load, from the affect picture system. The authors observed that earlier studies involved higher cortical functions (attention, language and memory) that might obscure atypical activation patterns. When viewing positive pictures, psychopaths compared to controls showed increased activation of the left orbitofrontal region, and reduced activity in the right medial frontal and medial temporal regions. When viewing negative pictures, psychopaths compared to controls showed increased activation on the right in prefrontal regions and the amygdala, and reduced activation of the right subgenual cingulate and temporal regions, and the left dorsal cingulate and parahippocampal gyrus. Thus, different activation patterns were shown in criminal psychopaths, with parts of the emotionrelated circuit being significantly overactive while other parts were underactive. Abnormalities occurred in both cortical and subcortical regions. The findings support the notion that psychopathy is associated with dysregulation of frontal and related regions of the brain during the processing of affective material.
Vollum et al. [42] compared a group of people with cluster B personality disorder (APD and borderline) and healthy controls using fMRI during the performance of a Go/NoGo task. In the control group, during response inhibition, the main sites of activation were in the prefrontal cortex, the right dorsolateral and the left orbitofrontal cortex. In the patient group, the activation was bilateral and more extensive, including the medial, superior and inferior frontal gyri and the anterior cingulate. Thus, during response inhibition, the cluster B and healthy patients activated different neural pathways.
Kiehl et al. [43] studied groups of incarcerated criminal psychopathic individuals and healthy volunteers using fMRI during the processing of abstract and concrete words. Two sets of words were presented, concrete words interspersed with pseudowords, and abstract words interspersed with pseudowords, and patients were to indicate as soon as they identified a real English language word. Psychopathic individuals were slower to respond to both abstract and concrete words, but there was no difference in accuracy. For both groups, processing ofword stimuli was associated with activation in bilateral fusiform gyrus, anterior cingulate, left middle temporal gyrus, right posterior superior temporal gyrus and left and right inferior frontal gyrus. However, consistent with the study hypothesis, psychopathic individuals failed to show a difference in activation of the right anterior temporal gyrus when processing abstract and concrete words. These results supported the researchers proposition that psychopathy is associated with dysfunction of the right hemisphere during the processing of abstract material. They speculated that complex social emotions such as love, empathy and guilt may call for abstract functioning, and that abstract processing deficits, based in the right temporal lobe, may be a fundamental abnormality in psychopathy.
Discussion
The comparison of neuroimaging studies in psychopathy, as with neuroimaging studies of other metal conditions, is made difficult by many differences between the studies. There are frequently differences in technical equipment and methodology. Although the structural studies of the last decade have not featured the older coaxial tomography and have all used MRI technology, differences still exist in the power of the machines, the thickness of slices and a host of other variables. The recent functional studies have used SPECT, PET and fMRI, which have basic differences in the physiological events they examine and the physics they use. Functional studies explore metabolism [30–32],[35] and blood flow [33], [34], [40], and may used conditioning theory [37], response inhibition [38], [42] affective processing [39], [41] and semantic processing [43]. Thus, the comparison of studies must be approached with great caution.
A difficulty in all psychiatric research is to identify a homogeneous population. Optimally, all research groups should use the same concepts, terminology and diagnostic criteria. On these points, the current review is severely hampered. The authors have chosen the term psychopathy, as this is widely used in clinical psychiatry. However, the major diagnostic systems use the terms dissocial (ICD-10) and antisocial (DSM-IV) personality disorder, and these carry slight differences in the emphasis of certain symptoms. Many authors, additionally or alternatively, use the PCL in the diagnostic or quantification process, most usually the revised version [34], [38], [40] but occasionally the screening version [22]. Whereas most conceptualized the PCL-R as being composed of two core components [27], [28], [37], others discerned three [22]. In one study, the combination of diagnostic procedures led to the designation, ‘psychopathic antisocial individuals’ [29]. Other authors have studied psychiatric patients with a history of violence [32], and accused murderers [35], [36], psychopathic inmates [38], violent offenders [40], criminal psychopaths [41] and psychopathic criminals [43]. Some have studied incarcerated individuals [27], [28], [32], [37], [38] and others, individuals living in the community [22], [26].
The choice of control groups had been a problematic issue. Incarcerated or hospitalized individuals have been compared with healthy community volunteers 30– [32], [35], [42], [43], or institution staff members [21], [28]. Incarcerated criminal psychopaths have been compared with incarcerated criminal non-psychopaths [39]. Community volunteer psychopathic individuals have been compared with community non-psychopathic individuals [29], APD individuals have been compared to borderline personality individuals [30], and ADP individuals with alcohol problems have been compared to non-APD individuals with alcohol problems [33]. Another approach compared substance abuse patients who were psychopathic, substance abuse patients who were non-psychopathic and healthy controls [34]. A similar comparison was between community living psychopathic individuals, psychiatric patients and healthy individuals [26]. Soderstrom et al.[40] found that generating control groups to match all confounders ‘next to impossible’, and comparisons have been made within groups of violent offenders on the basis of PCL-R results [27], [40].
The focus of this review is psychopathy. Some symptoms of psychopathy, such as irritability and impulsivity, are shared by peoplewith no psychiatric diagnosis, and by people with other psychiatric diagnoses. Decisions had to be made regarding the inclusion/exclusion of imaging studies that focused on such shared symptoms. Studies were included where the majority of those who were violent suffered psychopathy, other personality or had no psychiatric diagnosis [31], [35]. Studies were excluded in which the majority of patients suffered axis I diagnosis [44–47]. Criticism of this review can be made regarding the wisdom of the inclusion/exclusion of particular studies, and we make no claim to infallible wisdom. The study by Goyer et al. [30] was included although the majority of the patients were comorbid for and axis I diagnosis, because itwas one of the very earliest neuroimaging studies of personality disorder, and is widely quoted in the neuroimaging of psychopathy literature.
In this review, five structural imaging studies from the last decade have been considered. In view of the difficulties in selecting homogeneous index cases and control groups, conclusions cannot be made. Decreased prefrontal grey matter [26], decreased posterior hippocampal volume [27] and increased callosal white matter [29] have been reported, but these have not been substantiated [21], [27] and further studies are needed.
Fifteen functional imaging studies of the last decade have been considered. There is some evidence of reduced metabolism in the frontal cortex associated with a life history of aggression [30], having murdered [31], [35], and a history of repetitive violent behaviour [32]. Murderers may also show reduced metabolism in the angular gyrus and the corpus callosum [35].
There is also evidence of reduced frontal perfusion with antisocial behaviour [32], [33], [40]. Abnormalities have been shown in the frontal lobes when index cases have performed tasks such as response inhibition [38], [42]. Increases in activity have been shown in the frontal and temporal lobes of index cases associated with classical conditioning [37] and the processing and remembering of emotional words [34], [39]. This increased activity with the processing of emotional material may reflect that in psychopathy, additional effort and resources are required to process this type of information.
In a recent study of the processing of emotional words, dysfunction was shown in the right anterior temporal gyrus [43]. The use of positive and negative emotionally laden pictures has shown wide spread differences in activation patterns in the frontal and temporal lobes when criminal psychopaths were compared to normal volunteers [41]. These results are consistent with the clinical picture of psychopathy in which the patient is lacking normal emotional responses.
The functional but not the structural neuroimaging strongly suggests dysfunction of the frontal and temporal lobes, and possibly other structures including the angular gyrus and corpus callosum, in psychopathy. However, replication studies are required before conclusions can be drawn.
The findings reviewed can be placed in a broader context. The new and ever advancing technologies are 864 NEUROIMAGING IN PSYCHOPATHY progressively identifying non-specific findings in a range of psychiatric disorders [48]. It may be some time before diagnosis, specific findings provide information on the underlying pathology, which can be related to our clinical experience.
Should further studies substantiate that psychopathy is associated with abnormal brain activity, caution will be required, and many questions will remain. It is important to avoid an excessively reductionistic approach. Responses depend on genetic endowment, early life experience and the sociocultural context, in addition to the nature of the current stimulus and its significance to the individual.