Keywords

A series of high-profile cases in the last decade have raised public awareness of some of those with autism who have broken the law.Footnote 1 They have all made front page headlines and received prime time press coverage, with the diagnosis of autism a key feature of the story and include Gary McKinnon (Kushner, 2011), Adam Lanza (Soloman, 2014), “RXG” (Dodd, 2021) and Lauri Love (BBC, 2018). They have been characterised as ‘geeks’, and ‘loners’, as ‘shy’, ‘awkward’, or ‘odd’, and this makes them vulnerable to sensationalist tabloids portraying them as a threat. However, autism lies beyond such interpretations of behavior, and façades of personality, and takes many different forms, otherwise all individuals with autism would be law breakers.

Since inception, the ‘auto’(‘self’) aspect of autism has been interpreted as indicating antisocial traits. Asperger himself referred to the presentation as ‘autistic psychopathy’ (Asperger & Frith, 1991), whilst Kanner framed it as a ‘difficulty to relate’ (Kanner, 1943), implying that the ability to develop relationships makes us ‘social’, rather than ‘anti-social’. Yet, many people with autism express a strong desire for friendships.

Whether for exchange or gain, procreation, protection or simply enjoyment from interaction, it has been suggested that to be human is to engage with others—“No man is an island” (Donne, 1988). A premise of community is inherent in notions of what it is to be social and those seen to ‘go it alone’ or who fail to conform to norms are often identified as ‘other’, which is easily transposed as ‘deviant’.

Autism therefore, as a condition characterised by atypicality in social interaction and behavior, with oft quoted ‘deficits’ in empathy and sometimes a lack of interest in others, might lend itself to expectations of criminality, which by definition involves omissions or the violation of the rights of others. However, as will be illustrated in this chapter, there is as yet no evidence-base suggesting individuals with autism are any more or less likely to break the law than those without autism, and those who do offend show marked similarities to offenders without autism. Furthermore, questions over whether empathy protects people from breaking the law (or re-offending) are considered and indeed, whether the cognitive and behavioral features of autism do in fact suggest an anti-social or even asocial profile.

What is clear from the limited research on offenders with autism is that while a diagnosis may present some vulnerability to committing an offence, and challenges to rehabilitation, best practice for the treatment needs of autistic offenders is yet to be determined.

Autism and Offending: Who, Where and How Many?

Information available regarding offenders with autism comes from the domains of justice, mental health and social care (see Table 1). As with other crime data, the picture is incomplete, information is often partial and can vary across services, health and social systems and nations, making accurate estimates of the number of individuals with autism who break the law challenging. In addition to being incomplete, several other factors can impact the variability in rates of prevalence or incidence. Table 1 shows figures from studies of offending populations, i.e. in prisons, secure, and forensic services. They provide answers to the question “Are autistic people overrepresented in these settings?” If they are not, then prevalence figures should be approximately 1% for the rates of autism in these settings (Baird et al., 2006). However, there are some complicating factors (see below).

Table 1 A selection of studies on prevalence and incidents of autism and offending
Full size table

Autism Diagnosis

How an offender with autism is defined can vary across studies and may inflate some figures, whilst leading to undercounting in others. For example whether an individual has a specific diagnosis i.e. Asperger’s Syndrome (Ghaziuddin et al., 1991) or only displays ‘autistic traits’ (Geluk et al., 2012; Sutton et al., 2013) can affect figures. (Also see Table 1). A diagnosis of autism spectrum disorder can be provided by a medical professional using diagnostic criteria from the DSM-V (APA, 2013) or ICD-10 (WHO, 2016), and/or from findings of a ‘gold standard’ assessment of autism such as the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2) (Lord et al., 2012), the Autism Diagnostic Interview-Revised (ADI-R) (Rutter et al., 2003) or Diagnostic Interview for Social and Communication Disorders (DISCO) (Wing 2006). Some of these diagnostic tools and assessments include specifiers such as Asperger’s Syndrome and Pathological Demand Avoidance (WHO, 2016), whilst others include broader classifications of autism and autism spectrum disorder (Lord et al., 2012). However, as will be illustrated, much of the current literature, particularly case studies and prevalence data in forensic and mental health include a putative autism diagnosis (Esan et al., 2015), scores on screening tests for autism, or diagnosis from file records that do not necessarily provide the source or method or classification of diagnosis (Kohn et al., 1998; Raina & Lunsky, 2010) (see Table 1). Additionally, some data have included suspected but not confirmed, autism diagnoses (Moultrie & Beckett, 2011).

The recognition and identification of autism has improved greatly over the last ten to 20 years with the Autism Act (2009) in England and Wales, the Australian Autism State Plan (2009), the National Hungarian Autism Strategy (Magyar Köztársaság Kormánya, 2010) and, the Spanish Strategy on Autism Spectrum Disorders (Ministerio de Sanidad, 2015) all improving services for assessment and support. Such improvements however, have not been equally dispersed. For example, England and Wales initially focused efforts on the assessment and support of children with autism creating the need for an Autism Act to fill the gap for adults, and the needs of particularly low priority or vulnerable groups, such as offenders, were considered much later. Autistic individuals with forensic needs were only specifically identified in the 2015 updates (Department of Health, 2015) of the policy implementing the Autism Act (2009). Information from countries with different systems or perhaps no recognised autism social policy, or legislation, further limits details on those with autism who break the law outside Western culture or societies with colonial-based care and justice systems. Currently no information regarding autistic perpetrators of crimes is available from victim surveys such The Crime Survey for England and Wales, or victim support agencies in the United Kingdom, including the national charities such as the National Society for the Prevention of Cruelty to Children (NSPCC) or Barnardo’s.

Improved awareness of autism has allowed for autistic offenders to be identified outside specialist or intellectual disability services, however the challenges of diagnostic overlap and frequent psychiatric co-morbidities, including personality disorders and intellectual disabilities, can create further challenges in providing a true estimate of offending in autistic populations, and of autism in offending populations. For example, many older offenders with autism or Asperger’s Syndrome may have been diagnosed with schizophrenia or a personality disorder (Luciano et al., 2014; Takara et al., 2015), or received no diagnosis at all, depending on when they were sentenced or committed their crime since autism has only been included as a diagnosis by the American Psychiatric Association and World Health organization since the 1980’s, and as such these individuals are not represented in prevalence data. Additionally, a lack of expertise or confidence in distinguishing between autism and intellectual disability, or autism and personality disorder, and a co-morbid presentation, including in specialist intellectual or developmental disability services, may further hinder accurate estimates of prevalence data (Cuccaro et al., 1996; Shah, 2001). This may be particularly true in young offenders, where there is hesitancy over diagnostic labels (Fernald & Gettys, 1980; Kite et al., 2013), and with the overlap in clinical presentation between autism and conduct disorder, oppositional defiant disorder and attachment disorders, particularly if focusing on the ‘auto’ features of autism, or traits such as ego-centricity, lack of interest/consideration with others, etc., adding further complexity to identifying and recognising autism in those who also display anti-social or criminal tendencies (Mandell et al., 2007; Mayes et al., 2017; Moran, 2010).

Rates of autism in offending populations vary by population and sample. For example, autism may be under-recognised in prisons due to lack of standardised screening procedures, and the assumption of a certain level of cognitive adaptive social functioning as implied by mens rea and proceeding through a trial or Court process (see Table 1). Information from mental health institutions, including forensic mental health and specialist intellectual developmental disability services, often indicate over-representation of individuals with autism (Esan et al., 2015; Scragg & Shah, 1994), in comparison to the prevalence estimate of 1% of the general public (Baird et al., 2006). However, this needs careful interpretation: It may be that having autism and a mental health condition increases the risk of offending, or it may be that offenders with autism get diverted disproportionately from prisons into mental health services.

With such variation in the figures for the percentage of people with autism in offending populations (Table 1), it is difficult to ascertain an accurate picture of autism in offending populations.

Far fewer studies ask the question, “What percentage of people with autism offend?” This is a very different question (compared to the question of how many people with autism there are in CJS systems), as comparison samples of non-autistic people are needed, and it is known that amongst young men large percentages of them have minor convictions by the time they are in their 20s and 30s. For example, in the Farrington sample of working class boys from south London 37% had convictions by 32 years (Farrington, 1995), and hence some literature suggests much lower rates of offending than this for autistic young people (Howlin, 2004; Woodbury-Smith et al., 2006).

The variation in methodology, diagnostic criteria and sample selection in many of the above mentioned studies have been strongly criticised for not taking account of the biased samples being used (King & Murphy, 2014). A systematic review of individuals with autism involved with criminal justice systems, compared to those without autism, found that unbiased community-based samples indicated no differences and no over-representation of autism in CJS systems (King & Murphy, 2014).

Autistic Offenders in the Criminal Justice System: Characteristics, Types of Crime and Vulnerabilities

Typically, offenders without autism are overwhelmingly male, whether samples have been gathered from the community, in courts, in prisons, or on probation (Coleman & Moynihan, 1986; Farrington, 1995; Harvey et al., 1992), so it might be expected that people with autism who have offended will also be predominantly male. In fact, of course, autism is also more often diagnosed in males than females and, although rates vary in different studies, a recent systematic review concluded that an autism diagnosis was 4.6 times as likely in males as in females (Loomes et al., 2017), though there have been some assertions that autism has been somewhat under-diagnosed in women (Lai et al., 2011). It is to be expected therefore that studies of people with autism in the CJS will report a high proportion of males, as indeed they do. For example, Tint et al. (2017), in a Canadian study of 462 families with children over 11 years or adults with autism, reported that 78.5% of their autistic sample were male, 16% of them had had encounters with the police, and of these 78.3% were male. Gender therefore did not predict police encounters in this study and nor did the presence of Intellectual Disability [ID]). However, significantly more of those who had encounters with the police were reported to have had difficulties with aggressive behavior, and had not had structured day services, and came from families who were apparently unable to afford services. In a similar US study of a nationally representative sample of 920 youth with autism (Rava et al., 2017), nearly 20% had had encounters with the police by age 22 years and approximately 5% had been arrested. In this US study, females with autism were significantly less likely to have been stopped by the police than males with autism, and those individuals with externalising behaviors like aggression and tantrums were more again likely to have been stopped and more likely to have been arrested. The presence of Attention Deficit Hyperactivity Disorder, degree of communication difficulties (a proxy for ID), and, total household income did not affect the likelihood of encounters with the police. Nevertheless, in two studies from Japan, some of the important background factors affecting offending for people with autism were adverse childhood events. The autistic criminal groups were significantly higher than the autistic non-criminal groups or non-criminal non-autistic groups on childhood adversity factors, such as family violence, physical and sexual abuse, neglect, parental death, divorce and other parental loss (Kawakami et al., 2012; Kumagami & Matsuura, 2009).

People with autism commonly have other psychiatric disorders such as anxiety, depression, Attention Deficit Hyperactivity Disorder, and schizophrenia, with recent estimates suggesting that approximately 70% have at least one additional disorder and approximately 40% have two (Simonoff et al., 2008). It seems that these additional difficulties increase the likelihood of involvement in the CJS, particularly in the case of Attention Deficit Hyperactivity Disorder (Lundström et al., 2014) and psychosis and personality disorder (Fazel et al., 2008). However, it seems that having an additional diagnosis of ID does not increase the risk of involvement in the CJS. Lindsay et al. (2014), for example, analysed over 400 referrals to forensic ID services and found that approximately 10% had autism. They argued that since approximately 10% of people with ID not referred to forensic services also have autism, then autism did not increase the risk of forensic referral. Other researchers have come to similar conclusions from different samples. Mouridsen et al. (2008), for example, recorded that in their autistic sample (N=313) in Denmark, those who had Asperger syndrome (N=114) had a similar rate of offences to the general population sample (N=933), while those with both autism and ID had much lower offence rates, suggesting that ID was a protective factor. Quite why this may be, is not known, but one possibility is that more disabled people are more likely to be accompanied by carers in the community who can assist them to stay out of trouble, whereas those with Asperger’s may be out alone.

A number of suggestions have been made about the way that the characteristics of those with autism may predispose them to offending. For example, that they may be led into crimes by their social naivety, or misunderstanding of social cues, by their reactions to disruptions of routines, or through their pursuit of obsessions and circumscribed interests, and/or through difficulties with theory of mind (Mouridsen et al., 2008). Nevertheless, direct attempts to test these characteristics as contributory factors to offending have unsuccessful. Woodbury-Smith et al. (2010) examined the circumscribed interests of a group of adults with autism who had offended and a group with autism who had not offended. All participants could describe their circumscribed interests and all said they would like to spend more time on them. Those with an offending history tended to have more violence-related special interests, but there were only two cases where there was a clear and specific link between the person’s interests and the crimes themselves, including one case of arson and one theft of electrical equipment. This suggested that although crimes may be linked to special interests occasionally, this is unusual, and that it is possible to have special interests in violent material (such as in World War 2) without committing violent offences.

Connections between offending and other characteristics of autism, such as theory of mind, emotional recognition and executive functioning, have also been examined. Woodbury-Smith et al. (2005), for example, compared people with autism who had offended with those who had not, and with a general population control group. They measured theory of mind, emotional recognition and executive functioning. They found that generally the comparison group did better than the autistic non-offending group in executive functioning, emotion recognition and theory of mind, but the autistic offending group only differed significantly from the comparison group in having poorer recognition of fear. This is also known to be impaired in people with psychopathy (Blair et al., 2004) and Woodbury-Smith et al. considered it might be possible that their offending sample were co-morbid for psychopathy. Subsequent research did demonstrate higher levels of callous-unemotional traits in adolescents with autism—about 50% of them were above the cut-off of the test—and poorer recognition of fear in those with high callous-unemotional traits, but there was no difference in levels of conduct disorder between those with high, vs. those with low, callous-unemotional traits (Leno et al., 2015).

As regards the types of offences which people with autism commit, early evidence suggested that people with autism were likely to commit violent offences, including sex offences, and arson (e.g., Hare et al., 1999). However, these studies were sometimes based on hospitalised samples, rather than unbiased samples, and they often had no comparison groups (King & Murphy, 2014 for a discussion of this issue). Later studies of more representative samples, with comparison groups, showed a wide range of crimes; there were some suggestions that youth with autism were proportionately less likely to commit crimes against property, while they were more likely to commit crimes against people, than were youth without autism (Cheely et al., 2012; Kumagami & Matsuura, 2009; Mouridsen et al., 2008; Woodbury-Smith et al., 2006). Some studies also found lower rates of drug offences (Woodbury-Smith et al., 2006) and lower rates of driving offences (Mouridsen et al., 2008) amongst those with autism, compared to control groups. Meanwhile, in their sample of over 400 people referred to ID forensic services, Lindsay et al. (2014) reported lower rates of sexual offences amongst those with autism and ID, compared to those with ID alone.

There have been suggestions recently that people with autism may be particularly likely to commit cyber-crimes, as has been reported in some high profile individual cases (e.g., Gary McKinnon, Kennedy, 2012; and Lauri Love, BBC, 2015). Payne et al. (2019) differentiated between cyber-enabled crime, where crimes do not necessarily require on-line activity, such as fraud, and cyber-dependent crime, where on-line activity is necessary, such as hacking and spreading malware. They examined cyber skills, cyber-dependent crime and autistic traits on the Autism Spectrum Quotient (Baron-Cohen et al., 2001) in 290 participants whom they had recruited from computer science programmes, only 23 of whom reported having autistic diagnoses. Of these 290 participants, 122 individuals reported 333 cyber-dependent crimes, none of which had been prosecuted. Those who had carried out cyber-dependent criminal activity had higher levels of digital skills and higher Autism Quotients, than those who had not but fewer of them actually had an autism diagnosis. The authors concluded that about 40% of the association between autistic-like traits and cyber-dependent crime was mediated by advanced digital skills.

Recently, the potential vulnerability to radicalisation in individuals with autism has been explored (Allely & Faccini, 2018; Faccini & Allely, 2017), including discussions of the case of “RXG”, a British adolescent who used the internet to incite acts of violence against Muslims in Australia on Anzac Day (Dodd, 2021). Individuals with autism are purported to be vulnerable to radicalisation due to their similarities to those typified by the ‘lone wolf’ terrorist who ‘buys’ into a cause (e.g., white supremacy), perhaps due to political and personal grievances for which they find support online, but they rarely have any direct or personal links or contact with organisation members or leaders. ‘Lone Wolf terrorists’ are typically socially isolated, with poor social skills, and limited social networks (Hamm & Spaaij, 2017). Very few cases involving autism and terrorism have been identified but they include impersonal threats and acts of terrorism, e.g., calling in a bomb threat to an airport, rather than being specific to those with a particular religious or political agenda (Faccini, 2010). In relation to potential treatment, there is currently no research exploring the impact of programmes such as Prevent, a UK nationwide school/policy programme, or other de-radicalisation or terrorism prevention interventions for individuals with autism.

Finally, as regards characteristics, people with autism seem likely to struggle with understanding the CJS and might have been expected to have vulnerabilities in the CJS, which disadvantage them. In fact, it seems that they are not particularly suggestible in interrogative situations (Maras, & Bowler, 2012a; North et al., 2008), though they may be more compliant than other people (North et al., 2008), and may not perform well in cognitive interviews (Maras, & Bowler, 2012b). There has been very little research in this field however so it is not possible to say, for example, whether they understand their rights adequately.

Current Research on Interventions for Autistic Offenders

As mentioned, the literature regarding treatment for offenders with autism contains limited empirical data and no systematic or controlled trials. Whether in prison, inpatient psychiatric care, including specialist intellectual and developmental disability, or general mental health services, community care, or on probation, there is a gap in specifying precisely how the needs of autistic offenders should be met, in both legislation and social policy. For example, opportunities for early intervention are frequently evident in case studies and qualitative data (Griffin-Shelley, 2010; Ray et al., 2004), though such interventions have not often been provided. Similarly, service provision is not necessarily standardised and sufficient across local authorities and catchment areas resulting in treatment potentially being unavailable to certain individuals in some places, as a result of residential location or demand for services in that area (Melvin et al., 2020a, 2020b. A similar pattern is echoed in an examination of Adam Lanza’s case, with a history of contact and support from child development, mental health and social care services, identifying multiple instances where the appropriate levels of assessment, diagnosis, treatment and support were not recognised or provided (Sturmey, 2019).

Treatment for offenders can be controversial, requiring the balancing of the needs of the offender against public protection and opinions (Jones & Newburn, 2013; Tonry, 2007). Nevertheless, there are a range of approaches, incorporating multiple psychological disciplines and psychiatric medication.

Cognitive Behavioral Programmes

Cognitive behavioral therapy (CBT) in groups is currently considered best practice for most offenders (Joy Tong & Farrington, 2006; Lipsey et al., 2001; Lipsey et al., 2007; Lipton et al., 2002), with offence specific options for sexual offences, and arson, and generic programmes for anger, including for violence crimes, and problem-solving (Doley et al., 2015; Lindsay et al., 2007; Mann & Thornton, 1998; McGuire, 2005; Novaco, 1975). Recent systematic reviews have suggested that the best outcomes are achieved when the programme is manualised and delivered to groups of offenders, by a psychologist (Gannon et al., 2019).

CBT interventions are founded upon a relapse prevention addiction model and they strive not to completely eradicate risk (unlike early behavioral programmes such as aversive therapy and orgasmic reconditioning (Abel et al., 1976; Marquis, 1970; Marshall, 1971), but to enhance recognition of high risk situations and develop offenders’ skills to manage their own risk of re-offending. Such programmes typically incorporate psychoeducation, and address cognitive distortions, increasing victim empathy and developing a relapse prevention plan (Laws et al., 2000; Laws & Marshall, 2003; Marshall & Laws, 2003).

Some aspects of such treatment programmes, whether for arson, sexual offending, or violence, have to be questioned with regards to offenders with autism, due to their specific cognitive and behavioral profile, including the issues of the group delivery of the programme, the difficulties of increasing empathy, and shifting cognitive disorders, all of which may prove challenging for those with a condition characterised by cognitive inflexibility, social interaction and communication difficulties and low/fewer displays of empathy (Higgs & Carter, 2015; Murphy, 2010a; Murphy, 2010b; Woodbury-Smith & Dein, 2014).

Early research indicated CBT programmes were effective in reducing recidivism in non-disabled offenders, with reduction rates of re-offending of up to 86% (Dowden et al., 2001; Hanson et al., 2002). However, more recent reviews and analyses of data, including carefully controlled research, has suggested that such studies may have over-estimated treatment effects and across the literature, effect sizes and outcomes are not as high or consistent as would be anticipated (Gannon et al., 2019; Jolliffe & Farrington, 2007; Mews et al., 2017; Polaschek & Collie, 2004; Schmucker & Losel, 2008), with re-offending rates typically placed between 10 and 15% after 15 years for sexual offenders (Hanson & Bussiere, 1998) and reductions in recidivism of up to 36% for violence (Robinson, 1995). Very few of these studies however refer to offenders with autism.

Other Treatments

Other forms of talking therapy are available to address criminal behaviors, including dialectal behavioral therapy (so-called third wave CBT), psychodynamic or psychoanalytic therapies (Bianchini et al., 2019; Mulay et al., 2017; Trupin et al., 2002), however evidence regarding their use with individuals with intellectual and/or developmental disabilities who break the law, is very sparse (Beail, 2001; Beail et al., 2005; Brown et al., 2013; Morrissey & Ingamells, 2011).

Additionally, pharmacological treatments can be used for certain crime types, e.g. testosterone lowering medication for sexual behaviors, and antipsychotics or benzodiazepines for aggression (Turner et al., 2013; Turner & Briken, 2019) but the consistency of findings with regarding to positive treatment outcome for offenders with autism is limited (Kohn et al., 1998; Milton et al., 2002), with the former reporting reductions in behavior and the latter not). Additionally, the appropriateness and ethical use of such treatment for individuals with intellectual and developmental disabilities is contended (Sawyer et al., 2014).

A range of applied behavioral techniques and therapeutic programmes have also been employed with developmental disability populations, including those with autism, to address aggressive behaviour and/or inappropriate sexual behaviors (Davis et al., 2015). Typically, these are considered over cognitive behavioural programmes for individuals with poor cognitive and /or verbal skills. They can be used for those with moderate-to-severe or-profound intellectual disabilities and as such, the aggressive and/or sexualised behaviors targeted are referred to as ‘inappropriate’ or ‘challenging’ rather than framed as criminal due to a lack of mens rea and criminal intent due to severity of intellectual disability (Michael Doyle, 2004). So for inappropriate sexual behavior, for example, these are behaviors deemed unacceptable by social or legal standards (Ward et al., 2001), and include acts committed without abusive or harmful intent, e.g. touching of genitals in public or disrobing for sensory or self-pleasure, with a lack of awareness of others rather than disregard for others (i.e. it is not done for exhibition purposes but due to ignorance of social norms regarding privacy, etc.), as well as those classed as criminal offences.

Some applied behavioral programmes however, such as the ACHIEVE! Programme (Pritchard et al., 2018), have been used within specialist residential schools for those with intellectual and developmental disabilities, including autism, and demonstrated positive outcomes in reducing problem behaviour, including harmful sexual behaviours. However, effectiveness for specific behaviours e.g. reductions in sexualised behaviors compared to aggression are not presently known, nor is there information on effectiveness regarding treatment of individuals with intellectual disabilities alone compared to those with a co-morbid diagnosis of autism.

Applied Behaviour Analysis (ABA) and Positive Behaviour Support (PBS) has also been explored within forensic settings, with a recent systematic review by Collins et al. (accepted) examining the effectiveness of ABA and PBS in forensic settings, including services for those with intellectual and developmental disabilities. The review included twenty-nine articles and findings indicated that behavioural modification techniques and behavioural analysis can be implemented within forensic settings with some degree of success, with a single study reporting reductions of frequency of challenging behaviour, including inappropriate sexual behaviour, in an autistic adolescent (Collins et al. 2019). The systematic review identified several challenges to the implementation of PBS in forensic settings, including barriers to collaborative work, inconsistent practice and a lack of resources, however robust conclusions from the literature were limited by the methodological quality of studies identified.

Evidence on Interventions Overall

Therefore, the current evidence base regarding offender treatment for autistic individuals, adults and young people, consists mainly of case studies or reports (Faccini, 2010; Faccini & Allely, 2016; Kelbrick & Radley, 2013; Kohn et al., 1998; Murphy, 2010a; Murphy & Melvin, 2020), a handful of qualitative studies (Melvin et al., 2020a, 2020b, 2019; Payne et al., 2020) and uncontrolled quantitative designs (Heaton & Murphy, 2013; Langdon et al., 2013; Murphy et al., 2007; Murphy et al., 2010), and a number of narrative or systematic reviews (Higgs & Carter, 2015;Melvin et al., 2017 ; Schnitzer et al., 2020). There are very few studies and little long-term data exploring recidivism in this group over time (but see Heaton & Murphy, 2013). However, what has been shown in several studies is that autistic offenders display recidivist behaviors during treatment (Melvin et al., 2019; Ray et al., 2004; Sotsec, 2002) and after treatment, and as such may benefit from repeating the treatment (Melvin et al., 2020a, 2020b; Murphy & Melvin, 2020; Sturmey, 2019) or receiving booster sessions. Overall the literature, including systematic reviews, illustrate variability in responsivity to treatment in autistic offenders (Melvin et al., 2017; Schnitzer et al., 2020). Studies identified in the Melvin et al. review included different treatment approaches for adults and adolescents but mainly CBT, with a few utilising cognitive analytical therapy, family therapy and psychopharmacology. An assortment of offences were noted in the literature including theft, aggression, sexual offences, firearms offences and manslaughter (Kelbrick & Radley, 2013; Melvin et al., 2020a, 2020b; Murphy, 2010a; Murphy et al., 2010; Murphy & Melvin, 2020). The level of detail regarding treatment approach and responsivity varied within the studies and, for the most part, they were delivered alongside other treatment approaches, and many participants had undergone multiple rounds of therapy, with more than one technique, e.g. CBT and Cognitive Analytic Therapy (Kelbrick & Radley, 2013). There was variability within the findings with some reporting positive outcomes from a treatment approach, whilst others did not. For example, Murphy, 2010a used CBT with a young man with autism convicted of manslaughter and reported ‘minimal effect’, whereas CBT was included by Kelbrick and Radley and treatment was deemed ‘successful’. However, all case studies within the Melvin systematic review and wider literature, identify the need for changes or adaptations to therapy for offenders with autism. The later systematic review by Schnitzer et al. (2020) on autistic adolescent offenders only included an additional three studies to the Melvin et al. systematic review and echoed similar inconsistent results regarding positive treatment outcomes and need for adaptations. One of the issues may be the degree of autism in different individuals, or the different degrees of, say, cognitive inflexibility. As yet it is not possible to identify the important variables in terms of degrees of autism symptomatology and treatment outcome.

The aforementioned literature illustrates that autistic offenders are frequently treated alongside non-autistic offenders. This may take place in prison (Vinter et al., 2020; Young et al., 2018), in specialist intellectual and developmental disability mental health services and general mental health services (Lunsky et al., 2008; Palucka et al., 2012), including secure/or forensic services (Allely, 2018; Lindsay, 2013; Lindsay et al., 2010; Lindsay et al., 2014; Murphy & Mullens, 2017) and community services (Faccini, 2014; Lindsay et al., 2011; Lindsay et al., 2014) or probation (Lewis et al., 2015). The limited data available from these and the wider research literature makes few or no suggestions regarding increased opportunity or likelihood of positive treatment outcomes in different settings, i.e. prison or hospital. However, as with non-autistic offenders, treatment should be delivered in a specialist intellectual or developmental disability service using an adapted programme if the level of cognitive function and social adaptability require it. However, no studies have compared treatment outcomes between community mental health and secure forensic intellectual or disability services or prisons.

There are no validated autism-specific treatment programmes for criminal behavior, and autistic offenders may be included in groups for those without autism but with other intellectual or developmental disabilities (Taylor et al., 2016); Clare & Murphy, 1993; Langdon et al., 2013; Melvin et al., 2020a, 2020b), or without autism but with mental health issues (Milton et al., 2002; Murphy, 2010a). Some may be in specialist autism only services (Cervantes et al., 2019; Kuriakose et al., 2018) and in general mental health services, with most of the literature related to children and young people rather than adults. Again, there is as yet no literature or evidence supporting the use of one treatment method over another, bar reports of offender preference i.e. not explicitly linked to recidivism rates.

Future of Research and Practice for Autistic Offenders

What then can be drawn from the current knowledge base regarding autistic offenders to inform research, practice and policy. Although this chapter has highlighted a dearth of empirical evidence, there is some evidence. The knowledge base is developing, and we can draw from other areas of practice to guide and shape treatment for offenders with autism, while more controlled research and robust findings are awaited.

There are multiple opportunities through the justice process, health, and social care systems where autistic offenders can benefit from informed practice, some of which are summarised in Tables 2 and 3, however in order for these findings to progress from guidance to best-practice, a full programme of research needs to be undertaken, with some aspects specific to autistic offenders, whilst others could provide wider benefit to the those with autism, and for the treatment of offenders in general.

Table 2 Considerations for Autism in the Justice and Social Care Systems
Full size table
Table 3 Suggested Treatment Adaptations for Offenders with Autism
Full size table

Many countries have recognised that their CJSs treat those with disabilities unfairly (Bradley, 2009), yet, none have incorporated disability-fair practice into all their routines in police stations, prisons, and probation services. While there is evidence of some screening for ID being available in some countries, at some police stations, and prisons (Hayes et al., 2007; Mason & Murphy, 2002; Murphy et al., 1995; Murphy, Gardner & Freeman, 2017), yet, very little screening for autism has even been trialled in prisons (Fazio et al., 2012; Young et al., 2018), let alone rolled into routine practice, and there are no studies of screening for autism in police stations. It seems unlikely that criminal justice procedures will improve much until the criminal justice agencies are able to identify who needs specialist provisions. In the UK, Liaison and Diversion services, set up in the wake of the Bradley report, were intended to provide such services to criminal justice agencies but it seems that they are failing to identify all those who need their services. One barrier here is the strict confidentiality of personal records, and one possibility of course is that agencies could construct information-sharing protocols, so that information known to education or health services could be shared with criminal justice, with the person’s consent. This might obviate or at least reduce the need for screening for multiple conditions in criminal justice agencies. However, information sharing protocols have been suggested numerous times to those determining policy in England and Wales but so far have not come to fruition.

Second, criminal justice agencies need to ensure that their procedures can be understood by those with disabilities. It is clear, in the UK at least, that people with intellectual disabilities, including those with ID and autism have great difficulty understanding criminal justice processes, so that they find it difficult to understand what is going on in the police station and in court, and struggle to access visitor rights and health appointments in prison (Talbot, 2009). Some provision has been made for this, in England and Wales, such as Appropriate AdultsFootnote 2 in police stations and IntermediariesFootnote 3 in courts, but evidence suggests that both are insufficiently available (Cooper, 2012).

Third, educational and rehabilitative programmes that are designed for those with disabilities such as autism need to be widely available. Ideally these need to start in schools, so that programs teaching about discrimination, bullying, emotional regulation, and sex and relationships should be delivered to all children, and specifically designed for those with special needs such as autism. Some programmes of this kind are being trialled in the UK (Melvin, Langdon and Murphy, in prep; Brown & Murphy, submitted) but they are a long way from being routinely available and there is certainly no evidence bearing on whether they will prevent or reduce future difficulties. Similar programs need to be available to adults with autism and while they exist in some areas, there is patchy provision and a need for more research on outcomes.

Useful Resources