Progress in Neurobiology 122 (2014) 1–23
Contents lists available at ScienceDirect
Progress in Neurobiology
jo urnal homepage: www.elsevier.com/locate/pneurobio
Brain alterations in paedophilia: A critical review
a, a b
Sebastian Mohnke *, Sabine Mu¨ ller , Till Amelung ,
c d e
Tillmann H.C. Kru¨ ger , Jorge Ponseti , Boris Schiffer ,
f b a,
Martin Walter , Klaus M. Beier , Henrik Walter **
a
Charite´ – Universita¨tsmedizin Berlin, Department of Psychiatry and Psychotherapy, Charite´platz 1, 10117 Berlin, Germany
b
Charite´ – Universita¨tsmedizin Berlin, Institute of Sexology and Sexual Medicine, Luisenstr. 57, 10117 Berlin, Germany
c
Hannover Medical School, Department of Clinical Psychiatry, Social Psychiatry, and Psychotherapy, Carl-Neuberg-Str. 1, 30625 Hannover, Germany
d
University Medical Center Schleswig-Holstein, Section for Sexual Medicine, Arnold-Heller-Str. 3, 24105 Kiel, Germany
e
Division of Forensic Psychiatry, Department of Psychiatry, Psychotherapy and Preventive Medicine, LWL-University Hospital, Bochum, Germany
f
Otto-von-Guericke-University Magdeburg, Department of Psychiatry, Leipziger Str. 44, 39120 Magdeburg, Germany
A R T I C L E I N F O A B S T R A C T
Article history: Psychosocial and biological factors have been implicated in paedophilia, such as alterations in brain
Received 5 December 2013
structure and function. The purpose of this paper is to review the expanding body of literature on this
Received in revised form 25 July 2014
topic including brain abnormality case reports, as well as structural and functional neuroimaging
Accepted 31 July 2014
studies. Case studies of men who have committed sexual offences against children implicate frontal
Available online 10 August 2014
and temporal abnormalities that may be associated with impaired impulse inhibition. Structural
neuroimaging investigations show volume reductions in paedophilic men. Although the findings have
Keywords:
been heterogeneous, smaller amygdala volume has been replicated repeatedly. Functional neuroimaging
Child sexual abuse
investigations demonstrate an overlap between paedophiles and teleiophiles during sexual arousal
Sexual arousal
processing. While it is controversial among studies regarding group differences, reliable discrimination
Sexual offending
Brain pathology between paedophilic and teleiophilic men may be achieved using functional activation patterns.
Magnetic resonance imaging Nevertheless, the heterogeneous findings published so far suggest further research is necessary to
disentangle the neurobiological mechanisms of paedophilic preference. A number of methodological
confounds have been identified, which may account for the inconsistent results that could prove to be
beneficial for future investigations.
ß 2014 Elsevier Ltd. All rights reserved.
Contents
1. Introduction ...... 2
1.1. Diagnosis ...... 2
1.2. Prevalence ...... 2
1.3. Paedophilia vs. sexual offending against children ...... 3
1.4. Aetiology ...... 4
1.5. The neurophenomenological model of sexual arousal ...... 4
1.6. Literature search ...... 4
Abbreviations: ICD-10, International Classification of Diseases and Related Health Problems, 10th Revision, World Health Organization; DSM-5, Diagnostic and Statistical
Manual of Mental Disorders, 5th Edition, American Psychiatric Association; CSA, child sexual abuse; CSO, child sexual offender; NSO, non-sexual offender; MRI, magnetic
resonance imaging; fMRI, functional magnetic resonance imaging; PET, positron emission tomography; CT, computed tomography; VBM, voxel-based morphometry; ROI,
region of interest; FWE, familywise error; FDR, false discovery rate.
* Corresponding author at: Charite´ – Universita¨tsmedizin Berlin, Campus Mitte, Department of Psychiatry and Psychotherapy, Division of Mind and Brain Research,
Charite´platz 1, 10117 Berlin, Germany. Tel.: +49 30 450 517 223; fax: +49 30 450 517 906.
** Corresponding author at: Charite´ – Universita¨tsmedizin Berlin, Campus Mitte, Department of Psychiatry and Psychotherapy, Division of Mind and Brain Research,
Charite´platz 1, 10117 Berlin, Germany. Tel.: +49 30 450 517 141; fax: +49 30 450 517 906.
E-mail addresses: [email protected] (S. Mohnke), [email protected] (S. Mu¨ ller), [email protected] (T. Amelung),
[email protected] (Tillmann H.C. Kru¨ ger), [email protected] (J. Ponseti), [email protected] (B. Schiffer), [email protected]
(M. Walter), [email protected] (K.M. Beier), [email protected] (H. Walter).
http://dx.doi.org/10.1016/j.pneurobio.2014.07.005
0301-0082/ß 2014 Elsevier Ltd. All rights reserved.
2 S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23
2. Neurological case reports ...... 5
2.1. Brain tumours ...... 5
2.2. Different forms of dementias ...... 7
2.3. Parkinson’s disease ...... 7
2.4. Further neurological disorders ...... 8
3. Neuroimaging ...... 8
3.1. Structural neuroimaging...... 8
3.1.1. CT studies...... 8
3.1.2. MRI studies ...... 8
3.2. Functional neuroimaging ...... 12
3.2.1. PET studies...... 12
3.2.2. fMRI studies...... 12
3.3. Correlational analyses in neuroimaging studies ...... 18
4. Discussion ...... 19
Acknowledgements ...... 22
References ...... 22
1. Introduction et al., 2013), implying that the sexual interest in prepubertal
children would be distinguishable from the sexual interest in
1.1. Diagnosis pubertal children. Contrary to early propositions, and after an
extensive debate, the diagnosis of hebephilia was not included in
Paedophilia is commonly viewed as an adult sexual responsive- the DSM-5 (for a summary of this discussion see Blanchard, 2013).
ness towards immature (prepubescent) subjects. Research distin- In addition, the ICD-10 requires the presence of recurrent sexual
guishes it from teleiophilia as the sexual responsiveness to mature fantasies or urges towards children, and these elements are not
(postpubertal) subjects and hebephilia, the sexual responsiveness necessary according to the DSM-5. Following DSM-criteria sexual
to pubertal children (Beier et al., 2013; Blanchard et al., 2000). behaviour with prepubescent children that persists more than six
Paedophilia appears in both psychiatric diagnostic systems, the months would be sufficient for the diagnosis of paedophilic
ICD-10 and the DSM-5, under disorders of sexual preference disorder, even in the absence of arousing fantasies or urges.
(paraphilia). Since the DSM-5, paedophilia is explicitly distin- The DSM-5 distinguishes between an exclusive and a non-
guished from paedophilic disorder, defined via recurrent, intense exclusive type of paedophilia (whether the person can be sexually
sexually arousing fantasies, and sexual urges towards prepubes- aroused only by children or also by elder persons), a gender
cent children. The latter additionally requires either that these preference, and a limitation to incest. Also, the criteria within each
sexual urges lead to marked distress, interpersonal difficulty, or diagnostic system have been changed significantly during its
that the individual has acted out on these sexual urges. Therefore, development (for DSM criteria, see Blanchard, 2010). Therefore,
by the redefinition of the term paedophilia (i.e., sexual focus on the different definitions of paedophilia comprise different popula-
children without distress, interpersonal difficulty, and sexual acts tions that have a common intersecting-set but are not identical
involving children), the mere sexual preference for children has (see Fig. 1).
been depathologized. This makes sense from both a clinical and a The stability of paedophilia is also controversial. In general,
forensic perspective, since paedophilia which causes neither paedophilia has been seen as a lifelong individual trait (e.g., Seto,
suffering nor child harm requires no therapy or prosecution 2008, 2012). Cases in which paedophilic behaviours occurred after
(although preventive measures could be indicated, because the brain lesions have thus been discussed as behavioural manifesta-
preference may still constitute a risk factor for later offending; see tions of pre-existent latent paedophilic urges due to general
Section 1.3). impulse disinhibition. In contrast, there are some reports claiming
Both ICD-10 and DSM-5 require that the affected person’s successful therapeutic alteration of sexual preference and reduced
sexual urges/fantasies have been acted upon or cause marked reoffending in paedophilic child sexual offenders (CSO; Marshall
distress or interpersonal difficulty. Furthermore, both systems et al., 2005; Marshall, 2008). Additionally, a change in psycho-
specify that the affected person must be 16 years and at least five physiological reaction to child sexual stimuli has been recently
years older than the sexually desired children. In contrast to the demonstrated in some men (Mu¨ ller et al., 2014). Whether the
ICD-10, the DSM-5 further specifies that individuals in late changes reported represent an ‘‘elimination’’ of paedophilia or
adolescence who are in ongoing sexual relationships with 12– rather a control of symptoms remains unclear. While the findings
13-year-olds do not fit the criteria for paedophilia. Further remain anecdotal, they might still challenge the state of the art
disagreement between the diagnostic systems exists with regard view of paedophilia as a stable and unchangeable condition.
to the (body) age of the desired children: ICD-10 defines Neurobiological correlates may be particularly useful to disentan-
paedophilia (F65.4) as a sexual preference for children who are gle whether paedophilia is mutable by: first, elucidating the
either prepubescent or in an early state of puberty, but might also aetiology of human sexual preference (see Section 1.4); and
be younger; in contrast, the DSM-5 defines paedophilia or second, uncovering the link between preference and behaviour
paedophilic disorder (302.2) by recurrent, intense sexually leading to the risk of potentially endangering others as suggested
arousing fantasies, sexual urges, or behaviours involving sexual by the hypothesis that paraphilic interest can become behaviou-
activity with prepubescent children (generally age 13 years or rally manifest following brain damage (see also Section 1.3).
younger), which have to persist over a period of at least six months.
Note that the sexual preference for pubertal children has been 1.2. Prevalence
termed hebephilia and was repeatedly described as a distinguish-
able phenomenon (Beier et al., 2013; Blanchard, 2010). Further- Due to the lack of epidemiological studies, the exact prevalence
more, a recent study using multiple taxometric analyses supports of paedophilia is unknown. However, in the Berlin Male Study,
the view of paedophilia as a separate diagnostic entity (Schmidt conducted on a population-based sample of men between 40 and
S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23 3
Fig. 1. The diagnosis of paedophilia/paedophilic disorder according to the ICD-10 and the DSM-5.
79 years, 367 participants volunteered to anonymously fill out they cannot be used synonymously. Not all CSOs have
questionnaires on paraphilic arousal patterns. Of these men 9.5% paedophilia, just as not all paedophilic men necessarily commit
reported sexual fantasies involving children, 6% masturbated to child sexual abuse (CSA). Therefore, CSOs can be grouped in
these fantasies, and 3.8% admitted sexual contact with children those: (1) without a sexual preference for children that have
below 13 years of age (Ahlers et al., 2009). Of note, only two sexually abused children (e.g., sexually inexperienced adoles-
participants (0.5%) reported to be distressed by their excitability by cents seeking a surrogate; persons with antisocial personality
children. A population-based Finnish study on 3967 male twins disorders or perpetrators within general traumatizing family
between the ages of 21 and 43 years found a 12-month-prevalence constellations; Seto, 2008), and, (2) with a sexual preference for
of 0.3% (95% confidence interval: [0.08–0.61%]) for sexual interest children (e.g., paedophilia and/or hebephilia). The proportion of
in children (judging from respective sexual fantasies or actions; paedophiles in CSOs is about 40–50% (Maletzky and Steinhauser,
Mokros et al., 2012; Alanko et al., 2013). Furthermore, 2.7% of the 2002; Seto and Lalumiere, 2001; Seto, 2008). Conversely, the
participants reported masturbation phantasies involving children proportion of paedophiles who sexually approach children
below the age of 16, and 0.3% had sexual contact with children of seems to be similarly high with about 43% (Seto et al., 2006).
that age group during the past 12 months. Taking together, the In that respect, it has to be kept in mind that a paedophilic
point prevalence for paedophilia could be estimated to be around inclination in an individual does not mean that the inclined
0.3–3.8% (Ahlers et al., 2009; Alanko et al., 2013). There are isolated person will necessarily act on his fantasies. On the other hand,
reports of paedophilia in women (Chow and Choy, 2002), although paedophilia is a major risk factor for committing sexual offences
it seems to be substantially more frequent in men (Denov, 2003). against children, particularly for sexual recidivism (Hanson and
Morton-Bourgon, 2005). The distinction between paedophilia
1.3. Paedophilia vs. sexual offending against children and CSA is important to consider, since most research has been
conducted on CSOs that have not always been reliably
The term paedophilia is often used interchangeably with sexual diagnosed. Results from studies on these populations may be
offending against children, although research clearly shows that additionally confounded by antisocial traits of the subjects and
4 S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23
the forensic settings. Very little is known about undiagnosed (Cantor et al., 2004; Joyal et al., 2014; Kru¨ ger and Schiffer,
paedophiles who have not sexually abused a child. 2011; Schiffer and Vonlaufen, 2011). Additionally, Blanchard et al.
(2002, 2003) detected an increased rate of head injuries with
1.4. Aetiology temporary unconsciousness in paedophilic sex offenders. Impor-
tantly, only the incidence of head injuries before the age of 13 was
The aetiology of paedophilia is uncertain, but seems to be elevated in paedophiles, which points to possibly damaging effects
multifactorial, comprising psychosocial and biological factors. on neuronal development.
The conditioning hypothesis points to the relevance of first sexual Recent research began to shed light on the manifestations of the
contact with same-aged peers during childhood (Marshall and presumed neurodevelopmental disruptions by exploring abnor-
Eccles, 1993). During these experiences, an association between malities in brain function and structure. The aim of this paper is to
formerly neutral stimuli like a prepubescent body schema and review the neuroimaging literature on paedophilic men to provide
unconditioned stimuli such as sexual gratification would be an overview of published findings concerning abnormalities in
learned and result in the persistent sexual preference for brain anatomy and functional processing during visual sexual
respective sexual partners. This theory is contrary to the fact that stimulation. Historically, the first clues on brain structures
most people have their first sexual experiences during early implicated in paedophilic sexual arousal stem from several
puberty, but only a small percentage develops paedophilically neurological case reports describing paraphilia-like behaviour
motivated interest. Furthermore, some paedophilic men report that occurred with brain pathology. Therefore, we will start by
that they were aware of their sexual interest in children before summarizing the findings from these case studies.
their first sexual contacts (Dandescu and Wolfe, 2003).
Experience of having been sexually abused: Several reports show 1.5. The neurophenomenological model of sexual arousal
concordantly that sexual offenders against children who have been
abused in childhood themselves, are more likely to develop A framework helpful to interpret the reviewed findings is the
paedophilic interest (e.g., Nunes et al., 2013). However, the neurophenomenological model of sexual arousal (Fig. 2; Redoute´
evidence for this relationship in non-offender groups is substan- et al., 2000; Stole´ru et al., 1999). This model attempts to explain the
tially weaker (Fromuth et al., 1991) and, only a small proportion of specific roles of different brain regions typically activated in
sexually abused children develop paedophilia. There is evidence healthy men during stimulation with sexually arousing stimuli
that this association might be moderated by further environmental (Ku¨ hn and Gallinat, 2011; Poeppl et al., 2014; Stole´ru et al., 2012).
factors (for instance, experiences of neglect in childhood, lack of It assigns sexual stimulation to intertwined psychological sub-
parental supervision, intrafamilial violence, poor parent-child processes: a cognitive component comprises the evaluation of
attachment; Marshall and Marshall, 2000; Salter et al., 2003), and stimuli as sexually incentive (mediated by activity of the right
certain characteristics of the experienced abuse (duration, timing, lateral orbitofrontal cortex and inferior temporal gyrus), the focus
use of violence, penetration, relationship to the perpetrator, having of attention on these stimuli (superior and inferior parietal lobule),
perpetrators of both sexes; Burton et al., 2002). The exact and the motor imagery of potential sexual behaviour (ventral
mechanism by which history of being sexually abused increases premotor area, supplementary motor area, inferior parietal lobule,
the likelihood of developing paedophilia is unknown. Theoretical- and cerebellum). The experience of the hedonic quality of sexual
ly, learning mechanisms (e.g., imitation) as well as the facilitation arousal is provided by an emotional component (somatosensory
of attitudes and beliefs supporting adult-child-sex are assumed cortex, amygdala, and posterior insula), while a motivational
(Seto, 2008). An alternative explanation for the facts that not all component would orientate behaviour towards the desired goal
sexually abused children and not all children who had sexual and enable the perception of sexual desire (anterior cingulate
contact with other children develop paedophilia is an interaction cortex, claustrum, posterior parietal cortex, hypothalamus, sub-
with biological factors, such as genetic predisposition. stantia nigra, and ventral striatum). A physiological component
Genetics: First clues regarding potential genetic influences on would control various somatic responses in order to prepare the
paedophilia came from a study by Gaffney et al. (1984). organism for sexual behaviour (anterior cingulate cortex, anterior
Retrospectively reviewing medical records they found that insula, putamen, and hypothalamus). Last, the model proposes
paedophilia was significantly more prevalent in first-degree different inhibitory processes, which include the prevention of
relatives of patients fulfilling DSM-III criteria for paedophilia than emergence of sexual arousal in unsuitable situations and cognitive
in families of patients with non-paedophilic paraphilias or appraisal to reduce the incentiveness of stimuli (medial and left
depression. While this investigation was limited by its small lateral orbitofrontal cortex, and lateral temporal cortex) as well as
sample size (n = 33 paedophilic patients, n = 21 non-paedophilic the control of executed sexual behaviour (caudate and caudal
paraphilic patients, n = 33 depressed patients) and did not rule out anterior cingulate cortex). When discussing the neural correlates
alternative explanations for the familial accumulation, a larger of paedophilia we will refer to this model, which might give further
recent study by Alanko et al. (2013) gives further support for the insights into the meaning of the observed abnormalities.
assumption of genetic underpinnings. In the population-based
Finnish sample, the amount of non-additive genetic variance that 1.6. Literature search
could be accounted for sexual interest in children and youth below
the age of 16 was 14.6%. However, this percentage also emphasizes An extensive literature search was conducted using the
the importance of environmental factors. electronic databases PubMed, ISI Web of Science, and PsycInfo
Neurodevelopmental perturbations: Further research on the in order to identify all relevant articles published until April 2014.
biological aetiology of paedophilia mostly builds on the assump- We used the following search terms in the title, abstract or
tion, which is based on a multitude of indirect evidence, that keywords: (‘‘pedophil*’’ or ‘‘child molest*’’ or ‘‘sexual abuse’’ or
paedophilia could result from neurodevelopmental abnormalities. ‘‘sexual offend*’’) and (‘‘imaging’’ or ‘‘magnetic resonance imaging’’
There are reports of neuropsychological deficits including lower or ‘‘computed tomography’’ or ‘‘positron emission tomography’’ or
intelligence, impaired response inhibition, dampened attention, ‘‘brain’’). Additional papers were searched by checking the
reduced verbal and visuospatial learning abilities, slower proces- literature cited in the articles identified by the electronic database
sing speed, diminished task switching, and cognitive reasoning search. Studies were included if they provided original data on (a)
abilities as well as an increased rate of left-handedness brain damage associated with paedophilia or sexual offending
S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23 5
Fig. 2. The neurophenomenological model of sexual arousal.
Adapted from Stole´ru et al. (2012).
against prepubescent children, (b) structural neuroimaging in specified in one case (Regestein and Reich, 1978). Another patient
subjects diagnosed with paedophilia or who sexually offended suffered from a brainstem glioma additionally affecting the pons,
against prepubescent children or (c) functional neuroimaging ventral midbrain, thalamus, and hypothalamus, which affected
studies on sexual arousal processing in paedophiles or CSOs. Both both hemispheres, but was more pronounced in the left
case reports and studies comparing samples of paedophilic hemisphere (Miller et al., 1986). In addition to the abuse or
subjects or CSOs against controls were included. molestation of children, an increase of general sexual interest, the
preoccupation with sex or a disinhibition of sexual behaviour was
2. Neurological case reports common to all cases. Therefore, none of the abnormalities were
limited to CSA, but instead suggested general hypersexuality.
Before controlled imaging studies were available, the first Furthermore, the abnormalities were not isolated to symptoms on
evidence implicating brain structures in paraphilia-like behaviours the sexual level; patients exhibited personality changes and
was provided by neurological case reports. These reports show that impulsivity (Burns and Swerdlow, 2003; Lesniak et al., 1972;
paedophilic behaviour can be provoked by a variety of diseases Regestein and Reich, 1978) or diminished financial judgement
such as brain tumours, different forms of dementias, Parkinson’s (Miller et al., 1986). Three of these patients were tested
disease, Huntington’s disease, hippocampal sclerosis as well as neuropsychologically (the one not tested was described to have
encephalitis, and may also result from surgical treatment, e.g., normal intelligence; Miller et al., 1986) and demonstrated deficits
temporal lobotomy or pallidotomy (see Table 1). In the following including reduced intelligence (Lesniak et al., 1972), reduced
section we will review the literature pertaining to these cases cognitive flexibility and attention (Regestein and Reich, 1978), and
separately for the different diagnoses. We only consider reports disturbed memory recall, apraxia and agraphia (Burns and
about sexual behaviour or interest in prepubescent children. There Swerdlow, 2003). In light of this additional symptomatology it
are some papers in which sexual interest in pubertal children was appears unlikely that the tumours caused a specific change in
termed paedophile, which will not be discussed here. sexual preference. The variety of symptoms reported rather speak
in favour of the assumption that the neoplasms led to unspecific
2.1. Brain tumours disinhibition, which also affected sexual experience and behaviour,
thereby increasing the risk for sexual abusive behaviour. It is
There are reports of CSA emergent from brain neoplasms in four conceivable that such a disinhibition could result in acting out
married men (Table 1). In three cases these were located in the premorbid paedophile interest (Burns and Swerdlow, 2003;
right frontal lobe: medial paracentral (Lesniak et al., 1972), medial Mendez and Shapira, 2011). An elevated sexual drive alone could
orbitofrontal (Burns and Swerdlow, 2003), and not further potentially lead to behaviours involving minors of age even in the
6 S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23
Table 1
Summary of case reports of paedophilic behaviour occurring with neurological disorders.
Refs. Age Diagnosis/Localization Paedophilic symptoms Further symptoms Neuropsychological findings
Brain tumours
Lesniak et al. (1972) 56 Right-sided medial- Sexual intercourse with Sodomy, masochism, sadism, Reduced intelligence
paracentral glioma or minor daughter, bathing repeated sex proposals to
meningioma nude with children of both different woman, impulse
sexes, exposition of penis control problems, personality
in front of children and changes
encouragement to touch
it, urination into the face
of minor boy
Regestein and Reich (1978) 56 Right-sided frontal Multiple molestation of Increased sexual desire, more Reduced attention and
meningioma pre- and pubertal children frequent sexual intercourse cognitive flexibility
of both sexes with wife, personality
changes, inability to sustain
erection, inability to learn
new music
Miller et al. (1986) 34 Brainstem glioma involving Multiple sexual advances Collection of pornography, Normal intelligence
the thalamus, hypothalamus, towards young children, difficulties with erection and
ventral midbrain and pons including prepubertal ejaculation, conversation
(more pronounced in left daughter filled with sexual innuendo,
hemisphere) Idiopathic showing pornographic
hydrocephalus, Weber‘s pictures to visitors at home,
syndrome, Benedikt‘s personality changes, poor
syndrome, hypothyroidism financial judgement
Burns and Swerdlow (2003) 40 Right-sided orbitofrontal Interest in child Increased sexual desire Delayed recall, agraphia,
hemangiopericytoma pornography, advances (pornography, incl. child apraxia
towards prepubertal pornography), sexual
stepdaughter, denied advances towards female
previous attraction to medical staff and other
children patients, suicidal ideation,
balance problems, urinated
on himself being
unconcerned of it; severe headaches
Dementias
Mendez et al. (2000) 60 Frontotemporal dementia Molestation of Sexualized behaviour, Verbal stereotypies, mild
(hypometabolism in right prepubertal children, disinhibited and aggressive echolalia, diminished
inferior temporal region, exhibitionism behaviour, compulsive memory and executive
more subtle left) behaviour, decline in social functions
and personal awareness,
decreased judgement
Mendez and Shapira (2011) 67 Frontotemporal dementia Sexual advances towards Preoccupation with sexuality, Reduced verbal fluency,
(frontal atrophy and daughters multiple daily sexual diminished memory and
hypoperfusion, more advances towards wife, executive functions
pronounced in right fondling of wife in public,
hemisphere) touching of women’s breasts
in magazines, decline in work
performance, disinhibited
behaviour, compulsive
behaviour
Mendez and Shapira (2011) 82 Vascular dementia following Molestation of Frequent masturbation, Deficits in attention,
stroke (subcortical lacunes in stepdaughter talking about sex, sexualized verbal fluency, working
left caudate head and right language, inappropriate memory, executive
globus pallidus, touching of others, functions and visuospatial
hypometabolism in right disinhibited behaviour construction
posterior cingulate cortex)
Rainero et al. (2011) 49 Frontotemporal dementia Sexual arousal and urges Mutation in PGRN verbal and Mild episodic memory
(bilateral frontal atrophy and towards 9-year-old- physical hetero-directed impairment
hypoperfusion) daughter aggressive behaviour, no
further disinhibition
Parkinson’s disease
Berger et al. (2003) 57 Unknown Repeated sexual Increased sexual urges, Below average
molestation of 10-year- problems with erections and intelligence
old boy and 6-year-old girl orgasms
Mendez and Shapira (2011) 59 No known abnormalities Interest in child Increased sexual urges, use of No abnormalities
pornography pornography (incl. child
pornography)
Mendez and Shapira (2011) 59 Pallidotomy, no further Molestation of 5-year-old Increased sexual urges No abnormalities
abnormalities granddaughter, (hiring strippers and
masturbation while prostitutes, use of
viewing photo of pornography, intrusive
granddaughter sexual thoughts, forced wife
to have sex with him several
times a day, propositions to
wife’s female friends)
S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23 7
Table 1 (Continued )
Refs. Age Diagnosis/Localization Paedophilic symptoms Further symptoms Neuropsychological
findings
Further neurological disorders
Fairweather (1947) 24 (of 154) patients with Sexual molestation of Unknown Unknown
postencephalitic ‘‘little girls’’
parkinsonism
Mendez et al. (2000) 67 Hippocampal sclerosis Molestation of 5-year-old Sexual preoccupation, past Deficits in working
(volume loss in mesial boy; son of patient drug abuse, advances memory
temporal lobes and reported to have been towards female physician
hypometabolism in right molested by him in
temporal lobe) childhood
Devinsky et al. (2010) 39 Epilepsy (symptoms started Use of child pornography Hypersexuality, coprophilia, Unknown
after resecting posterior more frequent masturbations
temporal parts) and sexual intercourse with
wife, increased use of
pornography (incl. child
pornography)
Mendez and Shapira (2011) 32 Huntington’s disease Inappropriate touching of Verbally and physically Deficits in verbal fluency
(bilateral atrophy of caudate 6-year-old girl aggressive, constricted affect, and executive functions
and putamen nuclei; moodiness, depression,
hypometabolism of striatum) irritability, disorganization
absence of paedophilic preference. It might foster an unselective 2011). As in most cases described above, CSA did not arise in an
search for sexual partners. The neurophenomenological model isolated manner, but in the context of marked hypersexuality. In
supports this interpretation since behavioural control functions one patient the symptomatology appeared after a right pallidoto-
are ascribed to the implicated medial frontal areas (Redoute´ et al., my, which he underwent after suffering from Parkinson’s disease
2000; Stole´ru et al., 1999). for 16 years (Mendez and Shapira, 2011). As for the other two
examined patients, neuroimaging did not reveal any structural
2.2. Different forms of dementias alterations (Mendez and Shapira, 2011). All of these cases were
medicated with dopamine agonists, and no further psychiatric or
There are three reports of frontotemporal dementia (Mendez neuropsychological abnormalities were reported. Despite a
and Shapira, 2011; Mendez et al., 2000; Rainero et al., 2011) and sexually satisfying relationship with his wife, the patient described
one of vascular dementia following stroke (Mendez and Shapira, by Berger et al. (2003) seems to have been aware of a pre-existing
2011) leading to sexual child molestation. Two men with paedophilic disposition. However, he never acted on it until his
frontotemporal dementia exhibited frontal hypoperfusion and wife refused sexual contact, which was mainly due to the patient’s
atrophies, while one had no structural abnormalities on the erectile and orgasmic dysfunctions. These symptoms developed at
magnetic resonance imaging (MRI) scan; however, positron the same time as an increased libido. Both followed dopaminergic
emission tomography (PET) revealed right anterior temporal medication and were even more pronounced after the patient
hypoperfusion that may have been extended into the orbitofrontal increased medication on his own authority. This supports the idea
cortex. Computed tomography (CT) of the patient with vascular that brain pathology can lead to paraphilia-like behaviour without
dementia showed subcortical lacunes in the left caudate head and changing an actual existing sexual preference. Instead, neural
the right globus pallidus while PET additionally revealed a alterations might have dampened inhibiting mechanisms, which
hypometabolism in the right posterior cingulate. Three of these allowed for the manifestation of aberrant sexual behaviour,
four men showed hypersexual behaviour not limited to children. especially in the absence of other possibilities to act on strongly
This provides further evidence that brain pathology led to increased sexual urges. While again underlining a possible role of
disinhibition rather than to a specific change in sexual age the pallidum, this also implies that anatomical abnormalities must
preference. Additionally, in all cases personality changes, as well as not be a necessary prerequisite for the exertion of CSA, pointing to
obsessive-compulsive, impulsive or hetero-directed aggressive the importance of changes on the functional level. This is further
behaviour were described. Furthermore, all patients also suffered supported by the fact that all patients received dopaminergic
from neuropsychological deficits in executive functions, memory, medication, which implies a potential role of this transmitter
attention, verbal fluency, visuospatial abilities or demonstrated system in evoking hypersexuality. A recent cross-sectional study
verbal stereotypies, and echolalia. These reports suggested that the on 3090 patients with Parkinson’s disease confirmed effects of
temporal cortex, the basal ganglia, and the frontal regions, are dopaminergic drugs on impulse control disorders, including
involved in CSA via disinhibited sexual behaviour. A disinhibited compulsive sexual behaviour (Weintraub et al., 2010). A possible
sexual drive following lesions of the temporal lobes and the explanation is that some substances non-selectively bind to
amygdala is a well-known symptom of the Klu¨ ver–Bucy syndrome. dopamine D1 and D2 receptors mainly located in the dorsal
Klu¨ ver and Bucy (1939) concluded that the temporal cortex exerts striatum, and probably mediating motor effects, but also to D3
an inhibiting function on sexual behaviour. This is in line with the receptors abundant in the ventral striatum. The ventral striatum
neurophenomenological model (Redoute´ et al., 2000; Stole´ru et al., (which includes the pallidum) is well-known to be associated with
1999), which ascribes inhibiting and action control functions to the behavioural addiction and substance dependence (Potenza, 2013;
implicated structures. Schacht et al., 2013). Dopaminergic medication has repeatedly
been shown to alter activity of brain areas involved in reward and
2.3. Parkinson’s disease motivation (Kassubek et al., 2011). A functional magnetic
resonance imaging (fMRI) investigation suggests that such
CSA was also reported to occur in male patients suffering from dopamine-mediated effects might already occur in very early
Parkinson’s disease (Berger et al., 2003; Mendez and Shapira, stages of sexual arousal processing. Oei et al. (2012) presented
8 S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23
healthy males with sexual photographs, which were shown so fast, 3. Neuroimaging
that they could not be processed consciously. During this
subconscious processing, dopamine administration led to in- In this section we will review structural (Section 3.1) and
creased brain activation in brain regions associated with the functional neuroimaging studies (Section 3.2) on paedophilia. In
motivational component of sexual arousal (Redoute´ et al., 2000; addition to identifying brain areas associated with sexual arousal
Stole´ru et al., 1999), namely the nucleus accumbens and dorsal processing in paedophilia and analysing differences in brain
anterior cingulate. Since these regions are thought to regulate structure and function compared to a control group, numerous
behavioural drive towards a desired target (‘‘wanting’’; Berridge studies also reported correlations between brain volume or
and Robinson, 1998), dopaminergic medication could contribute to activity and parameters for paedophilic interest, offence char-
compulsive behaviour as a result of stimulating a ‘‘running start’’ of acteristics or psychopathology. These results are discussed in
this circuit. Section 3.3.
2.4. Further neurological disorders 3.1. Structural neuroimaging
There are isolated accounts of CSA in other disorders. Mendez 3.1.1. CT studies
and Shapira (2011) described a patient suffering from Huntington’s The first studies on structural brain abnormalities in paedo-
disease, who had bilateral atrophy of the caudate and putamen philes employed CT. Graber et al. (1982) observed reduced brain
nuclei as well as striatal hypometabolism. In one instance this man density in three men who had molested children (judging from the
inappropriately touched a 6-year-old-girl. Amongst other symp- case reports at least two of these did not show exclusive
toms, he had impulse control problems, heightened irritability and paedophilic interest). This was also found by Hendricks et al.
aggression, and executive deficits. Another patient suffering from (1988) who collected data from 12 men incarcerated because of
hippocampal sclerosis had bilateral mesial temporal volume molestation of children or adolescents (no diagnosed paedophiles),
reductions and right temporal hypometabolism (Mendez et al., 10 patients from radiology without cerebral abnormalities, and 16
2000). While also showing increased sexual interest and preoccu- healthy controls. The control group was inadequate since it
pation, this man was reported to have molested prepubescent and included women (three in the radiology patient group and two in
pubescent boys. Clues for a possible premorbid paedophilic the healthy control group), whereas all CSOs were male. Hucker
interest existed, particularly the patient’s son claimed to have et al. (1986) found enlargement of the left temporal and anterior
been molested by him as a child. Devinsky et al. (2010) reported a horns to be more common in paedophilic sex offenders (n = 41)
patient suffering from epilepsy, who began collecting child than in a non-sex offender control group (n = 14). Yet, these results
pornography following a second temporal lobotomy. Interestingly, were not replicated in subsequent CT studies (Langevin et al., 1989,
this activity as well as hypersexuality did not evolve until his 1988; Wright et al., 1990). However, Wright et al. (1990) noted
second temporal lobotomy during which more posterior temporal greater brain asymmetry in paedophiles, mostly with smaller left-
areas were resected. The first surgical intervention, in which more hemispheric widths, which was mainly attributable to the
anterior temporal regions and the amygdala were removed, did not temporal lobe. While the results from these studies are hardly
entail the symptomatology. Finally, Fairweather (1947) noted that comparable, two of them implicate some temporal abnormalities.
in his sample of 154 postencephalitic male patients, 24 had These abnormalities seem to be unspecific, since they were not
committed sexual offences against ‘‘little girls’’. only present in paedophilic sex offenders (all CT studies were
In none of the above cited cases, brain pathology specifically carried out with forensic samples), but also in other sexually
led to paedophilia. Rather, in the majority of cases CSA occurred aggressive men (e.g., non-paedophilic incest offenders; Langevin
in the context of hypersexuality, broader changes in personality, et al., 1988). This further supports the notion that temporal
impulse control problems and/or neuropsychological deficits. abnormalities might play a role in disinhibited behaviour. Besides
Thus, a change in sexual age preference caused by the involved the low spatial resolution, CT studies are limited to the analyses of
pathological processes may not be a sufficient explanation for the frequencies of rater-classified structural pathology, thereby
full spectrum of symptoms observed. It is more likely that the possibly missing more subtle abnormalities. In addition, conclu-
abnormalities resulted in reduced behavioural control leading to sions about paedophilia from these reports are impossible because
paedophilia-like behaviour on the background of a premorbid the necessary distinction between paedophilic men and CSOs is
sexual interest in children or a lack of more preferred sexual missing.
partners. Brain damages described were located mainly frontal,
temporal or within the basal ganglia. According to the 3.1.2. MRI studies
neurophenomenological model (Redoute´ et al., 2000; Stole´ ru The first study using MRI to study structural brain abnormali-
et al., 1999) these structures are implicated in inhibition and ties in a sample including paedophilic men was conducted in
appraisal processes, which fits well with the interpretation made Austria. Eher et al. (2000) investigated a sample of 38 imprisoned
above. sexual offenders of which 15 had offended against minors and
Case reports naturally suffer from the shortcoming that they are eight met diagnostic criteria for paedophilia. Similar to previous CT
based on observations of single patients. In these studies studies, and unlike subsequent MRI studies (which used more
paedophilic preferences often have not been diagnosed thorough- quantitative methods, see below), a qualitative analysis approach
ly. These studies were the first to uncover brain regions, which was chosen investigating narrative reports of neuroradiologists.
could provide evidence for why some patients act out abnormal These were coded for the presence of brain abnormalities and
sexual interests. It should be stressed that patients who acted on subsequently, respective frequencies were compared between
their urges were investigated, which could be the reason for the groups. In a total of 38 sexual offenders, 17 were found to have
detection of areas primarily implicated in control processes. In structural abnormalities: four cortical atrophy, one periventricular
order to identify brain regions involved in the sexual preference for lesions, six deep white matter lesions, one ventricular enlarge-
children, diagnosed paedophile patients would have to be directly ments, and five multiple lesions. Additionally, in offenders without
compared to non-paedophile control subjects. This has been anatomical abnormalities self-perceived aggressiveness was cor-
accomplished in several neuroimaging studies conducted within related with self-reported interpersonal problems, and social
the last years, which we will discuss in the following sections. anxiety. This association with aggressiveness was not found in the
S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23 9
Table 2
Summary of structural MRI studies.
Refs. Sample Sexual orientation Medication Method Main findings Significance
Schiltz et al. (2007) 15 CSO, inpatient Exclusive-type None in at VBM/MM GM reductions in p < .05 corrected
15 community paedophilia in at least 13 CSOs paedophiles: for multiple
controls least 13 CSOs Amygdala R, comparisons
Gender orientation Hypothalamus L/R, within ROIs
unknown Innominate substance L/R,
Bed nucleus striae
terminalis L/R, Septal
region R
Enlargement of the
temporal horn R in
paedophiles
Schiffer et al. (2007) 18 CSO, inpatient Exclusive-type None VBM GM reductions in p < .05 FDR
24 community paedophilia paedophiles: corrected for whole
controls Attracted to males: Cerebellum L/R*, Inferior brain
9 paedophiles frontal gyrus L/R*, *Additionally
12 teleiophiles Cingulate gyrus L/R*, p < .05 FWE
Posterior cingulate gyrus corrected within
L*, Insula L/R*, Precuneus ROIs
L/R, Parahippocampal
gyrus L/R*, Superior
temporal gyrus L/R,
Middle temporal gyrus R,
Putamen L/R*
(Amygdala L/R in
unpublished re-analysis)
Cantor et al. (2008) 44 CSO, outpatient Unknown Unknown VBM/MM WM reductions in paedo- p < .05 FDR
(paedo- & and hebephiles: corrected for whole
hebephilic) Superior fronto-occipital brain
53 NSO, outpatient fasciculus L, Arcuate
fasciculus R
Cantor & Blanchard 19 paedophilic CSO, Unknown Unknown MM WM reductions in paedo- p < .05 uncorrected
(2012) outpatient and hebephiles in bilateral (2 models)
49 hebephilic CSO, temporal (p = .04) and
outpatient parietal lobes (p = .06)
47 NSO, outpatient
Poeppl et al. (2013) 9 CSO, inpatient Exclusivity of age None VBM GM reduction in p < .05 FWE
11 NSO, inpatient preference paedophiles: corrected within
unknown Amygdala R ROIs
Attracted to males:
7 paedophiles
0 teleiophiles
Annotations: MRI magnetic resonance imaging, CSO child sex offender, NSO non-sex offender, VBM voxel-based morphometry, MM manual morphometry, ROI region of
interest, L/R left/right, GM grey matter, WM white matter, FWE familywise error correction, FDR false discovery rate.
17 offenders with structural abnormalities. Eher et al. (2000) included in these studies. In four of the investigations voxel-based
speculated that the observed multi-locus alterations might lead to morphometry (VBM) was used for data analysis. Two studies
violent behaviour by interfering with social information proces- additionally employed traditional morphometry; one study used it
sing. Although the study’s findings were interesting, the reported exclusively. In the following sections we will review the results
results are unspecific and the localization of anatomical abnor- concerning group differences between paedophiles and teleio-
malities remains unclear. Furthermore, conclusions regarding the philes from these studies (see Fig. 3 for a summary of
paedophilic subgroup are impossible (although this was not the methodological details).
main aim of the investigation), because paedophilic CSOs were The first published study by Schiltz et al. (2007) analysed MRI
analysed jointly with CSOs without paedophilia. Last, a control data of 15 paedophilic sex offenders recruited from a forensic
group was missing. clinic, and 15 community controls using VBM and manual
Further MRI studies appeared from 2007 onwards. These used morphometry. Six of the paedophile patients had committed
higher spatial resolution (Eher et al. (2000) employed a T2- offences exclusively against girls, three exclusively boys, and six
weighted spin-echo imaging sequence in axial plane with a slice against both sexes. Further sample characteristics were not
thickness of 5 mm at 0.5 Tesla, while subsequent studies utilized reported, but as Schiltz et al. (2007) analysed data from the same
T1-weighted 3-dimensional sequences with a maximum slice sample as Walter et al. (2007) (see Section 3.2.2.2 and Tables 2 and
thickness of 1.5 mm at 1.5 Tesla), more sophisticated quantitative 4), it can be concluded that at least 13 of the 15 patients were
analysis methods and focused investigation of well-defined exclusively oriented towards children and did not receive
paedophilic samples (diagnosed by structured clinical procedures). medication (Walter, personal communication). The VBM results
To date, five such studies have been published (summarized in were restricted to four predefined regions, known to be associated
Table 2), with three of them conducted with German samples and with the regulation of sexual function: (1) The left and (2) right
two stemming from Canada, with the latter two relying on the amygdala, (3) the bilateral septal region and the bed nucleus striae
same data set. Overall 61 paedophilic (and 49 hebephilic men; terminalis as well as the (4) hypothalamus and the substantia
Cantor and Blanchard, 2012) and 103 teleiophilic men were innominata. The authors employed a significance level of p < .05,
10 S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23
Fig. 3. Relevant methodological specifics of MRI studies. See refs. (Ashburner and Friston, 2000; Genovese et al., 2002; Kriegeskorte et al., 2009).
corrected for multiple comparisons within these regions of interest handedness, and gender orientation. In both groups, 50% of
(ROIs). VBM and manual morphometry exhibited a reduction of participants were exclusively attracted to males, while the other
right amygdala volume accompanied by an enlargement of the 50% were exclusively attracted to females. The paedophilic
right temporal horn in paedophilic offenders. While manual patients were exclusively attracted to prepubescent children and
morphometry showed a similar effect for the left amygdala, this they were not medicated. The authors applied both whole brain
could not be confirmed by VBM. VBM did reveal volume reductions (thresholded at p < .05, false discovery rate correction, FDR) as
in all other ROIs. Schiltz et al. (2007) suggested that neurodevelop- well as ROI analyses (p < .05, familywise error correction, FWE)
mental problems cause the abnormalities, proposing that a carried out if an a priori ROI was significant at an uncorrected
dysfunction of the structures implicated might interfere with threshold of p < .001 and a cluster extent of k 100 voxels) to
sexual maturation. They further speculated that a devaluation of their data. Assuming a relation of paedophilia to obsessive-
infantile sexual interest, which would usually develop during compulsive spectrum disorders, Schiffer et al. (2007) predicted
puberty, might not occur (Freund and Kuban, 1993). associations within the frontrostriatal, limbic system, insulo-
The second study was published by Schiffer et al. (2007), opercular segments, and cerebellum. The whole brain analysis
applied VBM and compared 18 paedophilic inpatients with 24 yielded grey matter reductions in paedophilic patients in the
controls. The sample did not differ significantly in age, bilateral orbitofrontal cortex, insula, ventral striatum (putamen),
S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23 11
cingulate, precuneus, parahippocampal gyrus, superior temporal ‘‘The larger region followed the superior fronto-occipital
gyrus, cerebellum, left posterior cingulate, and the right medial fasciculus; it ran along the dorsal border of the caudate
temporal gyrus. Of the 15 ROI analyses carried out additionally, nucleus, extending posteriorly toward the occipital pole and
only an effect in the left parahippocampal gyrus would remain extending anteriorly to terminate primarily in the middle
statistically significant at p < .05 FWE if one applied Bonferroni frontal gyrus. The anterior extension appeared also to terminate
correction (for the 15 separate tests). Consistent with their in two other, weaker clusters, one in the frontal operculum and
hypothesis, the authors interpreted their results as evidence for one in the opercular part of the inferior frontal gyrus. [. . .] The
relatedness of paedophilia with the obsessive-compulsive second region occurred only in the right hemisphere, following
spectrum. In addition, in an unpublished post hoc analyses of the arcuate fasciculus from the parietal opercular region to the
the data, Schiffer (personal communication) confirmed ROI- temporal opercula, wrapping around the posterior extension of
corrected (p < .05 FWE) volume reductions within the bilateral the Sylvian fissure and descending to the inferior temporal and
amygdala, consistent with the results of Schiltz et al. (2007). fusiform gyri’’ (Cantor et al., 2008, p. 137).
The third German study in which data of paedophilic patients
were analysed with VBM was published by Poeppl and colleagues In addition, a phallometric paedophilia index score correlated
in 2013. They compared nine unmedicated paedophilic CSOs with with brain regions that largely overlapped with the areas revealed
11 non-sexual offenders (NSOs, who had mainly committed by the manual and voxel-based morphometry analyses (see also
property offences). Importantly, as both groups were recruited Section 3.3 and Tables 2 and 7). This study had some major
from forensic high security hospitals, the authors controlled for methodological strengths like the large sample, availability of
potential effects of forensic inpatient status. Seven of the phallometric data, stringent statistics, and the inclusion of a NSO
paedophilic patients were attracted to males but none of the control group. However, the results were limited by the possibility
controls, which limits interpretability of the results. Moreover, that hebephilic subjects might have been included in the data set; a
information regarding the exclusivity of paedophilic interest is limitation which led to a reanalysis by Cantor and Blanchard
lacking and the paedophilic group was significantly older than the (2012).
NSOs. The authors applied whole brain analyses and small volume The reanalysis, based on phallometric data or self-report,
corrections for all regions in which Schiltz et al. (2007) and Schiffer classified subjects as paedophilic if there was evidence for sexual
et al. (2007) found grey matter reductions at a FWE corrected interest in children below 10 years of age, hebephilic for interest in
significance level of p < .05. The only region for which Poeppl et al. children between 11 and 14 years and teleiophilic if both the
(2013) replicated grey matter reductions in was the right phallometric response as well as self-report confirmed interest in
amygdala, although this finding would not reach the required men and women older than 17 years, and no sexual offences
significance threshold if corrected for the number of ROI analyses, against minors were known. Accordingly, 19 subjects were
as the authors remark themselves. The authors concluded that an classified as paedophilic (as opposed to 44 in the original
anatomical reduction of amygdala volume may not directly cause analysis), 49 as hebephilic, and 47 as teleiophilic. There was no
paedophilia. Instead, in line with Schiltz et al. (2007), they differences in age or years of education between these groups. The
suggested that the observed abnormality might be evidence for a analyses were restricted to a manual morphometric examination
deficient developmental differentiation of this structure. Further, of those regions for which significant reductions were found
this aberrant differentiation could result in the non-initiation of a previously (Cantor et al., 2008), namely the white matter portions
devaluation process of infantile erotic interest. of bilateral temporal and parietal lobes. The analyses showed no
In addition to the three German studies described above, Cantor differences between paedophilic and hebephilic subjects for
et al. (2008) examined a Canadian sample of 44 CSOs who either temporal white matter but reduced volumes in both groups
confirmed paedophilic interest in a diagnostic interview, showed a compared to teleiophilic men. Group differences for parietal
stronger phallometric response to photos of children compared to regions had a similar pattern, although these were only
adults or (when phallometry was not available or unusable) had marginally significant (p = .06). Therefore, this reanalysis showed
committed at least one sexual offence against a child below the age comparable white matter abnormalities in paedophilia and
of 14. Hence, although the collection of phallometric data is a hebephilia, although the spatial resolution here was low. Cantor
strength of this study, the diagnostic heuristic leaves the et al. (2008) suggested that the observed white matter reductions
possibility that non-paedophilic offenders might have been could be indicative of diminished connectivity between regions
included here, since children below the age of 14 might very well involved in the processing of sexual arousal. However, connec-
have reached puberty and because there may be other reasons than tivity analyses (techniques to investigate the interaction of
paedophilia to molest children (see Section 1.3). In the offender different brain regions) or diffusion tensor imaging (DTI; An
group, 91% of men had molested girls and 32% had molested boys. MRI technique which allows inferences on white matter structure
No further information regarding sexual preference or medication by measuring the diffusion of molecules, especially water) would
were given. The control group consisted of 53 non-paedophilic be needed to support this conclusion.
NSOs, who did not differ from the patient sample in age, In summary, there is one structural MRI finding observed in
intelligence, years of education, and handedness. A major three of the four published samples. Schiltz et al. (2007), Schiffer
difference to the other studies is that subjects were not (personal communication), and Poeppl et al. (2013) all found
incarcerated but on probation or parole. In addition to VBM, the amygdala volume reductions, pointing to disturbances in emo-
authors manually parcellated 39 regions for which they tested tional processing of sexual arousal (Redoute´ et al., 2000; Stole´ru
group differences combined in four models (for cortical and et al., 1999). The fact that this finding was replicated twice is
subcortical grey matter, white matter, and cerebrospinal fluid) to promising, still, it requires further confirmation as the results were
circumvent multiple testing. Only whole brain analyses corrected obtained with multiple testing and at least in one case (Poeppl
for FDR were performed. With both techniques, no group et al., 2013) would not withhold statistical correction for these.
differences in CSF or grey matter were observable, which stands Aside from this, the available investigations led to rather
in contrast to the three German reports. Instead, manual heterogeneous results. While the three German studies conducted
morphometry pointed to temporal and parietal white matter with smaller samples found grey matter reductions in varying
deficiencies in the CSO group. Consistent with this, VBM showed brain structures (Poeppl et al., 2013; Schiffer et al., 2007; Schiltz
that especially two regions were involved here: et al., 2007), studies relying on one Canadian data set showed
12 S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23
white matter reductions exclusively (Cantor and Blanchard, 2012; fMRI studies on different psychological processes (response
Cantor et al., 2008). inhibition and face processing) in paedophilic men, which are
described in Section 3.2.2.3.
3.2. Functional neuroimaging
3.2.2.1. fMRI case reports on sexual arousal processing and neural
3.2.1. PET studies effects of LH-RH agonists. To date, four case reports providing data
There are two studies already mentioned (Graber et al., 1982; on sexual arousal processing in paedophilic CSOs have been
Hendricks et al., 1988) in which, additionally to CT scans, PET was published (Table 3). All of the findings summarized in the
employed to investigate alterations in brain activity in CSOs. following section have to be taken cautiously because the results
Results of both investigations showed reduced cerebral blood flow were acquired from individual subjects that probably show
in the CSO groups. In the study by Hendricks et al. (1988) this substantial inter-individual variability in activation patterns.
difference was greater at anterior than posterior sites. In a further Furthermore, they differed in their gender orientation, exclusivity
PET investigation Cohen et al. (2002) compared seven CSOs of paedophilic preference, and medication status. Also, stimuli and
diagnosed with non-exclusive paedophilia (attracted to girls) who statistical analysis (such as contrasts and significance levels)
underwent an outpatient treatment with seven healthy controls. varied across studies.
They used narrative scenarios about sexual contact between a man In three of the four fMRI case studies intra-subject contrasts
and either (a) a prepubescent girl, (b) a 30-year-old woman or (c) were performed comparing the neural processing of preferred
words from the dictionary being read out loud. During the last (nude children or children dressed in swimsuits) against non-
condition they observed diminished glucose metabolism in preferred stimuli (dressed children or nude adults). There is some
paedophiles within the right inferior temporal gyrus and ventral overlap in activation of occipital areas, the right prefrontal cortex,
superior frontal gyrus. In addition to no group differences during anterior cingulate cortex, and superior temporal gyrus (Table 3).
the condition of interest, observed effects were detected via ROI Activation of all of these areas have also been observed by group
analyses, of which a total of 120 were conducted. The authors studies on paedophilic men (see Section 3.2.2.2 and Table 5).
discuss that no effects survived correction for multiple compar- However, only two of these three studies can be compared directly
isons. Hence, although the results of Cohen et al. (2002) pointed to (which is further complicated by imprecise coordinates given in
regions previously mentioned in case reports, these first results one study; Dreßing et al., 2001). The findings reported by Schiffer
remained inconclusive and interpretability suffers from methodo- et al. (2009) were obtained from a patient who was on antilibidinal
logical flaws like small sample sizes, unclear diagnoses and/or medication for three months (while the others were not).
imprecise localizations. Narrowing comparisons to the two remaining studies (Dreßing
et al., 2001; Habermeyer et al., 2012) decreases concordant
3.2.2. fMRI studies activation to the fusiform gyrus. This is likely due to high inter-
There are seven fMRI investigations and four single case reports subject variability in brain activity because there is large overlap
(three of these primarily studied neural effects of antilibidinal for intra-group contrasts across respective studies (see Section
medication) on sexual arousal processing in paedophilia. A total of 3.2.2.2).
87 paedophilic and 100 teleiophilic men have been examined. Because of the small number of subjects included, most authors
Eight of these investigations were performed in Germany, two in of case reports abstained from making direct comparisons between
Switzerland, and one was conducted in France. Voxel-wise them with the exception of Moulier et al. (2012). After masking out
analyses over the whole brain were conducted in 10 of the 11 activation observed in the control subject from the same contrast
studies. In the remaining study analyses were constricted to the as analysed in the paedophilic patient (male children in
amygdala. swimsuits > female children in swimsuits), the patient showed
In all of these studies erotic pictures were presented in order to additional activity within the left calcarine fissure, left anterior
compare their neural processing in paedophilic men and non- insula, caudal anterior cingulate, and left cerebellar vermis. These
paedophilic control groups. Erotic pictures typically were photo- findings show little agreement with results from inter-group
graphs of children or adults, either nude or dressed in bathing suits contrasts (Table 6), although, results of overall group comparisons
or underwear. These were compared to control conditions during in general are quite inconsistent (see Section 3.2.2.2).
which either fully dressed people, people of a non-preferred age or Finally, the main focus of three of the fMRI case reports was to
neutral stimuli were shown. Some fMRI paradigms included investigate effects of pharmacological treatment with a long-
control tasks, like pressing a button whenever a certain stimulus acting luteinizing hormone-releasing hormone agonist (LH-RH,
appeared on screen to ensure attention of subjects. Further leuprolide acetate) on neural activation in paedophilia. LH-RH
information on the employed paradigms can be found in Table 4. agonists were shown to suppress deviant sexual urges and
Studies typically investigated the following questions: arousability (Ro¨sler and Witztum, 1998). The report by Schiffer
et al. (2009) only allows indirect comparisons of the activity found
(1) On the intra-group level: Which regions are activated during in the medicated patient with those seen in their previous
visual sexual stimulation? unmedicated sample (Schiffer et al., 2008b) at best, since he was
(2) On the inter-group level: Are there differences in brain activity scanned only once while medicated. The lack of recruitment of a
between paedophiles and control subjects during stimulation number of areas involved in emotional, motivational, and
with child and/or adult stimuli? autonomic aspects of sexual arousal (e.g., thalamus, amygdala,
(3) On the inter-group level: Are there differences in brain activity insula, substantia nigra, hippocampus, and the rostral anterior
between paedophiles and control subjects when only proces- cingulate), which the authors observed in their sample of
sing of the respective arousing stimuli is taken into account unmedicated paedophilic CSOs (Schiffer et al., 2008b), suggested
(children in paedophiles and adults in teleiophiles)? that LH-RH agonists might primarily suppress subcortical brain
activity. Two longitudinal studies on LH-RH effects on neural
In the following sections, we will first review the publications activation used longitudinal designs, scanning the patient before
on sexual arousal processing in single cases using fMRI (Section and some months after the onset of treatment. Partially confirming
3.2.2.1), before we will turn to studies providing group statistics on the hypothesis by Schiffer et al. (2009), Moulier et al. (2012) found
that matter (Section 3.2.2.2). In addition, there are two further that after five months of leuprolide acetate treatment, activity was
S. Mohnke et al. / Progress in Neurobiology 122 (2014) 1–23 13
Table 3
Summary of fMRI case reports.
Refs. Subjects Treatment/Design Paradigm/Stimuli Main findings
Dreßing et al. (2001) Patient: Treatment unknown Passive viewing of women in Stronger recruitment of the visual