The contribution of brain imaging to the understanding of psychopathy

Abstract Psychopathy is a personality type characterized by both callous emotional dysfunction and deviant behavior that affects society in the form of actions that harm others. Historically, researchers have been concerned with seeking data and arguments to support a neurobiological foundation of psychopathy. In the past few years, increasing research has begun to reveal brain alterations putatively underlying the enigmatic psychopathic personality. In this review, we describe the brain anatomical and functional features that characterize psychopathy from a synthesis of available neuroimaging research and discuss how such brain anomalies may account for psychopathic behavior. The results are consistent in showing anatomical alterations involving primarily a ventral system connecting the anterior temporal lobe to anterior and ventral frontal areas, and a dorsal system connecting the medial frontal lobe to the posterior cingulate cortex/precuneus complex and, in turn, to medial structures of the temporal lobe. Functional imaging data indicate that relevant emotional flow breakdown may occur in both these brain systems and suggest specific mechanisms via which emotion is anomalously integrated into cognition in psychopathic individuals during moral challenge. Directions for future research are delineated emphasizing, for instance, the relevance of further establishing the contribution of early life stress to a learned blockage of emotional self-exposure, and the potential role of androgenic hormones in the development of cortical anomalies.


Introduction
Psychopathy refers to a personality type expressed in the form of emotional callousness, lack of empathy, a grandiose estimation of self, impulsivity, and persistent antisocial behavior, among other traits (Hare, 2003). This profile notably overlaps with features defining the antisocial personality disorder (American Psychiatric Association, 2013). However, such terms are not synonymous. To meet criteria for psychopathy, an individual must exhibit emotional dysfunction (Blair, 2012). Emotional features such as callousness, lack of empathy and inflated selfappraisal are not necessary to operatively define antisocial personality disorder in the main Diagnostic and Statistical Manual of Mental Disorders (DSM)-V section, which predominantly considers antisocial deviance. However, the alternative trait-based definition section of the DSM-V includes an antisocial personality disorder variant characterized by low anxiety and a bold interpersonal style, which is closer to the traditional view of psychopathy (American Psychiatric Association, 2013;Venables et al., 2014). Analogously, the presence of callous-unemotional traits may be used to define a 'psychopathic' variant of conduct disorders in children (Patrick, 2014). A historical account of the use of these and related terms (e.g. sociopathy and dissocial disorder) can be found in Kiehl and Hoffman (2011).
The most widely used and validated instrument for the assessment of psychopathy is the interview-based Psychopathy Checklist-Revised (PCL-R; Hare, 2003), which contains distinct affective-interpersonal and impulsive-antisocial factors. Various self-report instruments also exist for assessing psychopathy, as well as instruments adapted for children and adolescents (Patrick, 2014). Some inventories are inspired by the PCL-R, while others represent new developments, for instance, the Psychopathic Personality Inventory (Lilienfeld and Andrews, 1996) and the Triarchic Psychopathy Measure (Patrick et al., 2009). Current brain imaging tools offer a unique window to explore the structural and functional bases of normal and deviant behavior. Historically, researchers have been concerned with seeking arguments to support a neurobiological foundation of psychopathy, as one of the most enigmatic personality disorders. In the last few years, increasingly abundant data have been provided. Nevertheless, further effort is necessary to integrate the new information. In this review, we describe the anatomical and functional features that characterize the brain of psychopathic individuals from a synthesis of neuroimaging research and discuss how such brain anomalies may account for psychopathic behavior. The available neuroimaging literature on psychopathy, defined in accordance with Hare's concept (Hare, 2003), is analyzed and the review is complemented by a secondary consideration of selected neuroimaging studies assessing young people showing antisocial behavior and callous emotional dysfunction.

Global brain assessment
The brain of primary psychopathic individuals (as opposed to patients with psychopathic-like behavior secondary to focal brain damage) does not show any gross anatomical anomalies upon visual inspection. The whole brain volume is similar to the general population in most studies that have provided the measurement (Dolan et al., 2002;Yang et al., 2005aYang et al., , 2005bNarayan et al., 2007;Tiihonen et al., 2008;Contreras et al., 2015). Significant brain volume reductions have only rarely been reported (Barkataki et al., 2006). When volumes have been measured for specific lobes using conventional anatomical limits, significant volume reduction was solely observed in the temporal lobe and only a minimal effect was appreciated in the prefrontal lobe (Dolan et al., 2002).
Within the samples of high socioeconomic risk, an association has been reported between the persistence of the cavum septum pelluciduma fluid-filled space between the two leaflets of the septumand higher scores of psychopathy . However, the cavum is similarly prevalent in disruptive behavior disorders in the absence of psychopathy Toivonen et al., 2013;White et al., 2013), suggesting that it is mainly a non-specific neuroradiological finding (Saba et al., 2013).

Changes in the content of gray matter
A large number of studies have focused on measuring the regional content of gray matter. These studies were based on trace delineation of gray/white matter boundaries, voxel-based morphometry and automated measurements of cortical thickness. To a large extent, there is general agreement regarding psychopaths' showing regional cortical reductions in relative gray matter content.
Earlier imaging studies predicted gray matter reductions in the prefrontal lobe of psychopaths (Raine et al., 2000;Yang et al., 2005aYang et al., , 2005b. Subsequent research has both largely confirmed such a prediction and demonstrated that tissue alteration extends to other brain locations (Laakso et al., 2001(Laakso et al., , 2002Narayan et al., 2007;de Oliveira-Souza et al., 2008;Müller et al., 2008;Tiihonen et al., 2008;Yang et al., 2009aYang et al., , 2009bYang et al., , 2010Boccardi et al., 2011Boccardi et al., , 2013Raine et al., 2011;Sato et al., 2011;Ermer et al., 2012;Gregory et al., 2012;Ly et al., 2012;Bertsch et al., 2013;Kolla et al., 2014;Kumari et al., 2014;Contreras-Rodríguez et al., 2015;Walters et al., 2015;Jiang et al., 2016), as summarized in Table 1. Although the number of reported brain areas indicates a tendency for changes to affect the cortex globally, gray matter volume reduction is particularly consistent across studies in (i) the rostral temporal lobe and the rostral and ventral frontal lobe including the ventral-medial and orbitofrontal cortex; (ii) the dorsal aspect of the brain medial wall involving a dorsal/medial frontal area extending to the anterior cingulate cortex and the posterior cingulate cortex/precuneus; (iii) the posterior parahippocampal gyri/medial visual cortex; and (iv) the sensorimotor cortex. Figure 1 provides an overall depiction of this consistent pattern of gray matter volume reduction in psychopaths.
The set of cortical regions shown in Fig. 1 represents a combination of neocortical and paralimbic areas. Although the involvement of other (limbic) elements has frequently been emphasized in psychopathy, only a few studies have actually detected significant volume reduction in structures such as the amygdala, hippocampus, and insula, as indicated in Table 1. Moreover, some analyses specifically targeting the individual limbic structures using manual tracing failed to find the predicted volume reduction in the amygdala (Barkataki et al., 2006;Boccardi et al., 2011), hippocampus (Raine et al., 2004;Barkataki et al., 2006;Boccardi et al., 2010), and anterior cingulate cortex (Glenn et al., 2010a). However, the limbic/paralimbic system as a whole (integrated using latent-variable modeling) does appear to be a major contributor to psychopathy .
In support of a developmental dynamic basis of brain tissue alterations in psychopathy, one study reported a significant association of childhood physical abuse with gray matter volume reduction in the temporal pole of adult psychopaths (Kolla et al., 2014). Child abuse was also associated with reduced frontotemporal cortical thickness predicting antisocial behavior in adolescents (Busso et al., 2017). In another study, the combination of relatively lower cortical volumes and higher volumes in subcortical regions was found to partially mediate the relation between adverse life events and antisocial behavior in a sample of 1741 adolescents (Mackey et al., 2017). At the age of 25 years, a voxel-based morphometry analysis demonstrated the association of early life poverty with gray matter volume reduction precisely in the ventral-medial frontal cortex (Holz et al., 2015). Symptoms related to early onset cumulative adversity have similarly been associated with reduced gray matter volumes in persistent delinquent youths (Raine et al., 2003).

Changes in the content of white matter
Despite consistency across studies, the extent to which reductions of gray matter volumes in adult psychopaths described above express true gray matter loss (i.e. atrophy) is not obvious. Indeed, results from white matter analyses suggest that white matter increases may contribute to relative reductions of gray matter measurements in the absence of relevant brain volume changes.
When results have been provided, white matter volume changes have generally paralleled the findings of gray matter changes in the opposite direction, with volume increases in global, prefrontal, and corpus callosum white matter correlating with psychopathy scores (Raine et al., 2003;Yang et al., 2005aYang et al., , 2005b. Larger volumes have also been demonstrated for basal ganglia elements with a very high content of white matter (Barkataki et al., 2006;Glenn et al., 2010b;Pujara et al., 2014).
A similar tendency has also been observed in disorders overlapping with psychopathy. In persistently violent offenders with antisocial personality disorder significantly larger white matter volumes were observed in broadly distributed brain regions grossly coinciding with areas of gray matter volume reduction (Tiihonen et al., 2008). In young males with psychopathic traits, a mixed pattern of changes has been reported combining both increases and decreases in regional white matter volume (De Brito et al., 2011).
It has been argued that the relative increase of white matter in psychopaths may be related to non-optimal brain remodeling (deficient axonal pruning), resulting in excessive anatomical connections (Raine et al., 2003). However, alternatively, we suggest here that larger white matter volumes in psychopathy may also relate to significantly accelerated myelination. It is important to note that white matter segments in conventional (T1-weighted) anatomical magnetic resonance imaging (MRI) scans correspond to myelinated white matter (Paus et al., 2001). Myelination is an active process throughout life (Yakovlev and LeCours, 1967;Pujol et al., 1993;Narayan et al., 2007), which is physiologically accelerated in early postnatal years and during adolescence (Paus et al., 2001;Pujol et al., 2006), and enhanced by repetitive use or skill learning (McKenzie et al., 2014;Pujol et al., 2016a). Pathological acceleration of myelination has been suggested, for example, in melancholic depression (Soriano-Mas et al., 2011), childhood obesity (Ou et al., 2015), heavy cannabis use (Matochik et al., 2005), and pathological lying (Yang et al., 2005a(Yang et al., , 2005b.
To summarize, measurable changes in gray and white matter regional tissue content can be detected in the brain of psychopaths in the absence of changes of total brain volume. This overall pattern of findings argues against gray matter atrophy as a sole explanation for gray matter volume reductions in psychopaths. We propose that accelerated myelination may be one contributing factor, in that it affects image tissue segmentation by displacing the boundary between white and gray matter.
Changes in the structure of white matter pathways Potential alterations in the structure of white matter pathways have been investigated using diffusion tensor imaging (DTI).
The common measurement across all studies is fractional anisotropy (FA), which quantifies the degree of structural maturation of white matter by estimating water diffusivity along tracts (Lebel et al., 2012). DTI studies of psychopathic populations coincide in showing FA reductions mostly involving frontal lobe connections (Craig et al., 2009;Motzkin et al., 2011;Sundram et al., 2012;Hoppenbrouwers et al., 2013;Sethi et al., 2015;Wolf et al., 2015;Jiang et al., 2017b). Alterations in the ventral connections between the frontal lobe and the temporal lobe would appear to be a solid finding, which mostly implicate the uncinate fasciculus (Craig et al., 2009;Motzkin et al., 2011;Sundram et al., 2012;Hoppenbrouwers et al., 2013;Wolf et al., 2015;Jiang et al., 2017b). Significant FA reductions have also been identified in dorsal pathways connecting the frontal lobe with the parietal lobe (Sundram et al., 2012;Sethi et al., 2015;Jiang et al., 2017b). Interestingly, the cingulum is altered both ventrally and dorsally, with FA reductions detected in the segment connecting the posterior cingulate cortex with the medial temporal lobe and the segment connecting the posterior cingulate cortex with the frontal lobe (Sethi et al., 2015). Frontal lobe structural connectivity alterations in psychopaths therefore seem to implicate dorsal and ventral bundles that connect brain areas that broadly demonstrate gray matter volume reductions (summarized in Fig. 1). However, frontal lobe connectivity changes are not limited to the anterior-posterior direction, as reduced FA has also been demonstrated in the genu of corpus callosum (transverse connections) (Sundram et al., 2012) and in frontal-basal ganglia connections (Sundram et al., 2012;Hoppenbrouwers et al., 2013).
Reductions in FA can be interpreted as expressing a maturation delay in late maturing bundles (Sethi et al., 2015). Nevertheless, accelerated, as opposed to delayed, maturation may better account for the majority of pathway alterations in psychopaths. Indeed, the increase in FA from normal childhood to adulthood reaches a peak between 20 and 40 years of age, and subsequently decreases (Lebel et al., 2012). Accordingly, from early adulthood, the normal physiological evolution of FA is one of the progressive reductions, presumably reflecting increased white matter structural complexity accompanying brain maturation (Douaud et al., 2011;Jones et al., 2013). In DTI studies of psychopathy, the mean age of adult samples has broadly coincided with the FA peak, or with the beginning of FA decreases (age range across studies, 30-40 years), with the sole exception of one study (Jiang et al., 2017b). Thus, within this age range, reduced FA is better explained by accelerated maturation, or psychopaths reaching FA maturity earlier than what is normally observed. Interestingly, in the DTI study featuring the youngest population (mean age 23 years), a combination of FA decreases (predominant) and increases was reported (Jiang et al., 2017b).
DTI analyses in younger populations with callous-unemotional traits have provided mixed results, with different studies reporting higher, lower or no differences in FA v. healthy controls for diverse white-matter tracts (see Waller et al. (2017) for a review). These findings offer further support for the proposal that abnormally accelerated white matter maturation contributes to the nature of anatomical changes in psychopathy and callousness. FA increase will be identified in youths, compared with typically developing controls, at ages prior to the FA age-related peak (the ascending part of the curve) for a given bundle, and FA reduction in older individuals closer to the age at which FA begins to decrease (Waller et al., 2017). Here too, early life stress could play a significant role. In one study, adults with life-time exposure to parental abuse showed FA reduction in the hippocampal extension of the cingulum bundle, the fornix, and the arcuate fasciculus (Choi et al., 2009).

Functional connectivity
Functional connectivity is considered a measurement of activity synchrony between brain regions sharing functional properties (Biswal et al., 1995). In antisocial offenders, the assessment of functional connectivity changes using whole-brain measures has provided evidence of both reduced functional integration and segregation in the organization of large-scale brain networks (Jiang et al., 2017a). In global terms, such findings may be interpreted as reflecting deficient brain functional maturation (Fair et al., 2009;Pujol et al., 2016b). The integration and segregation of large-scale brain networks is a life-time process that is especially active during adolescence (Sherman et al., 2014). Thus, the hypothetical acceleration of white matter maturation does not seem to imply a more efficient coupling among late maturing networks in the case of psychopaths (i.e. maturation may be faster but deficient).
As in the case of other imaging alterations, global brain changes in functional connectivity have been observed in association with changes that are particularly prominent in specific brain systems. Once again, the frontal cortex is the structure that shows the most relevant connectivity alterations in psychopaths. Some evidence indicates reduced functional connectivity between ventral frontal areas and the anterior temporal lobe and amygdala at rest (Motzkin et al., 2011) and during emotional stimulation (Decety et al., 2013a;Yoder et al., 2015;Volman et al., 2016). Other studies have demonstrated reduced functional connectivity of the dorsal frontal cortex (and anterior cingulate cortex) with several limbic-paralimbic structures (Ly et al., 2012;Pujol et al., 2012;Contreras et al., 2015;Philippi et al., 2015) and, particularly, with the posterior cingulate cortex/precuneus complex (Pujol et al., 2012;Contreras et al., 2015;Philippi et al., 2015).
In addition to long-distance connectivity reduction, psychopaths appear to display higher functional connectivity within the dorsal aspect of the frontal lobes at rest (Contreras et al., 2015), and between dorsal frontal areas and the striatum during reward expectancy (Geurts et al., 2016). Higher frontal cortex connectivity was associated with more severe lifestyle/antisocial traits (Hare factor 2) in the study carried out by Philippi et al. (2015). In general agreement with such functional connectivity findings, studies of structural connectivity (structural covariance) estimated from cortical thickness measurements, also implicates the superior frontal cortex as a relevant connectivity hub in psychopaths (Yang et al., 2012). Reduced functional connectivity also seems to combine with increased connectivity in antisocial personality disorder (Tang et al., 2016).
Brain response to aversively conditioned stimuli (e.g. with pain) has also been tested in psychopaths. Once again, the results have been divergent, with evidence for both attenuated Larson et al., 2013;Decety et al., 2014) and enhanced (Schneider et al., 2000;Schultz et al., 2016) brain activations, which further suggests that the experimental context is critical in determining the magnitude and direction of the response. Attention seems to be a relevant factor that may explain some apparent discrepancies. For example, compared with control subjects, psychopaths showed lower right amygdala and higher left frontal response to a short (200 ms) exposure to a painconditioned stimulus when attention was engaged in a neutral task (Larson et al., 2013). By contrast, long exposure (8 s) to a pain-conditioned complex image generated a greater response, in this case in the left amygdala and in areas processing visual stimuli features (Schultz et al., 2016).
There is also evidence relating psychopathy to altered responses to incentive stimulation. Specifically, psychopathy has been associated with lower basal ganglia activation related to monetary loss (Pujara et al., 2014), lower medial frontal activation during reward anticipation (Veroude et al., 2016), and lower anterior cingulate cortex activation in response to reward uncertainty , while increased activation has been observed in the posterior cingulate cortex and anterior insula during the task reversal phase . Also, higher psychopathy scores have been associated with reduced response to drug abuse-related pictures in the basal ganglia and connected brain structures (Cope et al., 2014a(Cope et al., , 2014b.

Brain activity during moral challenge
Neuroimaging research has demonstrated the involvement of a well-defined brain network in the mediation of moral judgment in the normal population. This network overlaps with the so-called default mode network connecting the medial frontal cortex, the posterior cingulate cortex, and the angular gyri (Greene et al., 2001), but also medial structures of the temporal lobe and upper brainstem (Harrison et al., 2008;Pujol et al., 2012). In functional MRI studies, psychopaths required to make a judgment in a moral dilemma situation (Fig. 2) have consistently shown deficient activation in the medial frontal and posterior cingulate cortex (Glenn et al., 2009;Harenski et al., 2010;Veit et al., 2010;Pujol et al., 2012;Yoder et al., 2015) and, remarkably, in the hippocampus and parahippocampal gyrus extending to the periaqueductal gray (Pujol et al., 2012) and amygdala (Glenn et al., 2009;Pujol et al., 2012;Marsh and Cardinale, 2014;Yoder et al., 2015). Such experiments may therefore show the inappropriate use of a network that both mediates attention to self as an agent and has access to autobiographical storage (Moll et al., 2007;Bado et al., 2014;Leech and Sharp, 2014). Consistent with this view, criminal psychopaths precisely showed lower activity in the posterior (and anterior) cingulate cortex, parahippocampal gyrus, hippocampus, and amygdala during the performance of an affective memory task (Kiehl et al., 2001).
A functional 'breakdown' in the brain network subserving moral judgment may also extend to neutral situations unrelated to the moral context. Functional MRI signal reduction or deactivation typically observed in the default mode network during conventional cognitive tasks is deficient in psychopaths (Pujol et al., 2012;Juárez et al., 2013;Freeman et al., 2015). Moreover, the anterior and posterior elements of the default mode network appeared to be significantly uncoupled during the resting state in functional connectivity analyses (Pujol et al., 2012). Therefore, the network abnormally responding during moral challenge would also seem to be primarily altered. This idea is consistent with the proposal that attentional deficits in psychopaths are not limited to situations of moral conflict, but extend more generally to other attentional domains (Newman et al., 2010;Aharoni et al., 2013;Newman and Baskin-Sommers, 2016;Rodman et al., 2016).
The arousal generated by a moral challenge in psychopaths appears to depend on how the self is taken as a reference. While watching morally laden scenes, psychopaths showed higher frontal activation when identifying the emotional state of the victim and lower frontal activation when identifying the emotional state of the predator . However, other data suggest that if the psychopath is forced to feel like the predatoras opposed to simply having to identify the predator's emotional statesuch a response attenuation may be less evident (Sommer et al., 2010;Decety et al., 2013aDecety et al., , 2013bMeffert et al., 2013), further suggesting an attentional selection bias (Newman et al., 2010;Newman and Baskin-Sommers, 2016).
In brief, while psychopaths are reactive to different forms of emotional stimulation, emotional processing may be incomplete with poor temporal lobe transmission of emotional flow to the ventral frontal system. Also, moral challenge experiments have revealed the inappropriate use of a dorsal network mediating attention to our inner emotional world.
In addition to functional imaging of evoked brain activation, other studies have assessed brain function based on a variety of metabolic parameters. Regional glucose metabolism and cerebral blood flow have generally been reported to be lower in psychopaths, particularly in frontal and temporal areas (Volkow et al., 1995;Raine et al., 1997;Soderstrom et al., 2002). One study reported lower levels of monoamine oxidase-A, an enzyme that regulates neurotransmitters, in the orbitofrontal cortex and ventral striatum in offenders with high-psychopathic traits (Kolla et al., 2015). Other authors have reported a positive correlation
between striatal serotonin 1B receptor binding and the level of psychopathy (da Cunha-Bang et al., 2017).

General implications
In this review, we provide a synthesis of available neuroimaging research directly concerning to psychopathy. A variety of imaging methods has been used to date to provide distinct perspectives of brain structure and function. There are notable differences across the findings of existing studies, which is likely to reflect important differences in study methodology, but may also reflect the underlying biological heterogeneity of psychopathy. Nevertheless, most studies coincide in suggesting that the brain of psychopaths differs notably from the typical brain in terms of both anatomy and function. There is a tendency for changes to affect the brain globally, but variations from the normal pattern are particularly evident in a number of functionally related brain structures. One system includes the rostral aspect of both frontal and temporal lobes, which subcortically report to the ventral striatum and anterior hippocampus/amygdala, respectively. Although this system is involved in a vast variety of processes, in the context of psychopathy, we emphasize its role in conveying emotional input to motivated actions (Simpson and Balsam, 2016). The second system combines the medial frontal (and anterior cingulate) cortex with the posterior cingulate cortex/precuneus, core elements of the default mode network, which are critical in focusing attention on self and accessing autobiographical storage (Bado et al., 2014;Leech and Sharp, 2014). This dorsal system is connected to caudal temporal lobe structures also altered in psychopathy. Relevant changes have been identified in the parahippocampus/posterior hippocampus and visual areas involved in the processing of sensory stimulation in general, but also relevant for the storage of significant autobiographical events (Jeong et al., 2015). In addition, alterations have also been reported involving the sensorimotor cortex, which is the primary representation of our bodies. Functional data suggest that emotionally evocative stimuli are indeed capable of activating the brain and even generating excessive responses in some (mostly dorsal) emotion-relevant brain areas in psychopaths. However, the emotional processing flow may appear to be disrupted in the transition from the temporal lobe to ventral frontal areas, with the consequent failure to integrate emotion into cognition and subsequent decisions, as often proposed (Kiehl et al., 2001;Raine and Yang, 2006;Blair, 2007;Moll et al., 2008;Contreras et al., 2014;Blair et al., 2018). The alteration in emotional processing resulting from such ventral temporal-frontal blockage may be sufficient to account for psychopathic behavior if taken to the extreme. Indeed, in frontotemporal dementia and after focal damage of rostral frontotemporal areas, the presence of psychopathic-like behavior is not exceptional (Anderson et al., 1999;Brower and Price, 2001;Koenigs et al., 2007;Moll et al., 2007;Diehl-Schmid et al., 2013;Birkhoff et al., 2016;Darby et al., 2018).
Nevertheless, imaging data indicate that emotional processing blockage may also occur as a result of the abnormal functioning of a dorsal brain system. Probably the most consistent finding in functional imaging research in the context of adult psychopathy is the alteration of the network involved in the mediation of moral judgment overlapping with the default mode network. The normal activation of this network serves to focus attention on the self as an agent and thus (as we propose) exposing the individual to be aroused (or 'touched') from emotional (positive or negative) memories stored in temporal lobe medial structures. Activation of the default mode network during moral conflict is certainly abnormal in psychopaths (Glenn et al., 2009;Harenski et al., 2010;Veit et al., 2010;Pujol et al., 2012;Yoder et al., 2015), which concurs with a similarly poor activation of key memory structures such as the parahippocampal gyrus, hippocampus, and amygdala (Glenn et al., 2009;Pujol et al., 2012;Marsh and Cardinale, 2014;Yoder et al., 2015).
All in all, a failure to integrate emotion into cognition would appear to result from at least two complementary breakdowns; bottom-up disruption of the emotional flow in the ventral frontotemporal system and a top-down blockage of emotional selfexposure in the dorsal network mediating moral judgment.

Future directions
In the current review, we have focused on how current neuroimaging research may contribute to a better understanding of pathophysiological mechanisms leading to psychopathic behavior. Nevertheless, another question is whether neuroimaging can provide clues as to its etiology? We would emphasize three potential causal factors deserving future research, in addition to genetic predisposition, which always plays a significant role in complex behavioral disorders.
As discussed earlier, one of the factors that potentially contributes to the development of brain pathology in psychopaths is life-time stress, which may well accelerate brain maturation. Life-time stress could indeed contribute to regional gray matter volume reduction through excessive neural activity leading to subsequent atrophy (a 'burnout' effect) or by increasing myelinated white matter with an apparent reduction of the gray matter tissue segment, or both. Future analyses may specifically focus on determining the life-time evolution of cortical white matter content. Additionally, it may be valuable in this context to test the hypothesis of accelerated brain maturation from a functional perspective, by assessing both brain activity and functional network connectivity in children and adolescents with psychopathic traits.
Androgens have a relevant effect on brain shaping. Critical periods include the prenatal androgenization of the brain and androgen activation during adolescence (Arnold and Gorski, 1984;Sato et al., 2008). The male and female brain differs, in part, due to androgen effects (Heany et al., 2016) and antisocial behavior is substantially more frequent in males (Yildirim and Derksen, 2012). Thus, anomalies in developmental androgenic brain modeling may increase the predisposition to psychopathy (Yildirim and Derksen, 2012). Relevantly, the effects of prolonged high-dose administration of androgenic steroids on brain anatomy are largely reminiscent of gray matter alteration patterns in psychopaths showing reduced cortical thickness in rostral frontotemporal areas, the medial frontal cortex, the posterior cingulate cortex/precuneus, and visual areas (Fig. 3) (Bjørnebekk et al., 2017).
Finally, the results from some analyses suggest that the described brain alterations could show distinct and opposite patterns of association with both affective disturbances and antisocial behavior. Indeed, some brain alterations are reported to be associated with more affective disturbances (Hare factor 1) and less antisocial behavior (Hare factor 2), including reductions in gray matter volume (Contreras et al., 2015), altered functional connectivity in large-scale cortical networks (Philippi et al., 2015), and amygdala reactivity (Hyde et al., 2014). A recent study further indicates that the severity of antisocial behavior is associated with larger prefrontal and striatal subregion volumes and higher functional connectivity between the several areas of the prefrontal cortex (Korponay et al., 2017a(Korponay et al., , 2017b. Future research on this issue is therefore of great interest to ascertain whether the described alterations in brain structure and function in psychopaths necessarily predispose them to crime. Interestingly, there are behavioral  Bjørnebekk et al. (2017). Note the notable regional coincidence with the areas showing the most consistent volume reduction in psychopaths illustrated in Fig. 1. The largest coincidences are present in the ventral temporal-frontal system (1) and the dorsal-medial system (2), but cortical reductions are also present in visual (3) and motor cortices (4).

26
Jesus Pujol et al. data indicating that antisocial behavior, as measured in Hare factor 2, is better explained by poorer cognitive functioning (Baskin-Sommers et al., 2015).

Conclusions
This review has sought to provide a consolidated perspective on the contribution of brain imaging to understanding of psychopathy. The brain of psychopaths differs from the typical brain in terms of both anatomy and function. Anatomical alterations involve primarily a ventral system connecting the anterior temporal lobe to anterior and ventral frontal areas, and a dorsal system connecting the medial frontal lobe to the posterior cingulate cortex/precuneus complex and, in turn, to medial structures of the temporal lobe. Functional imaging data indicate that a significant disruption or 'breakdown' in the flow of emotional information processing may occur in both these two brain systems and suggest specific mechanisms via which emotion is anomalously integrated into cognition in psychopathic individuals during moral challenge. To this end, broader reviews are of interest comparing the results from different fields including, for instance, EEG, genetics, neurochemistry, and neuropsychology.