Molecular Psychiatry (2009) 14, 946–953 & 2009 Nature Publishing Group All rights reserved 1359-4184/09 $32.00 www.nature.com/mp ORIGINAL ARTICLE Altered connections on the road to MC Craig1,2, M Catani1,2, Q Deeley1, R Latham1, E Daly1, R Kanaan3, M Picchioni3, PK McGuire3, T Fahy4 and DGM Murphy1 1Section of Brain Maturation, Institute of Psychiatry, De Crespigny Park, London, UK; 2Natbrainlab, Institute of Psychiatry, De Crespigny Park, London, UK; 3Section of Neuroimaging, Institute of Psychiatry, De Crespigny Park, London, UK and 4Department of Forensic Mental Health Science, Institute of Psychiatry, De Crespigny Park, London, UK

Psychopathy is strongly associated with serious criminal behaviour (for example, rape and murder) and recidivism. However, the biological basis of psychopathy remains poorly under- stood. Earlier studies suggested that dysfunction of the and/or (OFC) may underpin psychopathy. Nobody, however, has ever studied the connections (such as the uncinate fasciculus (UF)) linking these structures in psychopaths. Therefore, we used in vivo diffusion tensor magnetic resonance imaging (DT-MRI) tracto- graphy to analyse the microstructural integrity of the UF in psychopaths (defined by a - pathy Checklist Revised (PCL-R) score of X25) with convictions that included attempted murder, manslaughter, multiple rape with strangulation and false imprisonment. We report significantly reduced fractional anisotropy (FA) (P < 0.003), an indirect measure of micro- structural integrity, in the UF of psychopaths compared with age- and IQ-matched controls. We also found, within psychopaths, a correlation between measures of antisocial behaviour and anatomical differences in the UF. To confirm that these findings were specific to the limbic amygdala–OFC network, we also studied two ‘non-limbic’ control tracts connecting the posterior visual and auditory areas to the amygdala and the OFC, and found no significant between-group differences. Lastly, to determine that our findings in UF could not be totally explained by non-specific confounds, we carried out a post hoc comparison with a psychiatric control group with a past history of drug abuse and institutionalization. Our findings remained significant. Taken together, these results suggest that abnormalities in a specific amygdala– OFC limbic network underpin the neurobiological basis of psychopathy. Molecular Psychiatry (2009) 14, 946–953; doi:10.1038/mp.2009.40; published online 9 June 2009 Keywords: psychopathy; ; white matter connections; diffusion tensor imaging;

Introduction ‘acquired sociopathy’ after injury, the orbitofrontal cortex (OFC) and other regions of the Psychopathic personality disorder (psychopathy) is prefrontal cortex (PFC) have been considered impor- characterized by features of emotional detachment tant for personality and social behaviour.6 For exam- and antisocial traits,1 and is strongly associated with ple, OFC is crucial to successful reversal learning criminal behaviour and recidivism.2 It has been esti- (in which previously rewarded stimuli are associated mated, for example, that 15% of the prison popula- with punishment) and reversal learning is signifi- tion are psychopaths and they commit approximately cantly impaired in adult psychopaths7 and in young 50% more criminal offences than non-psychopathic people with psychopathic traits.8 It has also been criminals.3 reported that violent personality-disordered offenders The development of reliable and valid methods have reduced PFC grey matter volume9 and glucose for diagnosing psychopathy (for example, the Hare metabolism,10 and impaired OFC (and limbic) activa- Psychopathy Checklist Revised; PCL-R4) in conjunc- tion during aversive conditioning.11 In contrast, other tion with brain-imaging techniques is converging researchers have argued that amygdala dysfunction is towards the identification of neurobiological mecha- central to the affective deficits and impaired moral nisms that may underpin psychopathy. Since the socialization of psychopathy.12 This latter view is report of the case of Phineas Gage,5 who showed supported by evidence that psychopaths show per- formance deficits in tasks sensitive to amygdala Correspondence: Dr MC Craig, Psychological Medicine, Institute damage,13,14 and have significantly reduced amygdala of Psychiatry, PO50, 16 De Crespigny Park, Denmark Hill, London volume15 and decreased amygdala activation during SE5 8AF, UK. 16 E-mail: [email protected] verbal learning and decreased activity in brain Received 7 August 2008; revised 25 February 2009; accepted 13 regions modulated by amygdala during facial fear April 2009; published online 9 June 2009 processing.17 Altered connections on the road to psychopathy MC Craig et al 947 More recently, the dichotomy between researchers PCL-R scores (that is, X25). Sixteen patients initially postulating whether OFC or amygdala dysfunction is agreed to participate and nine were suitable for central to psychopathy18 has narrowed, and it has magnetic resonance imaging (MRI). Healthy controls been suggested instead that the social and emotional were recruited from the general population through deficits of psychopaths may reflect an interaction the Institute of Psychiatry, Kings College London, and between OFC and amygdala dysfunction.19,20 Analy- the absence of psychopathy was confirmed using sis of the functional and anatomical links between the Hare Psychopathy Checklist: Screening Version these structures offers the potential to move beyond (PCL-SV).26 theories of regional dysfunction towards a more Ethical approval was obtained from the Ethical coherent understanding of the possible brain net- Committee of the South London and Maudsley Trust works underlying psychopathy. However, to date, and Institute of Psychiatry, and St Georges Healthcare nobody has studied the white matter tracts linking Trust. After complete description of the study to the these brain regions. The OFC and the amygdala are patients, written informed consent was obtained. interconnected by fibres belonging to the uncinate fasciculus (UF), whose volume and integrity can be Neuroimaging and data analysis analysed in vivo using diffusion tensor magnetic Magnetic resonance imaging of the brain was carried resonance imaging (DT-MRI) tractography. This is a out on GE Signa 1.5 Tesla LX MRI system (General non-invasive neuroimaging technique that can be Electric, Milwaukee, WI, USA) at the Maudsley used to reconstruct three-dimensional trajectories of Hospital, London. white matter tracts within the living brain,21 and to probe the microstructural integrity of white matter in DT-MRI acquisition. Data were acquired with a wide range of neuropsychiatric conditions.22 40 mT mÀ1 gradients, using an acquisition sequence We therefore used in vivo DT-MRI to dissect and fully optimized for DT-MRI of white matter, providing measure the volume and the microstructural integrity isotropic resolution (2.5 Â 2.5 Â 2.5 mm) and coverage of the UF.21 For each hemisphere, we dissected the of the whole head. The acquisition was gated to the fibres of the UF and counted the number of stream- cardiac cycle using a peripheral gating device placed lines (SLs) as a surrogate of the tract volume. We also on the patients’ forefinger. After the correction for measured the mean fractional anisotropy (FA)— image distortions introduced by the application of the which is an indirect measure of white matter spatial diffusion encoding gradients, the diffusion tensor was organization and integrity.23 We compared the psy- determined in each voxel following the method of chopaths (that is, PCL-R score X2524) with age- and Basser et al.27 The operator (Michael C. Craig) carried IQ-matched controls. We recruited psychopaths from out all dissections blind to the diagnosis. Anatomical three specialist forensic inpatient units. All were consistency with classical descriptions of the tracts of repeat violent offenders with index offences that interest was confirmed by using neuroanatomy text included attempted murder, manslaughter, multiple books and tractography atlases. rape with strangulation and false imprisonment. Tract reconstructions. The trajectories of the UF, inferior longitudinal fasciculus (ILF) and inferior Methods fronto-occipital fasciculus (IFOF) were each recon- Patients structed using an approach that involved dissecting We studied 18 normal intelligence right-handed adult out two regions of interest (ROIs)28 (Figure 1). male volunteers: nine with high PCL-R scores (mean The ILF is a ventral associative bundle with long 28.4, range 25–34), aged 34±12 years, with full-scale and short fibres connecting the occipital and temporal IQ (FSIQ) 94±7, and nine healthy male controls aged lobes.3,28 The long fibres are medial to the short fibres 37±9 years, with FSIQ 91±6. All patients (that is, in and connect occipital visual areas to the amygdala both groups) were unmedicated and screened by and hippocampus. The first (temporal) ROI used to formal psychiatric semi-structured interview using dissect the ILF was defined around the white matter the ICD-10 research criteria,25 and a review of case of the anterior , usually on five axial notes was carried out to exclude any co-morbid slices. The second (occipital) ROI was defined around psychiatric illness or neurological/extra-cerebral dis- the white matter of the occipital lobe, usually on orders that might affect brain function. 13–15 slices. Most psychopaths had a past history of alcohol The UF is a ventral anterior associative bundle and/or substance misuse (polysubstance misuse that connects the anterior temporal lobe (including (n = 3), combined polysubstance and alcohol misuse amygdala and hippocampus) with the medial and (n = 3) and alcohol misuse (n = 1)), but none fulfilled lateral OFC. The first (temporal) ROI was defined the criteria for a substance misuse or dependence in the anterior temporal lobe (MNI, À15 to À19), as disorder within 6 months before scanning, with the described for the ILF. The second (external/extreme exception of one patient who fulfilled the criteria for capsule) ROI was defined around the white matter of harmful use of cocaine. Psychopaths were recruited the anterior floor of the external/extreme capsule. from three specialist forensic inpatient units over a The IFOF is a ventral associative bundle that period of 6 years from a group of 34 patients with high connects the ventral occipital lobe and the OFC. In

Molecular Psychiatry Altered connections on the road to psychopathy MC Craig et al 948

Figure 1 Virtual dissection of the major association pathways connecting to the amygdala and orbitofrontal cortex (OFC). A two-region-of-interest (ROI)-approach was used to carry out virtual dissection of the major association pathways connecting the amygdala and OFC. An anterior ‘frontal’ ROI was defined around the anterior floor of the in five consecutive axial slices (from MNI À6toÀ14). A second ‘temporal’ ROI was defined around the white matter of the anterior temporal lobe in five consecutive axial slices (from MNI À22 to À30). A third ‘occipital’ region was defined in the white matter of the occipital lobe in 10 consecutive axial slices (shown here only from MNI À4 to 4). To dissect the uncinate tract, all fibres passing through the ‘frontal’ and ‘temporal’ regions are shown in yellow. All fibres passing through the frontal and occipital regions are shown in red and correspond to the interior fronto-occipital tract. Finally, all fibres passing through the temporal and occipital regions are shown in green and correspond to the interior longitudinal tract. Dissections were carried out for both hemispheres.

this occipital course, the IFOF runs parallel to the ILF. All fibres passing through the temporal and occi- On approaching the anterior temporal lobe, the fibres pital region are shown in light grey and attributed to of the IFOF gather together and enter the external the ILF. All the SLs passing through the temporal and capsule dorsally to the fibres of the UF. The first ROI external/extreme capsule are considered to belong was delineated around the occipital lobe on approxi- to the UF and are shown in white. Finally, all SLs mately 13–15 contiguous axial slices in the same passing through the occipital and external/extreme manner as the posterior ROI of the ILF. The second capsule are considered to belong to the IFOF. region was defined around the external/extreme At the termination of tracking, the FA—a measure capsule as described above. that quantifies the directionality of diffusion on

Molecular Psychiatry Altered connections on the road to psychopathy MC Craig et al 949 a scale from 0 (when diffusion is totally random) to 1 differences in FA and number of SL. Statistical (when water molecules are able to diffuse along one analyses were corrected for multiple comparisons direction only)—were sampled at regular (0.5 mm) using Bonferroni’s correction. intervals along the tract (facilitated by the B-spline 3 continuous tensor-field approximation ) and the Results means computed. We terminated the fibre tracking when the FA fell below an (arbitrary) threshold group had a similar number of SLs of 0.15. For each tract, the trajectory obtained was in the right and left UF compared with age- and checked to ensure consistency with neuroanatomical IQ-matched controls, but a significantly reduced mean atlases by reconstructing in three dimensions. FA in right UF (psychopaths 0.403±0.014, controls 0.435±0.023, P = 0.003). There were no differences in Analyses of tracts. In our initial analysis, we the FA of the left UF (psychopaths 0.419±0.027; compared the mean number of SL and FA of the UF controls 0.427±0.020, P = 0.448) (Table 1, Figure 2). in psychopaths and controls. We hypothesized that There was no significant difference in the number the mean number of SL and FA of the UF in the of SLs or FA of the left or right ILF and IFOF (Table 1, psychopaths would be significantly less than in the Figure 2). control group. Using a ‘two-factor’ model of psychopathy, we To test the hypothesis that the FA changes were report a significant negative correlation between specific to the UF, we carried out a secondary analysis ‘antisocial behaviour’ (factor 2) scores and total of the volume and microstructural integrity of con- number of SL in the left UF (Pearson’s correla- nections of two ‘non-limbic’ control tracts connecting tion = À0.880, P = 0.004) and right UF (Pearson’s the posterior visual areas to the amygdala (through correlation = À0.884, P = 0.004). Using a ‘four-factor’ the ILF) and the OFC (through the IFOF). We hypo- model, we report a trend towards a negative thesized that in these ‘non-limbic’ tracts, there would be no significant between-group differences in the Table 1 Analysis of between-group differences in DTI mean number of SL or FA. indices We also investigated whether the anatomical differ- ences in the UF of the psychopaths were associated Psychopaths Controls Differences with variation in symptom severity. Early factor (s.d.) (s.d.) (P-value) analyses of the PCL-R suggested that the two dimen- sions reflect ‘emotional detachment’ (Factor 1) and UF (SL) ‘antisocial behavior’ (factor 2).29 We therefore corre- L 19.8 (16.2) 23.0 (9.0) 0.609 lated anatomical variation with these PCL-R factors R 37.1 (31.9) 28.6 (17.4) 0.490 within the psychopaths. More recent factor analy- ILF (SL) ses of the PCL-R have also suggested that 18 of the L 62.6 (38.8) 70.3 (43.6) 0.695 items are underpinned by four factors: Interpersonal R 48.9 (26.2) 49.3 (18.1) 0.975 (factor 1), Affective (factor 2), Lifestyle (factor 3) and 30,31 Antisocial (factor 4). We therefore correlated IFOF (SL) anatomical variation with these PCL-R factors within L 77.2 (54.2) 67.9 (51.8) 0.714 the psychopaths using a ‘two-factor’ and a ‘four- R 62.4 (55.8) 59.1 (41.2) 0.887 factor’ model. Finally, as noted above, the psychopaths had a UF (FA) history of substance misuse and institutionalization, L 0.42 (0.03) 0.43 (0.02) 0.448 and this may have affected the significant difference R 0.40 (0.01) 0.44 (0.02) 0.003* we found in the UF. Hence, we carried out a post hoc ILF (FA) analysis comparing FA in the UF of psychopaths L 0.48 (0.02) 0.48 (0.02) 0.883 with that in 11 patients who had a past history of R 0.48 (0.02) 0.48 (0.03) 0.690 alcohol/substance misuse and institutionalized care for psychotic mental illness. These patients did not IFOF (FA) differ significantly in age (33±6 years), IQ (95±5) L 0.44 (0.02) 0.46 (0.02) 0.241 or handedness, and were screened to exclude the R 0.45 (0.02) 0.46 (0.03) 0.448 presence of an Axis II diagnosis (for example, antisocial personality disorder) using the Schedule Abbreviations: FA, fractional anisotropy; IFOF, inferior for Affective Disorders and , Lifetime fronto-occipital fasciculus; ILF, inferior longitudinal fascicu- version (SADS-L).32 lus; L, left; R, right; SL, streamlines; UF, uncinate fasciculus; DTI, diffusion tensor imaging. *P < 0.005. Statistical analysis Psychopaths had similar number of SL in the right and left Statistical comparisons of the data were carried out UF compared with controls, but a significantly reduced using SPSS software (SPSS Inc., Chicago, IL, USA). mean FA in the right UF (P = 0.003). There were no Student’s t-test (two-tailed) for independent samples differences in the FA of the left UF (P = 0.448) or in the was used to investigate tract-specific mean group FA/SL of two ‘non-limbic’ control tracts: the ILF and IFOF.

Molecular Psychiatry Altered connections on the road to psychopathy MC Craig et al 950

Figure 2 Tract-specific measurements of fractional anisotropy (FA) in the psychopathy and control group. Psychopaths had a significantly reduced mean FA in the right UF (P = 0.003). There were no differences in the FA of the left UF (P = 0.448) or in the FA of two ‘non-limbic’ control tracts: the ILF and IFOF.

correlation between antisocial (factor 4) scores and In addition, we examined (post hoc) whether our FA and total number of SL in the right UF (Pearson’s findings could simply be explained by differences in correlation = À0.797, P = 0.058 and Pearson’s correla- substance misuse and/or institutionalization. tion = À0.794, P = 0.059, respectively) and affective Taken together, our findings suggest that abnormal (factor 3) scores and total number of SL in the left ‘connectivity’ in the amygdala–OFC limbic network UF (Pearson’s correlation = À0.792, P = 0.06) in may contribute to the neurobiological mechanisms psychopaths (Table 2). underpinning the impulsive, antisocial behaviour In our post hoc analysis, comparing psychopaths and emotional detachment associated with psycho- with patients with an earlier history of substance pathy. This hypothesis is supported by findings of an misuse and institutionalization, the psychopath group association between UF dysfunction, impulsivity and had a significantly reduced mean FA in the right UF reactive aggression. For example, earlier studies have (psychopaths 0.403±0.014, controls 0.437±0.016, reported (a) UF damage in many cases of Kluver–Bucy P < 0.000). In addition, the mean FA was reduced in syndrome (that is, a disconnection syndrome char- the left UF (psychopaths 0.419±0.027, controls acterized by aggressive behaviour, loss of normal 0.460±0.024, P = 0.002) and there was no significant anger and fear responses, decreased inhibition and difference in the number of SLs or FA of the left other personality changes),33 (b) reduced UF FA in or right ILF and IFOF after correction for multiple children showing impulsive traits after early severe comparisons. socio-emotional deprivation34 and (c) reduced func- tional connectivity between the amygdala and OFC in impulsive aggressive borderline personality- Discussion disordered patients.35 It could therefore be argued that In summary, we report significantly reduced FA in our findings are not specific to psychopathy per se; the right UF of psychopaths compared with age- and but may underpin antisocial behaviour in general. IQ-matched controls. Further, within psychopaths, However, to date, there have been no studies that we report associations between measures of antisocial have specifically analysed the UF in individuals with behaviour and anatomical differences in the UF. antisocial personality disorder. Further, post-mortem To confirm that these findings were specific to the and DT-MRI studies that have analysed the UF in limbic amygdala–OFC network, we studied two ‘non- schizophrenia, that is, another mental illness asso- limbic’ control tracts connecting the posterior visual ciated with aggressive and violent behaviour,36 have and auditory areas to the amygdala and the OFC, reported equivocal results.37–39 However, this is clearly and found no significant between-group differences. an important issue for future studies to address.

Molecular Psychiatry Altered connections on the road to psychopathy MC Craig et al 951 Table 2 Correlation between the ‘two factors’ and ‘four factors’ that underpin the PCL-R score and the microstructural integrity of the uncinate fasciculus

Two-factor model F1 P-value F2 P-value Total P-value

UF (FA) L 0.010 0.981 À0.020 0.962 À0.081 0.836 R 0.134 0.752 À0.344 0.404 À0.476 0.195

UF (SL) L 0.608 0.110 À0.880 0.004* À0.056 0.888 R 0.274 0.512 À0.884 0.004* À0.318 0.404

Four-factor model f1 P-value f2 P-value f3 P-value f4 P-value

UF (FA) L À0.152 0.774 0.200 0.704 0.085 0.872 0.570 0.237 R 0.368 0.473 0.433 0.392 À0.628 0.182 0.797 0.058

UF (SL) L 0.361 0.482 0.327 0.527 À0.792 0.060 À0.528 0.282 R 0.349 0.497 0.004 0.994 À0.718 0.108 À0.794 0.059

Abbreviations: FA, fractional anisotropy; L, left; R, right; SL, streamlines; UF, uncinate fasciculus. Two-factor model: F1, factor 1‘emotional detachment’; F2, factor 2‘antisocial behavior’; *P < 0.005. Four-factor model: f1, factor 1‘interpersonal’; f2, factor 2 ‘affective’; f3, factor 3‘lifestyle’; f4, factor 4‘antisocial’. In a two-factor analysis, there was a significant negative correlation between factor 2 scores and total number of SL in the left UF (P = 0.004) and right UF (P = 0.004) in psychopaths. In a four-factor analysis, there was a trend towards a negative correlation between factor 4 (antisocial) scores and FA and the total number of SL in the right UF (P = 0.058, 0.059, respectively) and factor 3 (affective) scores and the total number of SL in the left UF (P = 0.06) in psychopaths.

Lesion studies also suggest that executive function Thus, in summary, earlier studies support our and impulse control may be lateralized to the right findings of an association between reduced micro- hemisphere,40 and this might help explain why our structural integrity of the UF and behavioural traits main findings were limited to the right UF. Further, that characterize psychopathy. However, it remains the right UF has been reported to play a pivotal role unclear whether this is because of a primary patho- in the recollection of affect-laden autobiographical logy in the UF white matter, or whether it is second- memory triggered by sensory stimuli (‘ecphory’).41 ary to abnormalities in the amygdala and/or the This relevance of this finding to our study is that OFC. Further, the biological mechanisms underlying psychopaths have been reported to show poorer reduced UF microstructural integrity also remain memory for affect-laden material compared with unclear. One possible mechanism may include tract- non-psychopathic offenders42 and healthy controls.43 specific deficits in axonal maturation and/or myeli- Therefore, although highly speculative, reduced nation. This suggestion is based on the fact that FA microstructural integrity of the right UF could signal indirectly reflects the degree of myelination contribute to deficits in the processing of emotional and anatomical arrangement of axonal fibres,45 and autobiographical memory that may underpin psycho- that the number of SL is an indirect index of tract pathic traits such as shallow affect and lack of volume (which may also reduce in proportion with empathy. axonal number and/or myelination). Finally, it could be argued that a past history of Our study has a number of weaknesses. For institutionalization, alcohol and/or substance misuse example, we only studied a relatively small number in most psychopaths is responsible for the anatomical of psychopaths, and so our findings need to be differences that we found. Although FA in the white replicated in a larger study. However, the difficulties matter of patients with a past history of alcohol involved with recruiting and scanning our study misuse has been reported earlier, changes tend to be group cannot be over-emphasized. To control for diffuse and affect multiple tracts rather than being potential confounds as best as possible, we limited localized to a single tract.44 Further, we carried out our investigation to a group of psychopaths without a post hoc analysis comparing psychopaths with co-morbid mental illness, who were off medication patients who had a past history of alcohol/substance and not currently engaging in substance misuse. misuse and institutionalized care. This post hoc Nonetheless, the psychopaths we recruited had analysis confirmed our earlier findings in that the committed the most serious criminal offences defined psychopath group still had a significantly reduced by Law (for example, rape, attempted murder, etc.) mean FA in the right UF. with minimal guilt or remorse. As predicted, most

Molecular Psychiatry Altered connections on the road to psychopathy MC Craig et al 952 individuals we identified in this group refused to 11 Veit R, Flor H, Erb M, Hermann C, Lotze M, Grodd W et al. engage in medical research. In psychopaths who Brain circuits involved in emotional learning in antisocial fulfilled the criteria for entering the study, and who behavior and social phobia in humans. Neurosci Lett 2002; 328: 233–236. agreed to aid our investigation, we then had to 12 Blair RJR. Dysfunctions of medial and lateral orbitofrontal cortex negotiate complex security/safety issues. Hence, it is in psychopathy. Ann NY Acad Sci 2007; 1121: 461–479. extremely difficult to find/recruit very dangerous 13 Levenston GK, Patrick CJ, Bradley MM, Lang PJ. The psychopath psychopaths—especially those who do not have as observer: emotion and attention in picture processing. J Abnorm co-morbid mental health problems and/or current Psychol 2000; 109: 373–386. 14 Blair RJR, Colledge E, Murray L, Mitchell DG. A selective drug abuse. These factors contributed to the small impairment in the processing of sad and fearful expressions in sample included in this study and are likely to children with psychopathic tendencies. J Abnorm Child Psychol impede larger studies of this population. Lastly, it is 2001; 29: 491–498. possible that early-onset, longstanding, abnormalities 15 Tiihonen J, Hodgins S, Vaurio O, Laasko M, Repo E, Soininen H et al. Amygdaloid volume loss in psychopathy. Soc Neurosci Abst in social interaction may themselves affect brain 2000; 15: 2017. development. Hence, future studies of neurobiologi- 16 Kiehl KA, Smith AM, Hare RD, Mendrek A, Forster BB, Brink J cal differences in ‘at-risk’ populations of young indi- et al. Limbic abnormalities in affective processing by criminal viduals that may precede significant violent offending psychopaths as revealed by functional magnetic resonance behaviour are also required. imaging. Biol Psychiatry 2001; 50: 677–684. 17 Deeley Q, Daly E, Surguladze S, Tunstall N, Mezey G, Beer D et al. In conclusion, our findings may reconcile the dichot- Facial emotion processing in criminal psychopathy: Preliminary omy suggested by earlier regional-based theories of OFC functional magnetic resonance imaging study. Br J Psychiatry or amygdala dysfunction18 in psychopathy, and lend 2006; 189: 533–539. support to a network-based model.19,20 Confirmation of 18 Abbott A. Into the mind of a killer. Nature 2001; 410: 296–298. 19 Blair RJR. The amygdala and ventromedial prefrontal cortex in these findings by larger studies may have wider medico- morality and psychopathy. Trends Cog Sci 2007; 11: 387–392. legal implications and may ultimately provide a focus 20 van Honk J, Schutter DJG. Unmasking feigned sanity: a neurobio- for the development of treatments for psychopathy. logical model of emotion processing in primary psychopathy. Cog Neuropsych 2006; 11: 285–306. 21 Catani M, Howard RJ, Pajevic S, Jones DK. Virtual in vivo Conflict of interest interactive dissection (VIVID) of white matter fasciculi in the . Neuroimage 2002; 17: 77–94. The authors declare no conflict of interest. 22 Catani M. Diffusion tensor magnetic resonance imaging tractogra- phy in cognitive disorders. Curr Opin Neurol 2006; 19: 599–606. 23 Beaulieu C. The basis of anisotropic water diffusion in the nervous Acknowledgments system—a technical review. NMR Biomed 2002; 15: 435–455. 24 Cooke DJ, Michel CM. Psychopathy Across Cultures. Kluwer We thank Drs Gill Mezey, Dominic Beer, Amory Academic Publishers: Norwell, MA, 1998. Clarke, Timothea Toulpoulou and John Dowsett for 25 The ICD-10 Classification of Mental and Behavioral Disorders— their help in this study. Diagnostic Criteria for Research. World Health Organisation. Geneva, 1983. 26 Hart SD, Cox DN, Hare RD. The Hare Psychopathy Checklist: References Screening Version (PCL:SV). Multi-Health Systems: Toronto, Canada, 1995. 1 Patrick CJ, Bradley MM, Latig PJ. Emotion in the criminal 27 Basser PJ, Matiello J, Le Bihan D. Estimation of the effective self- psychopath: Startle reflex modulation. J Abnorm Psychol 1993; diffusion tensor from the NMR spin echo. J Magn Reson Ser B 102: 82–92. 1994; 103: 247–254. 2 Hare RD, Cooke DJ, Hart RD. Psychopathy and sadistic personality 28 Catani M, Jones DK, Donato R, ffytche DH. Occipito-temporal disorder. In: Millon T, Blaney PH, Davies RD (eds). Oxford connections in the human brain. Brain 2003; 126: 2093–2107. Textbook of Psychopathology. Oxford University Press: Oxford, 29 Patrick CJ. Emotion and psychopathy: startling new insights. UK, 1999. Psychophysiology 1994; 31: 319–330. 3 Hart SD, Hare RD. Psychopathy: assessment and association with 30 Neumann CS, Kosson DS, Forth AE, Hare RD. Factor structure of criminal conduct. In: Stoff DM, Breiling J (eds). Handbook of the Hare Psychopathy Checklist: Youth Version (PCL: YV) in Antisocial Behaviour. New York: John Wiley & Sons, Inc, 1997. incarcerated adolescents. Psychological Assessment 2006; 18: 4 Hare RD (ed). The Hare Psychopathy Checklist-Revised, 2nd edn, 142–154. Multi-Health Systems: Toronto, Ontario, Canada, 2003. 31 Hare RD, Neumann CS. Structural models of psychopathy. Curr 5 Damasio H, Grabowski T, Frank R, Galaburda AM, Damasio AR. Psychiatry Reports 2005; 7: 57–64. The return of Phineas Gage: clues about the brain from the skull of 32 Spitzer RL, Endicott J. Schedule for Affective Disorders and a famous patient. Science 1994; 264: 1102–1105. Schizophrenia. Lifetime Version (SADS-L). New York State 6 Damasio AR. A neural basis for sociopathy. Arch Gen Psychiatry Psychiatric Institute: New York, 1978. 2000; 57: 128–129. 33 Lilly R, Cummings JL, Benson DF, Frankel M. The human Kluver- 7 Budhani S, Richell RA, Blair RJ. Impaired reversal but intact Bucy syndrome. Neurology 1983; 33: 1141–1145. acquisition: probabilistic response reversal deficits in adult in- 34 Eluvathingal TJ, Chugani HT, Behen ME, Juhasz C, Muzik O, dividuals with psychopathy. J Abnorm Psychol 2006; 115: 552–558. Maqbool M et al. Abnormal brain connectivity in children after 8 Budhani S, Blair RJR. Response reversal and children with early severe socioemotional deprivation: a diffusion tensor psychopathic tendencies: success is a function of salience of imaging study. Pediatrics 2006; 117: 2093–2100. contingency change. JChild Psychol Psychiatry 2005; 46: 972–981. 35 Grant JE, Correia S, Brennan-Krohn T, Malloy PF, Laidlaw DH, 9 Raine A, Lencz T, Bihrle S, LaCasse L, Colletti P. Reduced prefrontal Schulz SC. Frontal white matter integrity in borderline personality gray matter volume and reduced autonomic activity in antisocial disorder with self-injurious behavior. J Neuropsychiatry Clin personality disorder. Arch Gen Psychiatry 2000; 57: 119–127. Neurosci 2007; 19: 383–390. 10 Raine A, Buchsbaum M, Lacasse L. Brain abnormalities in 36 Brennan PA, Mednick SA, Hodgins S. Major mental disorders and murderers indicated by positron emission tomography. Biol criminal violence in a Danish birth cohort. Arch Gen Psych 2000; Psychiatry 1997; 42: 495–508. 57: 494–500.

Molecular Psychiatry Altered connections on the road to psychopathy MC Craig et al 953 37 Highley JR, Walker MA, Esiri MM, Crow TJ, Harrison PJ. networks involved in autobiographical memory. J Neurosci 1996; Asymmetry of the uncinate fasciculus: a post-mortem study of 16: 4275–4282. normal subjects and patients with schizophrenia. Cereb Cortex 42 Christianson S-A˚ , Forth AE, Hare RD, Strachan C, Lidberg L, 2002; 12: 1218–1224. Thorell L-H. Remembering details of emotional events: a compar- 38 Kubicki M, Park H, Westin CF, Nestor PG, Mulkern RV, Maier SE et ison between psychopathic and nonpsychopathic offenders. al. DTI and MTR abnormalities in schizophrenia: analysis of white Personality and Individual Differences 1996; 20: 437–443. matter integrity. NeuroImage 2005; 26: 1109–1118. 43 Dolan M, Fullam R. Memory for emotional events in violent 39 Kubicki M, Westin C-F, Maier SE, Frumin M, Nestor PG, Salisbury offenders with antisocial personality disorder. Personality and DF et al. Uncinate fasciculus findings in schizophrenia: a Individual Differences 2005; 38: 1657–1667. magnetic resonance diffusion tensor imaging study. Am J 44 Pfefferbaum A, Sullivan EV, Hedehus M, Adalsteinsson E, Lim Psychiatry 2002; 159: 813–820. KO, Moseley M. In vivo detection and functional correlates of 40 Aron AR, Robbins TW, Poldrack RA. Inhibition and the right white matter microstructural disruption in chronic alcoholism. inferior frontal cortex. Trends Cognitive Sci 2004; 8: 170–177. Alcohol Clin Exp Res 2000; 24: 1214–1221. 41 Fink GR, Markowitsch HJ, Reinkemeier M, Bruckbauer T, Kessler J, 45 Beaulieu C. The basis of anisotropic water diffusion in the nervous Heiss W-D. Cerebral representation of one’s own past: neural system—a technical review. NMR Biomed 2002; 15: 435–455.

Molecular Psychiatry