- Email: [email protected]
Brain abnormalities in high-risk violent offenders and their association with psychopathic traits and criminal recidivism
Accepted Manuscript Brain abnormalities in high-risk violent offenders and their association with psychopathic traits and criminal recidivism Verena Leutgeb, Mario Leitner, Albert Wabnegger, Doris Klug, Wilfried Scharmüller, Thomas Zussner, Anne Schienle PII: DOI: Reference:
S0306-4522(15)00821-0 http://dx.doi.org/10.1016/j.neuroscience.2015.09.011 NSC 16567
To appear in:
Neuroscience
Accepted Date:
3 September 2015
Please cite this article as: V. Leutgeb, M. Leitner, A. Wabnegger, D. Klug, W. Scharmüller, T. Zussner, A. Schienle, Brain abnormalities in high-risk violent offenders and their association with psychopathic traits and criminal recidivism, Neuroscience (2015), doi: http://dx.doi.org/10.1016/j.neuroscience.2015.09.011
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
Brain abnormalities in high-risk violent offenders and their association with psychopathic traits and criminal recidivism
Verena Leutgeba*, Mario Leitnerb, Albert Wabneggera, Doris Kluga,b, Wilfried Scharmüllera, Thomas Zussnera & Anne Schienlea
a
Clinical Psychology, University of Graz, BioTechMedGraz Postal address: Universitätsplatz 2/DG, 8010, Graz, Austria b
Graz-Karlau State Correctional Facility, Graz, Austria Postal address: Herrgottwiesgasse 50, 8200, Graz, Austria
* Corresponding author: Verena Leutgeb University of Graz Universitätsplatz 2/DG A - 8010 Graz, Austria [email protected]
2
ABSTRACT Measures of psychopathy have been proved to be valuable for risk assessment in violent criminals. However, the neuronal basis of psychopathy and its contribution to the prediction of criminal recidivism is still poorly understood. We compared structural imaging data from 40 male high-risk violent offenders and 37 non-delinquent healthy controls via voxel-based morphometry. Psychopathic traits and risk for violence recidivism were correlated with grey matter volume (GMV) of regions of interest previously shown relevant for criminal behavior. Relative to controls, criminals showed less GMV in the prefrontal cortex and more GMV in cerebellar regions and basal ganglia structures. Within criminals, we found a negative correlation between prefrontal GMV and psychopathy. Additionally, there was a positive correlation between cerebellar GMV and psychopathy as well as risk of recidivism for violence. Moreover, grey matter volumes of the basal ganglia and supplementary motor area (SMA) were positively correlated with anti-sociality. GMV of the amygdala was negatively correlated with dynamic risk for violence recidivism. In contrast, GMV of (para)limbic areas (orbitofrontal cortex, insula) was positively correlated with antisociality and risk for violence recidivism. The current investigation revealed that in violent offenders deviations in GMV of the prefrontal cortex as well as areas involved in the motor component of impulse control (cerebellum, basal ganglia, SMA) are differentially related to psychopathic traits and the risk for violence recidivism. The results might be valuable for improving existing risk assessment tools.
Key words: Grey matter volume, risk assessment, cerebellum, basal ganglia, violence
3
1. INTRODUCTION There is emerging interest in developing tools that provide information about recidivism of violent criminal behavior in addition to behavioral measures. Empirically developed risk assessment instruments provide a probabilistic estimate of the violence risk for a specified time period (Hanson and Howard, 2010; Swets et al., 2000). Risk factors that are unchangeable (e.g., school maladjustment, history of alcohol problems, age at index offense) are commonly referred to as ‘static’. In contrast, ‘dynamic’ risk factors (e.g., impulsivity, unstable interpersonal relationships, lack of cooperation) assess the client’s treatment readiness and possibility of change. The Violence Risk Appraisal Guide (VRAG) for example is most commonly used by clinicians for assessing the static risk of violent recidivism of violent offenders (Quinsey et al., 1998). In addition, the Violence Risk Scale (VRS) allows a prospective rating of static and dynamic risk of violence recidivism (Wong and Gordon, 2006). A recent meta-analysis found little variation amongst the mean effect sizes of common actuarial or structured risk instruments (2009; Campbell et al., 2009). Therefore, neuroscientific methods may have the potential to improve existing tools for prediction of violence recidivism (for a recent review, see Meixner, 2014). However, despite the recent growth of knowledge on structural and functional alterations in the brain of individuals, who commit violent crimes, the contribution to criminal law (e.g., questions of credibility, culpability, or prediction of future recidivism) is still low. Psychopathy has been reported to be strongly associated with violence and criminal recidivism (Hare, 1991, 2003). It is a personality construct characterized by deficits in interpersonal relations and affective processes (e.g., fearlessness, callousness, failure to form close emotional bonds, dishonesty, deficits in passive avoidance learning, and deficient empathic responses) as well as antisocial and
4
impulsive behavior (2008; Hare and Neumann, 2008). The Psychopathy ChecklistRevised (PCL-R) is a valid measure of psychopathy (Hare, 2003) and has been successfully introduced as an instrument for risk assessment (Hare et al., 2000). It comprises two factors reflecting emotional and interpersonal detachment (Factor 1) as well as antisocial behavior and parasitic lifestyle (Factor 2). Regarding the twofactor model emotional detachment is the key feature of psychopathy (Hare, 2003). There is an ongoing debate about the question, whether the predictive efficacy of the PCL-R should be attributed more to Factor 1 or to Factor 2. In a recent meta-analytic study (Yang et al., 2010) the authors questioned the predictive quality of Factor 1 for violence in men, as the subscale predicted violence only at chance level. In contrast, Factor 2 showed high predictive quality. A disruption of brain regions underlying moral thinking and feeling might be central for antisocial behavior (for a thorough discussion, see Glenn et al., 2012). In line with this notion, magnetic resonance imaging studies have found significantly reduced grey matter volume (GMV) of the prefrontal cortex in patients with antisocial personality disorder (APD) and/ or high levels of psychopathy which has been interpreted to lead to poor behavioral control (Müller et al., 2008; Raine et al., 2000; Yang et al., 2005). Gregory et al. (2012) found reduced prefrontal GMV in offenders with APD, who were classified as psychopaths, relative to offenders with APD ‘alone’, and suggested that psychopathy represents a distinct phenotype. Moreover, psychopathy and persistent violence have also been associated with decreased grey matter in limbic and paralimbic areas such as the amygdala, the insula, the hippocampus, or the orbitofrontal cortex (e.g., Boccardi et al., 2011; ContrerasRodríguez et al., 2014; Ermer et al., 2012; Yang and Raine, 2009; Oliveira-Souza et al., 2008). Existing results are somehow inconsistent, as Boccardi et al. (2011) reported enlargement of the lateral and central nucleus of the amygdala in offenders
5
relative to controls. Anyway, the notion that psychopathy is associated with frontolimbic grey matter reductions and dysfunctions concerning moral sensibility and behavior seems to be widely accepted (e.g., Oliveira-Souza et al., 2008; for a review, see Simpson, 2012). Despite the fact that the majority of studies found reduced structures in antisocial or psychopathic groups, there is evidence that some brain regions, like basal ganglia and cerebellar cortex, may actually be larger (Glenn et al., 2012; Barkataki et al., 2006; Tiihonen et al., 2008). In addition, there are reports of abnormalities in the supplementary motor area (SMA) in criminals with high PCL-R scores (Müller et al., 2008). As those brain regions are associated with motor control and higher-order functions of motor cognition, structural deviations in samples afflicted with APD might explain impulsivity and poor behavioral control as well as violence recidivism (Bari and Robbins, 2013; Picazio and Koch, 2015). The current investigation was designed to compare structural imaging data of high-risk violent offenders with non-delinquent controls. We expected enhanced GMV in criminals relative to controls in basal ganglia and cerebellar regions, as well as reduced GMV in prefrontal structures and (para)limbic areas (e.g., the amygdala). For psychopathic traits and risk of criminal recidivism we hypothesized positive correlations with GMV of basal ganglia and cerebellar regions as well as the supplementary motor area (SMA), and negative correlations with prefrontal and limbic areas (e.g., amygdala).
2. EXPERIMENTAL PROCEDURES 2.1 Subjects We investigated 40 male inmates from a maximum-security prison located in Graz (Austria) and 37 male, non-delinquent controls with comparable age (prisoners: M = 38.1 years, SD = 12.0; controls: M = 36.7 years, SD = 9.6; t(75) = 0.5, p = .589). To
6
avoid group differences in overall intelligence, participants of the two groups were matched according to their educational level (years of education; prisoners: M = 11.4 years, SD = 1.9; controls: M = 11.6 years, SD = 1.4; t(75) = -0,6, p = .520). Moreover, the groups were matched according to their handedness (73 right-handed; 4 lefthanded). All participating prisoners were high-risk violent offenders (with one or more index convictions for violence). Sexual offenders were not included in the sample. All participants were to be released to the community in the near term (which meant early release/ conditional release or completion of their sentence). A comprehensive clinical psychiatric assessment and structured risk assessment was performed by experienced forensic psychologists or psychiatrists and actuarial instruments (see below). Clinically relevant symptoms of depression, bipolar disorder, psychosis, attention-deficit hyperactivity disorder, as well as organic (e.g., hyper- or hypothyroidism) or neurological conditions (e.g., history of severe head injury) led to exclusion from the sample. Moreover, prisoners with a history of substance and/ or alcohol abuse during imprisonment were excluded. Community healthy control participants were recruited via announcements in a local newspaper. Control group participants who had been convicted for any crime, or had a history of a mental disorder or substance/ alcohol abuse, were excluded. The prisoners were either given permission to leave prison on a day-release or to go out for a few hours a day or were escorted to the MRI-Lab of the University of Graz by officials of the Graz-Karlau State Correctional Facility, Graz, Austria. All participants provided written informed consent after receiving a full explanation of the test procedure. The study was in accordance with the Operational Guidelines for Ethics Committees That Review Biomedical Research (WHO, 2000) and had been approved by the ethics committee of the University of Graz.
7
2.2 Clinical Assessments Clinical rating-data and questionnaire scores for prisoners and controls are summarized in Table 1. For both groups, two independent raters filled out the Hare Psychopathy Checklist-Revised (PCL-R), which is a reliable and valid measure of psychopathic traits in the prison system (Hare, 1991, 2003). The PCL-R consists of 20 items, which have to be rated on a 3-point-scale. It uses a semi-structured interview, case history information, and specific scoring criteria and consists of two factors (2003; Hare, 2003). Factor 1 is comprised by the following two aspects: interpersonal (glibness/ superficial charm, grandiose sense of self-worth, pathological deception, conning/ manipulative) and affective (lack of remorse or guilt, shallow affect, callous/ lack of empathy, failure to accept responsibility for actions). Factor 2 is comprised by the following two aspects: lifestyle (need for stimulation/ proneness to boredom,
parasitic
lifestyle,
lack
of
realistic
long-term
goals,
impulsivity,
irresponsibility) and antisocial (poor behavioral control, early behavioral problems, juvenile delinquency, revocation of conditional release, criminal versatility). Inter-rater reliabilities (Pearson’s correlations) for the sum score and subscales (Factor 1 and 2) of the PCL-R were sufficiently high and ranged for prisoners from r = .80 to .88, and for controls from .81 to .96. Two scales, focusing on risk of violence recidivism, and thus on offense history, were additionally filled out for the prisoners by two independent raters, but not for controls: we obtained the Violence Risk Appraisal Guide (VRAG), which was developed to assess risk for violent recidivism (Rossegger et al., 2009). The scale contains 12 items reflecting static risk factors (e.g., age at index offense, index victim injury, sex of index victim). VRAG scores allow assignment of an offender to one of nine risk categories, ranging from 1 (lowest risk) to 9 (highest risk). The inter-rater reliability (Pearson’s correlation) for the sum score of the VRAG was r = .99. The
8
second scale was the Violence Risk Scale (VRS; Wong and Gordon, 2006). This scale was constructed for purposes of risk assessment and prediction. It integrates violence assessment, prediction, and treatment and consists of 6 items forming a Static Factor and 20 items forming a Dynamic Factor. Each item is rated on a 4-point scale. Static variables, such as offense history, remain unchanged regardless of treatment interventions. In contrast, dynamic variables such as interpersonal aggression or emotional control identify treatment targets linked to violence, and assess the client's treatment readiness and possibility of change. The inter-rater reliabilities (Pearson’s correlations) for the subscales of the VRS were r = .99 for static variables, and r = .98 for dynamic variables. All participants filled out the State-Trait Anger Expression Inventory (STAXI; Schwenkenmezger et al., 1992). The inventory consists of six scales: Trait Anger (frequency of anger experience), Temperament (intense anger in response to mild provocation), Rejection (sensitivity to criticism), Anger-Out (disposition to express anger aggressively), Anger-In (disposition not to act out feelings of anger), AngerControl (disposition to suppress feelings of anger). For this questionnaire, participants are asked to rate 44 items on 4-point scales that assess both the intensity of their anger at a particular time and the frequency that anger is experienced, expressed, and controlled.
2.3 MRI Acquisition and VBM Analyses T1-weighted scans were acquired using a 3T Siemens Skyra with a 32-channel head coil (Siemens, Erlangen, Germany). The scanning parameters were as follows: voxel size: = 0.88 x 0.88 x 0.88 mm; 192 transverse slices, slice thickness: 0.88mm, TE = 1.89 ms, TR = 1680 ms; TI = 1000 ms, flip-angle = 8o.
9
Brain imaging data were analyzed using SPM8 (Wellcome Trust Centre for Neuroimaging; http://www.fil.ion.ucl.ac.uk/spm/software/spm8/) including the VBM8 toolbox
(revision
435;
http://dbm.neuro.uni-jena.de/vbm)
for
voxel-based
morphometry in order to gain voxel-wise comparisons of grey matter volume (GMV). Anatomical scans were segmented into grey matter, white matter, and cerebrospinal fluid partitions. An optimized blockwise non-local means de-noising filter, a Hidden Markov Random Field approach, partial volume estimates, and normalization to MNI space by high-dimensional warping (DARTEL) with a standard template included in the VBM8-toolbox were used for pre-processing (final resolution: 1.5 mm × 1.5 mm × 1.5 mm). A Jacobian modulation for non-linear normalization was applied to correct for differences in head sizes and to obtain brain volume. Smoothing was executed with a Gaussian kernel with a full width at half maximum (FWHM) of 8 mm. Based on previous VBM studies with criminal, antisocial or psychopathic populations (Müller et al., 2008; Raine et al., 2000; Yang et al., 2005; ContrerasRodríguez et al., 2014; Ermer et al., 2012; Barkataki et al., 2006; Tiihonen et al., 2008) the following regions of interest (ROIs) were selected: dorsolateral prefrontal cortex (DLPFC), dorsomedial prefrontal cortex (DMPFC), orbitofrontal cortex (OFC) amygdala, insula, caudate nucleus, pallidum, putamen, cerebellar hemispheres, vermis, and supplementary motor area (SMA). For the present study we used ROIs from the automated anatomical labeling (AAL) template. The ROIs were constructed with the WFU PickAtlas (version 2.4; Wake Forest University School of Medicine). The DLPFC was built from masks of the middle frontal and the superior frontal gyrus, whereas the DMPFC was built from the frontal superior medial mask. Following masks were used (mask sizes in voxel (v) are provided in brackets): left DLPFC (21448 v), right DLPFC (23502 v), left DMPFC (8095 v), right DMPFC (6093 v), left OFC (13327 v), right OFC (13895 v), left amygdala (576 v), right amygdala (673 v),
10
left insula (4941 v), right insula (4842 v), left caudate nucleus (2649 v), right caudate nucleus (2737 v), left pallidum (1153 v), right pallidum (1060 v), left putamen (2688 v), right putamen (3776 v), left cerebellar hemisphere (25825 v), right cerebellar hemisphere (27042 v), vermis (4967 v), left SMA (5781 v), and right SMA (6291 v). We report analyses with initial thresholds .05 uncorrected. For purposes of comparability, we additionally provide analyses with initial thresholds .01 uncorrected. Peaks for voxel intensity tests are reported when significant (p < .05 corrected for family-wise error, small volume correction).
2.4 Statistical Analyses Statistical analyses were carried out using random effects models. To test for differences in GMV between patients and controls a two samples t-test (voxel intensity test) was conducted (comparisons: Criminals <> Controls). Scores of the PCL-R, the VRAG risk category, the VRS, and the STAXI were correlated via a simple regression with GMV within criminals and controls, respectively. Due to insufficient variability, we conducted no regression analyses for the PCL-R within controls. To restrict analysis to grey matter, images were thresholded for all analyses with an absolute threshold of 0.2.
3. RESULTS 3.1 Clinical Assessments The prisoners obtained higher scores than controls on both factors of the PCL-R and on the subscale Temperament of the STAXI (see Table 1).
Insert Table 1 here
11
3.2 VBM Analyses 3.2.1 Group differences In the ROI analysis, the prisoners had more GMV than controls in cerebellar regions, the caudate nucleus, and the pallidum (see Table 2). There were no group differences in other ROIs. Controls had more GMV in the DMPFC than prisoners (see Figure 1). A whole-brain analysis did not reveal any significant differences in GMV between prisoners and controls.
Insert Table 2 here Insert Figure 1 here
3.2.2 Regression analyses within prisoners Psychopathic traits (PCL-R, Factor 1) were positively correlated with cerebellar volume (see Figure 2) and negatively correlated with GMV of the DLPFC (see Table 3). Moreover, antisocial behavior (PCL-R, Factor 2) positively correlated with the volume of the SMA, basal ganglia regions (putamen and pallidum), as well as (para)limbic regions (OFC, insula). Scores for static risk prediction (VRAG Risk Category; VRS Static) positively correlated with GMV of cerebellar regions and (para)limbic regions (OFC, insula). VRS Dynamic negatively correlated with GMV of the
amygdala.
There
were
positive
correlations
between
the
subscales
‘Temperament’ and ‘Anger Out’ of the STAXI and GMV of cerebellar regions, as well as SMA.
3.2.3 Regression analyses within controls:
12
There was a negative correlation for the subscale ‘Anger Control’ of the STAXI and GMV of the OFC. Moreover, there was a positive correlation between the subscale ‘Temperament’ of the STAXI and GMV of the insula.
Insert Figure 2 here Insert Table 3 here
4. DISCUSSION This VBM study focused on differences in grey matter volume (GMV) between incarcerated high-risk violent offenders and non-delinquent healthy controls. Moreover, scores reflecting psychopathic traits (according to the Hare Psychopathy Checklist-Revised, PCL-R (Hare, 2003), risk for violent recidivism (Violence Risk Appraisal Guide, VRAG (Rossegger et al., 2009) and Violece Risk Scale, VRS (Wong and Gordon, 2006), as well as anger state-trait anger expression/ management, STAXI (Schwenkenmezger et al., 1992) were correlated with GMV. The main result of the current study revealed enhanced regional GMV in criminals relative to controls in areas associated with the motor component of impulse control (Picazio and Koch, 2015; Glenn et al., 2010), namely the cerebellum, and basal ganglia (caudate nucleus, pallidum). Moreover, within criminals there were consistent positive associations between GMV of the cerebellar cortex, (para)limbic structures (OFC, insula), basal ganglia, as well as the supplementary motor area (SMA) with psychopathic traits, anti-sociality, risk of criminal recidivism, and poor anger management. Similar results have been reported earlier by a research group, who found a 9.6% increase in striatal volume in psychopathic individuals relative to comparison subjects (2010; Glenn et al., 2010). Moreover, investigations showed increases in
13
putamen volume in patients afflicted with antisocial personality disorder (2006; Barkataki et al., 2006). As they also observed this increase in violent relative to nonviolent schizophrenics, the authors hypothesized a general involvement of this structure in violence. This seems plausible, as striatal structures have been found to be involved in impulsivity and poor behavioral control (Garavan, 2002). In line, our regression analysis within criminals revealed a positive relationship between antisocial behavior (Factor 2 of the PCL-R) and GMV of putamen and pallidum. Another brain region that might have a relevant contribution to psychopathic traits is the cerebellum, due to its involvement in motor control (for a discussion, see Picazio and Koch, 2015). To our knowledge, there is one investigation that found larger GMV of the cerebellum in persistently violent offenders with antisocial personality disorder relative to healthy controls (Tiihonen et al., 2008). Within the current study, there was a very consistent positive relationship between predictors of violent crimes and cerebellar GMV. This was true for interpersonal and affective problems associated with psychopathy (Factor 1 of the PCL-R), static risk factors and criminal relapse rates. Additionally, measures for temperament and expression of anger correlated positively with cerebellar GMV. Therefore, our data point to an involvement of the cerebellum in psychopathic traits and criminal recidivism. The enhanced cerebellar grey matter volume in criminals relative to controls might be interpreted to explain heightened impulsivity and poor behavioral control. In line, our regression analysis within criminals revealed a positive relationship between antisocial behavior (Factor 2 of the PCL-R) and GMV of the SMA, a brain region that is also known to be involved in motor control and impulsivity (for a recent review, see Bari and Robbins, 2013). In addition, expression of anger was positively correlated with the volume of the SMA.
14
The cerebellum has also been reported to be critically involved in moral behavior and aggression and possess various anatomical connections, also with the PFC (see Demirtas-Tatlidede and Schmahmann, 2013; Fumagalli and Priori, 2012; Turner et al., 2007). Thus, it seems interesting that in line with several structural studies with anti-social (e.g., Raine et al., 2000; Tiihonen et al., 2008) or psychopathic (e.g., Müller et al., 2008; Yang et al., 2005; Contreras-Rodríguez et al., 2014) individuals we found reduced GMV in the prefrontal cortex (PFC) in criminals relative to controls. Structural and functional changes of the PFC are the bestreplicated abnormalities across a wide range of antisocial groups and have also been obtained with different imaging methodologies (Glenn et al., 2012). The involvement of the PFC in higher-order executive functions can easily explain emotional deficits, poor decision-making, and the lack of behavioral control in violent offenders. In line, the regression analysis of our data within criminals revealed a negative correlation between interpersonal and affective problems associated with psychopathic traits (Factor 1 of the PCL-R; e.g., superficial charm, a manipulative nature, and a lack of empathy) and GMV of the dorsolateral PFC. Lastly, the current investigation failed to detect reductions in GMV of (para)limbic areas in offenders relative to controls. However, we found a negative relationship between dynamic risk factors for criminal relapse and grey matter of the amygdala. The amygdala is known to be crucial for emotional re-learning and conditioning (LeDoux, 2000). In fact, in an fMRI investigation psychopaths showed deficient fear conditioning and reduced amygdala activation relative to healthy controls (2005; Birbaumer et al., 2005). A deficiency to emotionally relate to neutral and biologically significant events might be central in psychopathy. Our study suggests that there is an association between reduced amygdala volume and dynamic risk factors for violence. In contrast, we found a positive relationship
15
between antisocial behavior (Factor 2 of the PCL-R) as well as criminal relapse rates and GMV of (para)limbic areas (OFC, insula). Summing up the correlations between GMV and risk assessment scales, both factors of the PCL-R revealed a relationship to GMV in violent offenders. Factor 2, reflecting antisocial behavior and parasitic lifestyle, was positively related to volumes of putamen, pallidum, and the SMA. In contrast, Factor 1, reflecting the core features of psychopathy, revealed a positive relationship with the cerebellum and a negative relationship with the dorsolateral PFC. Thus, although the predictive quality of Factor 1 for violence in men has been questioned in previous investigations (Yang et al., 2010), our data provide support for a differential relationship between the core features of psychopathic personality traits and GMV. Thus, in future investigations, the association between cerebellar volume and Factor 1 of the PCL-R should be further investigated. In addition, risk for violence recidivism (VRAG, VRS) was positively related to cerebellar GMV and negatively related to amygdala volume.
5. CONCLUSIONS Abnormalities in GMV of the prefrontal cortex as well as areas involved in the motor component of impulse control (cerebellum, basal ganglia, SMA) seem to be related to psychopathic traits and the risk of criminal recidivism in violent offenders. It remains unclear, if the abnormalities are causal for criminality, and moreover, if they are reversible. With respect to this and also as a limitation for the interpretation of group differences it has to be noted, that we did not control for childhood history (e.g., abuse or poverty) or history of alcohol and drug abuse prior to imprisonment. Science is still far from establishing neurobiological criteria as a contribution to final decisions in the court (for a discussion, see Witzel, 2012). However, the integration of objective neurobiological criteria, and an understanding of the interaction of biological and
16
psychological factors could provide a higher quality of risk assessment. In addition, understanding neurobiological correlates of mental disorders could be helpful for improving therapeutic interventions. Lastly, a neurobiological evaluation of treatment success in forensic psychiatry would be of great value for the decision if patients should be discharged, or not. Remedial efforts in the treatment of psychopathy might be most effective when implemented prior to the onset of severe antisocial behavior in the teenage years (2014; Anderson and Kiehl; Anderson and Kiehl, 2014). Future studies should therefore focus on changes in regional brain structures associated with therapeutic treatment. Moreover, future investigations should explore the neuronal basis of the relationship between self-report measures of psychopathic traits and criminal behavior.
ACKNOWLEDGEMENT We received no funding for the current study. There are no conflicts of interest or financial disclosures for any of the authors.
REFERENCES Anderson NE, Kiehl KA (2014) Psychopathy: Developmental perspectives and their implications for treatment. Restor Neurol Neuros 32:103–117. Bari A, Robbins TW (2013) Inhibition and impulsivity: Behavioral and neural basis of response control. Prog Neurobiol 108:44–79. Barkataki I, Kumari V, Das M, Taylor P, Sharma T (2006) Volumetric structural brain abnormalities in men with schizophrenia or antisocial personality disorder. Behav Brain Res 169:239–247.
17
Birbaumer N, Veit R, Lotze M, Erb M, Hermann C, Grodd W, Flor H (2005) Deficient fear conditioning in psychopathy: A functional magnetic resonance imaging study. Arch Gen Psychiat 62:799–805. Boccardi M, Frisoni GB, Hare RD, Cavedo E, Najt P, Pievani M, Rasser PE, Laakso MP, Aronen HJ, Repo-Tiihonen E, Vaurio O, Thompson PM, Tiihonen J (2011) Cortex and amygdala morphology in psychopathy. Psychiatry research 193:85– 92. Campbell MA, French S, Gendreau P (2009) The Prediction of Violence in Adult Offenders: A Meta-Analytic Comparison of Instruments and Methods of Assessment. Crim Justice Behav 36:567–590. Contreras-Rodríguez O, Pujol J, Batalla I, Harrison BJ, Soriano-Mas C, Deus J, López-Solà M, Macià D, Pera V, Hernández-Ribas R, Pifarré J, Menchón JM, Cardoner N (2014) Functional Connectivity Bias in the Prefrontal Cortex of Psychopaths. Biol Psychiat:in press. Demirtas-Tatlidede A, Schmahmann JD (2013) Morality: incomplete without the cerebellum? Brain a journal of neurology 136:e244. Ermer E, Cope LM, Nyalakanti PK, Calhoun VD, Kiehl KA (2012) Aberrant paralimbic gray matter in criminal psychopathy. J Abnormal Psych 121:649–658. Fumagalli M, Priori A (2012) Functional and clinical neuroanatomy of morality. Brain a journal of neurology 135:2006–2021. Garavan H (2002) Dissociable Executive Functions in the Dynamic Control of Behavior: Inhibition, Error Detection, and Correction. NeuroImage 17:1820–1829. Glenn AL, Raine A, Yaralian PS, Yang Y (2010) Increased Volume of the Striatum in Psychopathic Individuals. Biol Psychiat 67:52–58. Glenn AL, Yang Y, Raine A (2012) Neuroimaging in Psychopathy and Antisocial Personality Disorder: Functional significance and a neurodevelopmental
18
hypothesis. In: Neuroimaging in Forensic Psychiatry: From the clinic to the courtroom (Simpson JR, eds.), pp 81–98. Oxford, UK: Wiley-Blackwell. Gregory S, ffytche D, Simmons A, Kumari V, Howard M, Hodgins S, Blackwood N (2012) The antisocial brain: psychopathy matters. Archives of general psychiatry 69:962–972. Hanson RK, Howard PD (2010) Individual confidence intervals do not inform decision-makers about the accuracy of risk assessment evaluations. Law Hum Behav 34:275–281. Hare RD (1991) The Hare Psychopathy Checklist-Revised. Toronto, ON: Multi-Health Systems. Hare RD (2003) The Hare Pysychopathy Checklist-Revised. 2nd ed. Toronto, ON: Multi-Health Systems. Hare RD, Clark D, Grann M, Thornton D (2000) Psychopathy and the predictive validity of the PCL-R: an international perspective. Behav Sci Law 18:623–645. Hare RD, Neumann CS (2008) Psychopathy as a Clinical and Empirical Construct. Annu Rev Clin Psychol 4:217–246. LeDoux JE (2000) Emotion circuits in the brain. Annu Rev Neurosci 301:155–167. Meixner JB (2014) Applications of Neuroscience in Criminal Law: Legal and Methodological Issues. Curr Neurol Neurosci Rep 15:513. Müller JL, Gänßbauer S, Sommer M, Döhnel K, Weber T, Schmidt-Wilcke T, Hajak G (2008) Gray matter changes in right superior temporal gyrus in criminal psychopaths. Evidence from voxel-based morphometry. Psychiat Res-Neuroim 163:213–222. Oliveira-Souza R de, Hare RD, Bramati IE, Garrido GJ, Azevedo Ignácio F, TovarMoll F, Moll J (2008) Psychopathy as a disorder of the moral brain: Fronto-
19
temporo-limbic grey matter reductions demonstrated by voxel-based morphometry. NeuroImage 40:1202–1213. Picazio S, Koch G (2015) Is Motor Inhibition Mediated by Cerebello-cortical Interactions? The Cerebellum 14:47–49. Quinsey VL, Harris GT, Rice ME, Cormier CA (1998) Violent offenders: appraising and managing risk. 1st ed. Washington DC: American Psychological Association. Raine A, Lencz T, Bihrle S, LaCasse L, Colletti P (2000) Reduced prefrontal gray matter volume and reduced autonomic activity in antisocial personality disorder. Arch Gen Psychiat 57:119–127. Rossegger A, Urbaniok F, Danielsson C, Endrass J (2009) Der Violence Risk Appraisal Guide (VRAG): ein Instrument zur Kriminalprognose bei Gewaltstraftätern. Fortschr Neurol Psychiatr 77:577-584. Schwenkenmezger P, Hodapp V, Spielberger CD (1992) State-Trait-ÄrgerausdrucksInventar. Bern: Verlag Hans Huber. Simpson, J. R. ed. (2012) Neuroimaging in Forensic Psychiatry: From the clinic to the courtroom. Oxford, UK: Wiley-Blackwell. Swets JA, Dawes RM, Monahan J (2000) Better decisions through science. Sci Am 283:82–87. Tiihonen J, Rossi R, Laakso MP, Hodgins S, Testa C, Perez J, Repo-Tiihonen E, Vaurio O, Soininen H, Aronen HJ, Könönen M, Thompson PM, Frisoni GB (2008) Brain anatomy of persistent violent offenders: More rather than less. Psychiat Res-Neuroim 163:201–212. Turner BM, Paradiso S, Marvel CL, Pierson R, Boles Ponto, Laura L., Hichwa RD, Robinson RG (2007) The cerebellum and emotional experience. Neuropsychologia 45:1331–1341.
20
WHO (2000) Operational guidelines for ethics committees that review biomedical research. Geneva: World Health Organization. Witzel J (2012) Implications of neuroimaging for dangerousness assessment. In: Neuroimaging in Forensic Psychiatry: From the clinic to the courtroom (Simpson JR, eds.), pp 195–200. Oxford, UK: Wiley-Blackwell. Wong S, Gordon A (2006) The validity and reliability of the Violence Risk Scale: a treatment-friendly violence risk assessment tool. Psychol Public Pol L 12:279– 309. Yang M, Wong, Stephen C. P., Coid J (2010) The efficacy of violence prediction: a meta-analytic comparison of nine risk assessment tools. Psychol Bull 136:740– 767. Yang Y, Raine A (2009) Prefrontal structural and functional brain imaging findings in antisocial, violent, and psychopathic individuals: A meta-analysis. Psychiat ResNeuroim 174:81–88. Yang Y, Raine A, Lencz T, Bihrle S, LaCasse L, Colletti P (2005) Volume reduction in prefrontal gray matter in unsuccessful criminal psychopaths. Biol Psychiat 57:1103–1108.
FIGURE CAPTIONS
Figure 1: Group differences in grey matter volume.
Figure 2: Positive regression within prisoners between scores of Factor 1 (PCL-R) and grey matter volume of the right cerebellar hemisphere.
21
22
23
24
Table 1: Comparison of clinical rating-data and questionnaire scores between prisoners and controls.
Interview
PCL-R
VRAG Risk Category VRS
STAXI
Sum score Factor 1 Factor 2
Static Dynamic Trait Anger Temperament Rejection Anger In Anger Out Anger Control
Prisoners M (SD) 20.6 (7.7) 7.5 (3.9) 10.8 (5.3) 5.5 (1.7) 7.52 (4.6) 29.0 (11.9) 15.8 (3.4) 7.8 (2.3) 8.0 (1.8) 14.2 (5.1) 11.4 (3.5) 24.6 (4.7)
Controls M (SD) 1.6 (1.8) 0.5 (0.8) 1.1 (1.6) 15.1 (3.1) 6.7 (1.5) 8.4 (2.2) 14.7 (4.6) 10.4 (2.4) 25.2 (5.3)
t (df)
p
15.1 (43.5) 11.0 (43.0) 11.2 (46.4) 0.9 (75) 2.5 (75) 0.9 (75) 0.5 (75) 1.4 (75) 0.6 (75)
< .001 < .001 < .001 .352 .014 .353 .638 .158 .561
Df: degrees of freedom; M: mean; p: p-value; PCL-R: Hare Psychopathy Checklist-Revised; SD: standard deviation; STAXI: State-Trait Anger Expression Inventory; t: t-value; VRAG: Violence Risk Appraisal Guide; VRS: Violence Risk Scale.
25
Table 2: Grey matter volume differences between prisoners and controls.
Region
Prisoners > Controls Cerebellar hemisphere Vermis Caudate nucleus Pallidum Controls > Prisoners DMPFC
H
x
y
z
Cluster size .05
Cluster size .01
t
p(FWE)
R R L L
3 6 -6 -9
-51 -42 3 1.5
-45 -12 6 3
11473 310 148 71
989 0 9 7
4.66 3.15 3.46 3.67
.007 .022 .037 .010
R
12
44
30
1614
22
3.80
.027
Cluster size: number of voxels in cluster; DMPFC: dorsomedial prefrontal cortex; H L/R: left or right hemisphere; p(FWE): p-value (corrected for family-wise error); t: t-value; x,y,z: MNI coordinates.
26
Table 3: Results of the correlation analyses. Scale
Within prisoners PCL-R Factor 1 PCL-R Factor 1 PCL-R Factor 2 PCL-R Factor 2 PCL-R Factor 2 PCL-R Factor 2 PCL-R Factor 2 PCL-R Factor 2 PCL-R Factor 2 VRAG Risk Category
VRAG Risk Category VRAG Risk Category VRAG Risk Category VRS Static VRS Dynamic STAXI Temperament STAXI Temperament STAXI Anger Out STAXI Anger Out STAXI Anger Out Within controls STAXI Anger Control STAXI Temperament
Region
Cerebellar hemisphere R DLPFC R SMA R Putamen R Pallidum L OFC L OFC R Insula L Insula R Cerebellar hemisphere L Vermis R OFC L OFC R Vermis R Amygdala R Cerebellar hemisphere L Vermis L Cerebellar hemisphere L Vermis R/L SMA R
OFC L Insula R
Corr
Cluster size .05
Cluster size .01
t
p(FWE)
x
y
z
+
47
-77
-26
1666
335
4.98
.003
+ + + + + + + +
29 18 32 -23 -47 42 -23 36 -9
-11 12 -18 -5 24 35 12 -24 -66
68 65 6 -6 -5 -11 -20 23 -20
409 1028 846 267 7850 8741 2996 1054 2975
93 43 18 1 823 922 22 281 250
4.03 4.25 3.71 3.49 5.11 4.85 3.99 4.44 3.98
.049 .015 .023 .024 .003 .007 .028 .009 .023
+ + + + +
6 -47 44 5 30 -5
-68 27 35 -77 -5 -38
-18 -5 -12 -12 -27 -11
640 5302 6383 340 166 46
4 1148 862 29 0 0
3.43 5.31 5.22 3.77 3.20 3.05
.024 .002 .003 .043 .034 .034
+ +
-3 -18
-56 -65
3 -14
123 3225
1 269
3.36 4.60
.041 .022
+ +
0 17
-62 5
-3 56
1296 266
127 18
4.36 4.23
.010 .015
+
-21 51
45 9
-14 -2
4048 1992
318 90
4.78 3.94
.009 .034
Cluster size: number of voxels in cluster; Corr +/- : positive or negative correlation; DLPFC: dorsolateral prefrontal cortex; L: left; OFC: orbitofrontal cortex; p(FWE): p-value (corrected for familywise error); PCL-R: Hare Psychopathy Checklist-Revised; R: right; SMA: supplementary motor area; STAXI: State-Trait Anger Expression Inventory; t: t-value; VRAG: Violence Risk Appraisal Guide; VRS: Violence Risk Scale; x,y,z: MNI coordinates.
27
HIGHLIGHTS
•
We compared grey matter volume (GMV) of high-risk violent offenders and controls
•
Psychopathic traits and risk for violence recidivism were correlated with GMV
•
Criminals showed reduced GMV in the prefrontal cortex
•
Criminals showed enhanced GMV in the cerebellum and basal ganglia
•
Results might be valuable for improving existing risk assessment tools
S0306-4522(15)00821-0 http://dx.doi.org/10.1016/j.neuroscience.2015.09.011 NSC 16567
To appear in:
Neuroscience
Accepted Date:
3 September 2015
Please cite this article as: V. Leutgeb, M. Leitner, A. Wabnegger, D. Klug, W. Scharmüller, T. Zussner, A. Schienle, Brain abnormalities in high-risk violent offenders and their association with psychopathic traits and criminal recidivism, Neuroscience (2015), doi: http://dx.doi.org/10.1016/j.neuroscience.2015.09.011
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
Brain abnormalities in high-risk violent offenders and their association with psychopathic traits and criminal recidivism
Verena Leutgeba*, Mario Leitnerb, Albert Wabneggera, Doris Kluga,b, Wilfried Scharmüllera, Thomas Zussnera & Anne Schienlea
a
Clinical Psychology, University of Graz, BioTechMedGraz Postal address: Universitätsplatz 2/DG, 8010, Graz, Austria b
Graz-Karlau State Correctional Facility, Graz, Austria Postal address: Herrgottwiesgasse 50, 8200, Graz, Austria
* Corresponding author: Verena Leutgeb University of Graz Universitätsplatz 2/DG A - 8010 Graz, Austria [email protected]
2
ABSTRACT Measures of psychopathy have been proved to be valuable for risk assessment in violent criminals. However, the neuronal basis of psychopathy and its contribution to the prediction of criminal recidivism is still poorly understood. We compared structural imaging data from 40 male high-risk violent offenders and 37 non-delinquent healthy controls via voxel-based morphometry. Psychopathic traits and risk for violence recidivism were correlated with grey matter volume (GMV) of regions of interest previously shown relevant for criminal behavior. Relative to controls, criminals showed less GMV in the prefrontal cortex and more GMV in cerebellar regions and basal ganglia structures. Within criminals, we found a negative correlation between prefrontal GMV and psychopathy. Additionally, there was a positive correlation between cerebellar GMV and psychopathy as well as risk of recidivism for violence. Moreover, grey matter volumes of the basal ganglia and supplementary motor area (SMA) were positively correlated with anti-sociality. GMV of the amygdala was negatively correlated with dynamic risk for violence recidivism. In contrast, GMV of (para)limbic areas (orbitofrontal cortex, insula) was positively correlated with antisociality and risk for violence recidivism. The current investigation revealed that in violent offenders deviations in GMV of the prefrontal cortex as well as areas involved in the motor component of impulse control (cerebellum, basal ganglia, SMA) are differentially related to psychopathic traits and the risk for violence recidivism. The results might be valuable for improving existing risk assessment tools.
Key words: Grey matter volume, risk assessment, cerebellum, basal ganglia, violence
3
1. INTRODUCTION There is emerging interest in developing tools that provide information about recidivism of violent criminal behavior in addition to behavioral measures. Empirically developed risk assessment instruments provide a probabilistic estimate of the violence risk for a specified time period (Hanson and Howard, 2010; Swets et al., 2000). Risk factors that are unchangeable (e.g., school maladjustment, history of alcohol problems, age at index offense) are commonly referred to as ‘static’. In contrast, ‘dynamic’ risk factors (e.g., impulsivity, unstable interpersonal relationships, lack of cooperation) assess the client’s treatment readiness and possibility of change. The Violence Risk Appraisal Guide (VRAG) for example is most commonly used by clinicians for assessing the static risk of violent recidivism of violent offenders (Quinsey et al., 1998). In addition, the Violence Risk Scale (VRS) allows a prospective rating of static and dynamic risk of violence recidivism (Wong and Gordon, 2006). A recent meta-analysis found little variation amongst the mean effect sizes of common actuarial or structured risk instruments (2009; Campbell et al., 2009). Therefore, neuroscientific methods may have the potential to improve existing tools for prediction of violence recidivism (for a recent review, see Meixner, 2014). However, despite the recent growth of knowledge on structural and functional alterations in the brain of individuals, who commit violent crimes, the contribution to criminal law (e.g., questions of credibility, culpability, or prediction of future recidivism) is still low. Psychopathy has been reported to be strongly associated with violence and criminal recidivism (Hare, 1991, 2003). It is a personality construct characterized by deficits in interpersonal relations and affective processes (e.g., fearlessness, callousness, failure to form close emotional bonds, dishonesty, deficits in passive avoidance learning, and deficient empathic responses) as well as antisocial and
4
impulsive behavior (2008; Hare and Neumann, 2008). The Psychopathy ChecklistRevised (PCL-R) is a valid measure of psychopathy (Hare, 2003) and has been successfully introduced as an instrument for risk assessment (Hare et al., 2000). It comprises two factors reflecting emotional and interpersonal detachment (Factor 1) as well as antisocial behavior and parasitic lifestyle (Factor 2). Regarding the twofactor model emotional detachment is the key feature of psychopathy (Hare, 2003). There is an ongoing debate about the question, whether the predictive efficacy of the PCL-R should be attributed more to Factor 1 or to Factor 2. In a recent meta-analytic study (Yang et al., 2010) the authors questioned the predictive quality of Factor 1 for violence in men, as the subscale predicted violence only at chance level. In contrast, Factor 2 showed high predictive quality. A disruption of brain regions underlying moral thinking and feeling might be central for antisocial behavior (for a thorough discussion, see Glenn et al., 2012). In line with this notion, magnetic resonance imaging studies have found significantly reduced grey matter volume (GMV) of the prefrontal cortex in patients with antisocial personality disorder (APD) and/ or high levels of psychopathy which has been interpreted to lead to poor behavioral control (Müller et al., 2008; Raine et al., 2000; Yang et al., 2005). Gregory et al. (2012) found reduced prefrontal GMV in offenders with APD, who were classified as psychopaths, relative to offenders with APD ‘alone’, and suggested that psychopathy represents a distinct phenotype. Moreover, psychopathy and persistent violence have also been associated with decreased grey matter in limbic and paralimbic areas such as the amygdala, the insula, the hippocampus, or the orbitofrontal cortex (e.g., Boccardi et al., 2011; ContrerasRodríguez et al., 2014; Ermer et al., 2012; Yang and Raine, 2009; Oliveira-Souza et al., 2008). Existing results are somehow inconsistent, as Boccardi et al. (2011) reported enlargement of the lateral and central nucleus of the amygdala in offenders
5
relative to controls. Anyway, the notion that psychopathy is associated with frontolimbic grey matter reductions and dysfunctions concerning moral sensibility and behavior seems to be widely accepted (e.g., Oliveira-Souza et al., 2008; for a review, see Simpson, 2012). Despite the fact that the majority of studies found reduced structures in antisocial or psychopathic groups, there is evidence that some brain regions, like basal ganglia and cerebellar cortex, may actually be larger (Glenn et al., 2012; Barkataki et al., 2006; Tiihonen et al., 2008). In addition, there are reports of abnormalities in the supplementary motor area (SMA) in criminals with high PCL-R scores (Müller et al., 2008). As those brain regions are associated with motor control and higher-order functions of motor cognition, structural deviations in samples afflicted with APD might explain impulsivity and poor behavioral control as well as violence recidivism (Bari and Robbins, 2013; Picazio and Koch, 2015). The current investigation was designed to compare structural imaging data of high-risk violent offenders with non-delinquent controls. We expected enhanced GMV in criminals relative to controls in basal ganglia and cerebellar regions, as well as reduced GMV in prefrontal structures and (para)limbic areas (e.g., the amygdala). For psychopathic traits and risk of criminal recidivism we hypothesized positive correlations with GMV of basal ganglia and cerebellar regions as well as the supplementary motor area (SMA), and negative correlations with prefrontal and limbic areas (e.g., amygdala).
2. EXPERIMENTAL PROCEDURES 2.1 Subjects We investigated 40 male inmates from a maximum-security prison located in Graz (Austria) and 37 male, non-delinquent controls with comparable age (prisoners: M = 38.1 years, SD = 12.0; controls: M = 36.7 years, SD = 9.6; t(75) = 0.5, p = .589). To
6
avoid group differences in overall intelligence, participants of the two groups were matched according to their educational level (years of education; prisoners: M = 11.4 years, SD = 1.9; controls: M = 11.6 years, SD = 1.4; t(75) = -0,6, p = .520). Moreover, the groups were matched according to their handedness (73 right-handed; 4 lefthanded). All participating prisoners were high-risk violent offenders (with one or more index convictions for violence). Sexual offenders were not included in the sample. All participants were to be released to the community in the near term (which meant early release/ conditional release or completion of their sentence). A comprehensive clinical psychiatric assessment and structured risk assessment was performed by experienced forensic psychologists or psychiatrists and actuarial instruments (see below). Clinically relevant symptoms of depression, bipolar disorder, psychosis, attention-deficit hyperactivity disorder, as well as organic (e.g., hyper- or hypothyroidism) or neurological conditions (e.g., history of severe head injury) led to exclusion from the sample. Moreover, prisoners with a history of substance and/ or alcohol abuse during imprisonment were excluded. Community healthy control participants were recruited via announcements in a local newspaper. Control group participants who had been convicted for any crime, or had a history of a mental disorder or substance/ alcohol abuse, were excluded. The prisoners were either given permission to leave prison on a day-release or to go out for a few hours a day or were escorted to the MRI-Lab of the University of Graz by officials of the Graz-Karlau State Correctional Facility, Graz, Austria. All participants provided written informed consent after receiving a full explanation of the test procedure. The study was in accordance with the Operational Guidelines for Ethics Committees That Review Biomedical Research (WHO, 2000) and had been approved by the ethics committee of the University of Graz.
7
2.2 Clinical Assessments Clinical rating-data and questionnaire scores for prisoners and controls are summarized in Table 1. For both groups, two independent raters filled out the Hare Psychopathy Checklist-Revised (PCL-R), which is a reliable and valid measure of psychopathic traits in the prison system (Hare, 1991, 2003). The PCL-R consists of 20 items, which have to be rated on a 3-point-scale. It uses a semi-structured interview, case history information, and specific scoring criteria and consists of two factors (2003; Hare, 2003). Factor 1 is comprised by the following two aspects: interpersonal (glibness/ superficial charm, grandiose sense of self-worth, pathological deception, conning/ manipulative) and affective (lack of remorse or guilt, shallow affect, callous/ lack of empathy, failure to accept responsibility for actions). Factor 2 is comprised by the following two aspects: lifestyle (need for stimulation/ proneness to boredom,
parasitic
lifestyle,
lack
of
realistic
long-term
goals,
impulsivity,
irresponsibility) and antisocial (poor behavioral control, early behavioral problems, juvenile delinquency, revocation of conditional release, criminal versatility). Inter-rater reliabilities (Pearson’s correlations) for the sum score and subscales (Factor 1 and 2) of the PCL-R were sufficiently high and ranged for prisoners from r = .80 to .88, and for controls from .81 to .96. Two scales, focusing on risk of violence recidivism, and thus on offense history, were additionally filled out for the prisoners by two independent raters, but not for controls: we obtained the Violence Risk Appraisal Guide (VRAG), which was developed to assess risk for violent recidivism (Rossegger et al., 2009). The scale contains 12 items reflecting static risk factors (e.g., age at index offense, index victim injury, sex of index victim). VRAG scores allow assignment of an offender to one of nine risk categories, ranging from 1 (lowest risk) to 9 (highest risk). The inter-rater reliability (Pearson’s correlation) for the sum score of the VRAG was r = .99. The
8
second scale was the Violence Risk Scale (VRS; Wong and Gordon, 2006). This scale was constructed for purposes of risk assessment and prediction. It integrates violence assessment, prediction, and treatment and consists of 6 items forming a Static Factor and 20 items forming a Dynamic Factor. Each item is rated on a 4-point scale. Static variables, such as offense history, remain unchanged regardless of treatment interventions. In contrast, dynamic variables such as interpersonal aggression or emotional control identify treatment targets linked to violence, and assess the client's treatment readiness and possibility of change. The inter-rater reliabilities (Pearson’s correlations) for the subscales of the VRS were r = .99 for static variables, and r = .98 for dynamic variables. All participants filled out the State-Trait Anger Expression Inventory (STAXI; Schwenkenmezger et al., 1992). The inventory consists of six scales: Trait Anger (frequency of anger experience), Temperament (intense anger in response to mild provocation), Rejection (sensitivity to criticism), Anger-Out (disposition to express anger aggressively), Anger-In (disposition not to act out feelings of anger), AngerControl (disposition to suppress feelings of anger). For this questionnaire, participants are asked to rate 44 items on 4-point scales that assess both the intensity of their anger at a particular time and the frequency that anger is experienced, expressed, and controlled.
2.3 MRI Acquisition and VBM Analyses T1-weighted scans were acquired using a 3T Siemens Skyra with a 32-channel head coil (Siemens, Erlangen, Germany). The scanning parameters were as follows: voxel size: = 0.88 x 0.88 x 0.88 mm; 192 transverse slices, slice thickness: 0.88mm, TE = 1.89 ms, TR = 1680 ms; TI = 1000 ms, flip-angle = 8o.
9
Brain imaging data were analyzed using SPM8 (Wellcome Trust Centre for Neuroimaging; http://www.fil.ion.ucl.ac.uk/spm/software/spm8/) including the VBM8 toolbox
(revision
435;
http://dbm.neuro.uni-jena.de/vbm)
for
voxel-based
morphometry in order to gain voxel-wise comparisons of grey matter volume (GMV). Anatomical scans were segmented into grey matter, white matter, and cerebrospinal fluid partitions. An optimized blockwise non-local means de-noising filter, a Hidden Markov Random Field approach, partial volume estimates, and normalization to MNI space by high-dimensional warping (DARTEL) with a standard template included in the VBM8-toolbox were used for pre-processing (final resolution: 1.5 mm × 1.5 mm × 1.5 mm). A Jacobian modulation for non-linear normalization was applied to correct for differences in head sizes and to obtain brain volume. Smoothing was executed with a Gaussian kernel with a full width at half maximum (FWHM) of 8 mm. Based on previous VBM studies with criminal, antisocial or psychopathic populations (Müller et al., 2008; Raine et al., 2000; Yang et al., 2005; ContrerasRodríguez et al., 2014; Ermer et al., 2012; Barkataki et al., 2006; Tiihonen et al., 2008) the following regions of interest (ROIs) were selected: dorsolateral prefrontal cortex (DLPFC), dorsomedial prefrontal cortex (DMPFC), orbitofrontal cortex (OFC) amygdala, insula, caudate nucleus, pallidum, putamen, cerebellar hemispheres, vermis, and supplementary motor area (SMA). For the present study we used ROIs from the automated anatomical labeling (AAL) template. The ROIs were constructed with the WFU PickAtlas (version 2.4; Wake Forest University School of Medicine). The DLPFC was built from masks of the middle frontal and the superior frontal gyrus, whereas the DMPFC was built from the frontal superior medial mask. Following masks were used (mask sizes in voxel (v) are provided in brackets): left DLPFC (21448 v), right DLPFC (23502 v), left DMPFC (8095 v), right DMPFC (6093 v), left OFC (13327 v), right OFC (13895 v), left amygdala (576 v), right amygdala (673 v),
10
left insula (4941 v), right insula (4842 v), left caudate nucleus (2649 v), right caudate nucleus (2737 v), left pallidum (1153 v), right pallidum (1060 v), left putamen (2688 v), right putamen (3776 v), left cerebellar hemisphere (25825 v), right cerebellar hemisphere (27042 v), vermis (4967 v), left SMA (5781 v), and right SMA (6291 v). We report analyses with initial thresholds .05 uncorrected. For purposes of comparability, we additionally provide analyses with initial thresholds .01 uncorrected. Peaks for voxel intensity tests are reported when significant (p < .05 corrected for family-wise error, small volume correction).
2.4 Statistical Analyses Statistical analyses were carried out using random effects models. To test for differences in GMV between patients and controls a two samples t-test (voxel intensity test) was conducted (comparisons: Criminals <> Controls). Scores of the PCL-R, the VRAG risk category, the VRS, and the STAXI were correlated via a simple regression with GMV within criminals and controls, respectively. Due to insufficient variability, we conducted no regression analyses for the PCL-R within controls. To restrict analysis to grey matter, images were thresholded for all analyses with an absolute threshold of 0.2.
3. RESULTS 3.1 Clinical Assessments The prisoners obtained higher scores than controls on both factors of the PCL-R and on the subscale Temperament of the STAXI (see Table 1).
Insert Table 1 here
11
3.2 VBM Analyses 3.2.1 Group differences In the ROI analysis, the prisoners had more GMV than controls in cerebellar regions, the caudate nucleus, and the pallidum (see Table 2). There were no group differences in other ROIs. Controls had more GMV in the DMPFC than prisoners (see Figure 1). A whole-brain analysis did not reveal any significant differences in GMV between prisoners and controls.
Insert Table 2 here Insert Figure 1 here
3.2.2 Regression analyses within prisoners Psychopathic traits (PCL-R, Factor 1) were positively correlated with cerebellar volume (see Figure 2) and negatively correlated with GMV of the DLPFC (see Table 3). Moreover, antisocial behavior (PCL-R, Factor 2) positively correlated with the volume of the SMA, basal ganglia regions (putamen and pallidum), as well as (para)limbic regions (OFC, insula). Scores for static risk prediction (VRAG Risk Category; VRS Static) positively correlated with GMV of cerebellar regions and (para)limbic regions (OFC, insula). VRS Dynamic negatively correlated with GMV of the
amygdala.
There
were
positive
correlations
between
the
subscales
‘Temperament’ and ‘Anger Out’ of the STAXI and GMV of cerebellar regions, as well as SMA.
3.2.3 Regression analyses within controls:
12
There was a negative correlation for the subscale ‘Anger Control’ of the STAXI and GMV of the OFC. Moreover, there was a positive correlation between the subscale ‘Temperament’ of the STAXI and GMV of the insula.
Insert Figure 2 here Insert Table 3 here
4. DISCUSSION This VBM study focused on differences in grey matter volume (GMV) between incarcerated high-risk violent offenders and non-delinquent healthy controls. Moreover, scores reflecting psychopathic traits (according to the Hare Psychopathy Checklist-Revised, PCL-R (Hare, 2003), risk for violent recidivism (Violence Risk Appraisal Guide, VRAG (Rossegger et al., 2009) and Violece Risk Scale, VRS (Wong and Gordon, 2006), as well as anger state-trait anger expression/ management, STAXI (Schwenkenmezger et al., 1992) were correlated with GMV. The main result of the current study revealed enhanced regional GMV in criminals relative to controls in areas associated with the motor component of impulse control (Picazio and Koch, 2015; Glenn et al., 2010), namely the cerebellum, and basal ganglia (caudate nucleus, pallidum). Moreover, within criminals there were consistent positive associations between GMV of the cerebellar cortex, (para)limbic structures (OFC, insula), basal ganglia, as well as the supplementary motor area (SMA) with psychopathic traits, anti-sociality, risk of criminal recidivism, and poor anger management. Similar results have been reported earlier by a research group, who found a 9.6% increase in striatal volume in psychopathic individuals relative to comparison subjects (2010; Glenn et al., 2010). Moreover, investigations showed increases in
13
putamen volume in patients afflicted with antisocial personality disorder (2006; Barkataki et al., 2006). As they also observed this increase in violent relative to nonviolent schizophrenics, the authors hypothesized a general involvement of this structure in violence. This seems plausible, as striatal structures have been found to be involved in impulsivity and poor behavioral control (Garavan, 2002). In line, our regression analysis within criminals revealed a positive relationship between antisocial behavior (Factor 2 of the PCL-R) and GMV of putamen and pallidum. Another brain region that might have a relevant contribution to psychopathic traits is the cerebellum, due to its involvement in motor control (for a discussion, see Picazio and Koch, 2015). To our knowledge, there is one investigation that found larger GMV of the cerebellum in persistently violent offenders with antisocial personality disorder relative to healthy controls (Tiihonen et al., 2008). Within the current study, there was a very consistent positive relationship between predictors of violent crimes and cerebellar GMV. This was true for interpersonal and affective problems associated with psychopathy (Factor 1 of the PCL-R), static risk factors and criminal relapse rates. Additionally, measures for temperament and expression of anger correlated positively with cerebellar GMV. Therefore, our data point to an involvement of the cerebellum in psychopathic traits and criminal recidivism. The enhanced cerebellar grey matter volume in criminals relative to controls might be interpreted to explain heightened impulsivity and poor behavioral control. In line, our regression analysis within criminals revealed a positive relationship between antisocial behavior (Factor 2 of the PCL-R) and GMV of the SMA, a brain region that is also known to be involved in motor control and impulsivity (for a recent review, see Bari and Robbins, 2013). In addition, expression of anger was positively correlated with the volume of the SMA.
14
The cerebellum has also been reported to be critically involved in moral behavior and aggression and possess various anatomical connections, also with the PFC (see Demirtas-Tatlidede and Schmahmann, 2013; Fumagalli and Priori, 2012; Turner et al., 2007). Thus, it seems interesting that in line with several structural studies with anti-social (e.g., Raine et al., 2000; Tiihonen et al., 2008) or psychopathic (e.g., Müller et al., 2008; Yang et al., 2005; Contreras-Rodríguez et al., 2014) individuals we found reduced GMV in the prefrontal cortex (PFC) in criminals relative to controls. Structural and functional changes of the PFC are the bestreplicated abnormalities across a wide range of antisocial groups and have also been obtained with different imaging methodologies (Glenn et al., 2012). The involvement of the PFC in higher-order executive functions can easily explain emotional deficits, poor decision-making, and the lack of behavioral control in violent offenders. In line, the regression analysis of our data within criminals revealed a negative correlation between interpersonal and affective problems associated with psychopathic traits (Factor 1 of the PCL-R; e.g., superficial charm, a manipulative nature, and a lack of empathy) and GMV of the dorsolateral PFC. Lastly, the current investigation failed to detect reductions in GMV of (para)limbic areas in offenders relative to controls. However, we found a negative relationship between dynamic risk factors for criminal relapse and grey matter of the amygdala. The amygdala is known to be crucial for emotional re-learning and conditioning (LeDoux, 2000). In fact, in an fMRI investigation psychopaths showed deficient fear conditioning and reduced amygdala activation relative to healthy controls (2005; Birbaumer et al., 2005). A deficiency to emotionally relate to neutral and biologically significant events might be central in psychopathy. Our study suggests that there is an association between reduced amygdala volume and dynamic risk factors for violence. In contrast, we found a positive relationship
15
between antisocial behavior (Factor 2 of the PCL-R) as well as criminal relapse rates and GMV of (para)limbic areas (OFC, insula). Summing up the correlations between GMV and risk assessment scales, both factors of the PCL-R revealed a relationship to GMV in violent offenders. Factor 2, reflecting antisocial behavior and parasitic lifestyle, was positively related to volumes of putamen, pallidum, and the SMA. In contrast, Factor 1, reflecting the core features of psychopathy, revealed a positive relationship with the cerebellum and a negative relationship with the dorsolateral PFC. Thus, although the predictive quality of Factor 1 for violence in men has been questioned in previous investigations (Yang et al., 2010), our data provide support for a differential relationship between the core features of psychopathic personality traits and GMV. Thus, in future investigations, the association between cerebellar volume and Factor 1 of the PCL-R should be further investigated. In addition, risk for violence recidivism (VRAG, VRS) was positively related to cerebellar GMV and negatively related to amygdala volume.
5. CONCLUSIONS Abnormalities in GMV of the prefrontal cortex as well as areas involved in the motor component of impulse control (cerebellum, basal ganglia, SMA) seem to be related to psychopathic traits and the risk of criminal recidivism in violent offenders. It remains unclear, if the abnormalities are causal for criminality, and moreover, if they are reversible. With respect to this and also as a limitation for the interpretation of group differences it has to be noted, that we did not control for childhood history (e.g., abuse or poverty) or history of alcohol and drug abuse prior to imprisonment. Science is still far from establishing neurobiological criteria as a contribution to final decisions in the court (for a discussion, see Witzel, 2012). However, the integration of objective neurobiological criteria, and an understanding of the interaction of biological and
16
psychological factors could provide a higher quality of risk assessment. In addition, understanding neurobiological correlates of mental disorders could be helpful for improving therapeutic interventions. Lastly, a neurobiological evaluation of treatment success in forensic psychiatry would be of great value for the decision if patients should be discharged, or not. Remedial efforts in the treatment of psychopathy might be most effective when implemented prior to the onset of severe antisocial behavior in the teenage years (2014; Anderson and Kiehl; Anderson and Kiehl, 2014). Future studies should therefore focus on changes in regional brain structures associated with therapeutic treatment. Moreover, future investigations should explore the neuronal basis of the relationship between self-report measures of psychopathic traits and criminal behavior.
ACKNOWLEDGEMENT We received no funding for the current study. There are no conflicts of interest or financial disclosures for any of the authors.
REFERENCES Anderson NE, Kiehl KA (2014) Psychopathy: Developmental perspectives and their implications for treatment. Restor Neurol Neuros 32:103–117. Bari A, Robbins TW (2013) Inhibition and impulsivity: Behavioral and neural basis of response control. Prog Neurobiol 108:44–79. Barkataki I, Kumari V, Das M, Taylor P, Sharma T (2006) Volumetric structural brain abnormalities in men with schizophrenia or antisocial personality disorder. Behav Brain Res 169:239–247.
17
Birbaumer N, Veit R, Lotze M, Erb M, Hermann C, Grodd W, Flor H (2005) Deficient fear conditioning in psychopathy: A functional magnetic resonance imaging study. Arch Gen Psychiat 62:799–805. Boccardi M, Frisoni GB, Hare RD, Cavedo E, Najt P, Pievani M, Rasser PE, Laakso MP, Aronen HJ, Repo-Tiihonen E, Vaurio O, Thompson PM, Tiihonen J (2011) Cortex and amygdala morphology in psychopathy. Psychiatry research 193:85– 92. Campbell MA, French S, Gendreau P (2009) The Prediction of Violence in Adult Offenders: A Meta-Analytic Comparison of Instruments and Methods of Assessment. Crim Justice Behav 36:567–590. Contreras-Rodríguez O, Pujol J, Batalla I, Harrison BJ, Soriano-Mas C, Deus J, López-Solà M, Macià D, Pera V, Hernández-Ribas R, Pifarré J, Menchón JM, Cardoner N (2014) Functional Connectivity Bias in the Prefrontal Cortex of Psychopaths. Biol Psychiat:in press. Demirtas-Tatlidede A, Schmahmann JD (2013) Morality: incomplete without the cerebellum? Brain a journal of neurology 136:e244. Ermer E, Cope LM, Nyalakanti PK, Calhoun VD, Kiehl KA (2012) Aberrant paralimbic gray matter in criminal psychopathy. J Abnormal Psych 121:649–658. Fumagalli M, Priori A (2012) Functional and clinical neuroanatomy of morality. Brain a journal of neurology 135:2006–2021. Garavan H (2002) Dissociable Executive Functions in the Dynamic Control of Behavior: Inhibition, Error Detection, and Correction. NeuroImage 17:1820–1829. Glenn AL, Raine A, Yaralian PS, Yang Y (2010) Increased Volume of the Striatum in Psychopathic Individuals. Biol Psychiat 67:52–58. Glenn AL, Yang Y, Raine A (2012) Neuroimaging in Psychopathy and Antisocial Personality Disorder: Functional significance and a neurodevelopmental
18
hypothesis. In: Neuroimaging in Forensic Psychiatry: From the clinic to the courtroom (Simpson JR, eds.), pp 81–98. Oxford, UK: Wiley-Blackwell. Gregory S, ffytche D, Simmons A, Kumari V, Howard M, Hodgins S, Blackwood N (2012) The antisocial brain: psychopathy matters. Archives of general psychiatry 69:962–972. Hanson RK, Howard PD (2010) Individual confidence intervals do not inform decision-makers about the accuracy of risk assessment evaluations. Law Hum Behav 34:275–281. Hare RD (1991) The Hare Psychopathy Checklist-Revised. Toronto, ON: Multi-Health Systems. Hare RD (2003) The Hare Pysychopathy Checklist-Revised. 2nd ed. Toronto, ON: Multi-Health Systems. Hare RD, Clark D, Grann M, Thornton D (2000) Psychopathy and the predictive validity of the PCL-R: an international perspective. Behav Sci Law 18:623–645. Hare RD, Neumann CS (2008) Psychopathy as a Clinical and Empirical Construct. Annu Rev Clin Psychol 4:217–246. LeDoux JE (2000) Emotion circuits in the brain. Annu Rev Neurosci 301:155–167. Meixner JB (2014) Applications of Neuroscience in Criminal Law: Legal and Methodological Issues. Curr Neurol Neurosci Rep 15:513. Müller JL, Gänßbauer S, Sommer M, Döhnel K, Weber T, Schmidt-Wilcke T, Hajak G (2008) Gray matter changes in right superior temporal gyrus in criminal psychopaths. Evidence from voxel-based morphometry. Psychiat Res-Neuroim 163:213–222. Oliveira-Souza R de, Hare RD, Bramati IE, Garrido GJ, Azevedo Ignácio F, TovarMoll F, Moll J (2008) Psychopathy as a disorder of the moral brain: Fronto-
19
temporo-limbic grey matter reductions demonstrated by voxel-based morphometry. NeuroImage 40:1202–1213. Picazio S, Koch G (2015) Is Motor Inhibition Mediated by Cerebello-cortical Interactions? The Cerebellum 14:47–49. Quinsey VL, Harris GT, Rice ME, Cormier CA (1998) Violent offenders: appraising and managing risk. 1st ed. Washington DC: American Psychological Association. Raine A, Lencz T, Bihrle S, LaCasse L, Colletti P (2000) Reduced prefrontal gray matter volume and reduced autonomic activity in antisocial personality disorder. Arch Gen Psychiat 57:119–127. Rossegger A, Urbaniok F, Danielsson C, Endrass J (2009) Der Violence Risk Appraisal Guide (VRAG): ein Instrument zur Kriminalprognose bei Gewaltstraftätern. Fortschr Neurol Psychiatr 77:577-584. Schwenkenmezger P, Hodapp V, Spielberger CD (1992) State-Trait-ÄrgerausdrucksInventar. Bern: Verlag Hans Huber. Simpson, J. R. ed. (2012) Neuroimaging in Forensic Psychiatry: From the clinic to the courtroom. Oxford, UK: Wiley-Blackwell. Swets JA, Dawes RM, Monahan J (2000) Better decisions through science. Sci Am 283:82–87. Tiihonen J, Rossi R, Laakso MP, Hodgins S, Testa C, Perez J, Repo-Tiihonen E, Vaurio O, Soininen H, Aronen HJ, Könönen M, Thompson PM, Frisoni GB (2008) Brain anatomy of persistent violent offenders: More rather than less. Psychiat Res-Neuroim 163:201–212. Turner BM, Paradiso S, Marvel CL, Pierson R, Boles Ponto, Laura L., Hichwa RD, Robinson RG (2007) The cerebellum and emotional experience. Neuropsychologia 45:1331–1341.
20
WHO (2000) Operational guidelines for ethics committees that review biomedical research. Geneva: World Health Organization. Witzel J (2012) Implications of neuroimaging for dangerousness assessment. In: Neuroimaging in Forensic Psychiatry: From the clinic to the courtroom (Simpson JR, eds.), pp 195–200. Oxford, UK: Wiley-Blackwell. Wong S, Gordon A (2006) The validity and reliability of the Violence Risk Scale: a treatment-friendly violence risk assessment tool. Psychol Public Pol L 12:279– 309. Yang M, Wong, Stephen C. P., Coid J (2010) The efficacy of violence prediction: a meta-analytic comparison of nine risk assessment tools. Psychol Bull 136:740– 767. Yang Y, Raine A (2009) Prefrontal structural and functional brain imaging findings in antisocial, violent, and psychopathic individuals: A meta-analysis. Psychiat ResNeuroim 174:81–88. Yang Y, Raine A, Lencz T, Bihrle S, LaCasse L, Colletti P (2005) Volume reduction in prefrontal gray matter in unsuccessful criminal psychopaths. Biol Psychiat 57:1103–1108.
FIGURE CAPTIONS
Figure 1: Group differences in grey matter volume.
Figure 2: Positive regression within prisoners between scores of Factor 1 (PCL-R) and grey matter volume of the right cerebellar hemisphere.
21
22
23
24
Table 1: Comparison of clinical rating-data and questionnaire scores between prisoners and controls.
Interview
PCL-R
VRAG Risk Category VRS
STAXI
Sum score Factor 1 Factor 2
Static Dynamic Trait Anger Temperament Rejection Anger In Anger Out Anger Control
Prisoners M (SD) 20.6 (7.7) 7.5 (3.9) 10.8 (5.3) 5.5 (1.7) 7.52 (4.6) 29.0 (11.9) 15.8 (3.4) 7.8 (2.3) 8.0 (1.8) 14.2 (5.1) 11.4 (3.5) 24.6 (4.7)
Controls M (SD) 1.6 (1.8) 0.5 (0.8) 1.1 (1.6) 15.1 (3.1) 6.7 (1.5) 8.4 (2.2) 14.7 (4.6) 10.4 (2.4) 25.2 (5.3)
t (df)
p
15.1 (43.5) 11.0 (43.0) 11.2 (46.4) 0.9 (75) 2.5 (75) 0.9 (75) 0.5 (75) 1.4 (75) 0.6 (75)
< .001 < .001 < .001 .352 .014 .353 .638 .158 .561
Df: degrees of freedom; M: mean; p: p-value; PCL-R: Hare Psychopathy Checklist-Revised; SD: standard deviation; STAXI: State-Trait Anger Expression Inventory; t: t-value; VRAG: Violence Risk Appraisal Guide; VRS: Violence Risk Scale.
25
Table 2: Grey matter volume differences between prisoners and controls.
Region
Prisoners > Controls Cerebellar hemisphere Vermis Caudate nucleus Pallidum Controls > Prisoners DMPFC
H
x
y
z
Cluster size .05
Cluster size .01
t
p(FWE)
R R L L
3 6 -6 -9
-51 -42 3 1.5
-45 -12 6 3
11473 310 148 71
989 0 9 7
4.66 3.15 3.46 3.67
.007 .022 .037 .010
R
12
44
30
1614
22
3.80
.027
Cluster size: number of voxels in cluster; DMPFC: dorsomedial prefrontal cortex; H L/R: left or right hemisphere; p(FWE): p-value (corrected for family-wise error); t: t-value; x,y,z: MNI coordinates.
26
Table 3: Results of the correlation analyses. Scale
Within prisoners PCL-R Factor 1 PCL-R Factor 1 PCL-R Factor 2 PCL-R Factor 2 PCL-R Factor 2 PCL-R Factor 2 PCL-R Factor 2 PCL-R Factor 2 PCL-R Factor 2 VRAG Risk Category
VRAG Risk Category VRAG Risk Category VRAG Risk Category VRS Static VRS Dynamic STAXI Temperament STAXI Temperament STAXI Anger Out STAXI Anger Out STAXI Anger Out Within controls STAXI Anger Control STAXI Temperament
Region
Cerebellar hemisphere R DLPFC R SMA R Putamen R Pallidum L OFC L OFC R Insula L Insula R Cerebellar hemisphere L Vermis R OFC L OFC R Vermis R Amygdala R Cerebellar hemisphere L Vermis L Cerebellar hemisphere L Vermis R/L SMA R
OFC L Insula R
Corr
Cluster size .05
Cluster size .01
t
p(FWE)
x
y
z
+
47
-77
-26
1666
335
4.98
.003
+ + + + + + + +
29 18 32 -23 -47 42 -23 36 -9
-11 12 -18 -5 24 35 12 -24 -66
68 65 6 -6 -5 -11 -20 23 -20
409 1028 846 267 7850 8741 2996 1054 2975
93 43 18 1 823 922 22 281 250
4.03 4.25 3.71 3.49 5.11 4.85 3.99 4.44 3.98
.049 .015 .023 .024 .003 .007 .028 .009 .023
+ + + + +
6 -47 44 5 30 -5
-68 27 35 -77 -5 -38
-18 -5 -12 -12 -27 -11
640 5302 6383 340 166 46
4 1148 862 29 0 0
3.43 5.31 5.22 3.77 3.20 3.05
.024 .002 .003 .043 .034 .034
+ +
-3 -18
-56 -65
3 -14
123 3225
1 269
3.36 4.60
.041 .022
+ +
0 17
-62 5
-3 56
1296 266
127 18
4.36 4.23
.010 .015
+
-21 51
45 9
-14 -2
4048 1992
318 90
4.78 3.94
.009 .034
Cluster size: number of voxels in cluster; Corr +/- : positive or negative correlation; DLPFC: dorsolateral prefrontal cortex; L: left; OFC: orbitofrontal cortex; p(FWE): p-value (corrected for familywise error); PCL-R: Hare Psychopathy Checklist-Revised; R: right; SMA: supplementary motor area; STAXI: State-Trait Anger Expression Inventory; t: t-value; VRAG: Violence Risk Appraisal Guide; VRS: Violence Risk Scale; x,y,z: MNI coordinates.
27
HIGHLIGHTS
•
We compared grey matter volume (GMV) of high-risk violent offenders and controls
•
Psychopathic traits and risk for violence recidivism were correlated with GMV
•
Criminals showed reduced GMV in the prefrontal cortex
•
Criminals showed enhanced GMV in the cerebellum and basal ganglia
•
Results might be valuable for improving existing risk assessment tools