Abstract
The aim of this study was to examine the emotional states preceding and during sexual and violent offenses in a Dutch sample of male forensic inpatients. Moreover, the predictive impact of these emotional states on institutional violence in the first year of mandated care was examined using an incident scheme. Observer-ratings of emotional states by 103 male offenders and 97 sex offenders were examined using Mann-Whitney U tests. Using hierarchical multiple regression analyses, the predictive relationship between crime-related emotional states and incidents was examined. Sexual and violent crimes were equally preceded by painful emotions, primarily feelings of abandonment. During violent crimes, a state of bully and attack was dominant whereas sexual crimes were also characterized by self-aggrandizement and manipulation. These emotional states were not predictive for institutional violence. This study emphasizes the importance of emotional states in offending behavior and usefulness of schema therapy’s crime theory.
The cognitive behavioral model is one of the most widely applied and researched frameworks for understanding mental health and maladaptive behavior, including violence (Craske, 2010). A simplified take on this model is that people’s emotions and behaviors are influenced by cognitive representations of their environment. Other approaches, such as cognitive neuroscience or social psychology focus on cognitive and neural basis of mental processes that are linked to violent actions (e.g., emotions and cognitions emerge from interactions of brain networks), or allow to get a better understanding at what causes people to behave in a certain way in certain social situations. In the aforementioned models or approaches, emotions are an important factor. For example, emotions are a defining aspect of the human condition (van Kleef et al., 2016), and the emotions one feels can also compel to take action and guide how to interpret behavior displayed by others (Blagden et al., 2017; Svensson et al., 2013).
Problems in emotional functioning are associated with interpersonal violence and offending behavior (Byrne et al, 2016; Roberton et al., 2012, 2014). For example, individuals who sexually offend are often characterized by negative emotional states, such as loneliness, anxiety, and anger, but also by positive emotional states such as pleasure by sexual arousal (Gillespie et al., 2012; Gillespie & Beech, 2016). Emotional states are also associated with aggression and violence more generally (Novaco, 2016; Roberton et al., 2014; Velotti et al., 2017). For example, an inability to handle negative emotions has often been associated with committing several types of crimes (Gillespie & Beech, 2016; Novaco, 2016). Also, a sense of entitlement and instrumental aggression is associated with sexual coercion, aggression, and offending in general (Arbanas et al., 2022; Blinkhorn et al., 2016; Hepper et al., 2014; Krusemark et al., 2018).
Several treatments specifically target emotions (among other things) in offender populations, such as dialectical behavior therapy, cognitive analytic therapy, emotion-focused therapy, and schema therapy. Schema therapy is a commonly used approach for patients who receive mandated care in The Netherlands because they have committed a violent or sexual crime. Schema therapy focuses on changing ingrained maladaptive emotional states, also referred to as schema modes. A schema mode is a state-like concept that represents someone’s emotions as well as cognitions and behavior at a particular time so these states can fluctuate from moment-to-moment. These states are the result of a reciprocal relationship between childhood experiences, predisposing factors, and triggers that activate (mal)adaptive schema modes. Typically, various maladaptive schema mode domains are dominant in forensic patients (Keulen-de Vos & Bernstein, 2017; Keulen-de Vos et al., 2016, 2017). Child modes refer to feeling, thinking, and acting in a “child-like” manner; in such a mode, a patient feels similar to when s/he was a child when basic emotional needs (i.e., secure attachment and autonomy) have been neglected or frustrated. Avoidant schema modes involve attempts to protect oneself from pain by means of avoiding, for example by soothing oneself through substance use. Overcompensatory schema modes refer to extreme attempts to overcompensate painful feelings (i.e., child modes). Table 1 presents more information for dominant schema modes in forensic patients 1 .
Descriptions of the Schema Modes in This Study.
According to schema therapy’s crime theory (Bernstein et al., 2007; Keulen-de Vos & Bernstein, 2017), the events leading up to and culminating in violent and criminal behavior can be explained by a sequence of schema modes. The events preceding this behavior are often initiated by painful emotional triggers (i.e., child modes), for example situation in which someone feels abandoned, lonely, or hurt. These modes are triggered and subsequently covered up, avoided, or overcompensated (Keulen-de Vos & Bernstein, 2017). A recent study by Keulen-de Vos et al. (2016) supported this crime theory. In a sample of 96 forensic inpatients, schema modes were assessed based on descriptions of patients’ crimes and the events leading up to their crime. Criminal behavior was often preceded by frustration and painful feelings (e.g., loneliness, abandonment, and vulnerability, “child modes”) whereas criminal behavior itself was typically characterized by excessive states of anger involving threats and intimidation (i.e., overcompensatory modes). For example, a girlfriend breaks off the relationship with a male individual. This initially triggers feelings of abandonment and vulnerability (i.e., vulnerable child mode), and is followed by anger (i.e., angry child mode), and drug use to numb these feelings (i.e., detached self-soother mode). This subsequently leads to anger and impulsivity (i.e., impulsive and angry child modes), culminating in stalking and threatening behavior (i.e., bully and attack mode). Furthermore, schema modes moderately predicted later institutional transgressions: feelings of vulnerability (i.e., vulnerable child mode) and anger (i.e., angry child mode) experienced in the events leading up to criminal behavior and overcompensatory schema modes during the crime were predictive for the degree of physically aggressive behavior in the first year of admission to forensic mandated care (Keulen-de Vos et al., 2017). Keulen-de Vos et al. (2016) was the first to assess schema modes related to criminal behavior but other studies examined the relationship between schema modes and aggression, and between schema modes and recidivism risk (Bernstein et al., 2023; Clercx et al., 2021; Dunne et al., 2018; Lewis et al., 2021). For example, Lewis et al. (2021) reported that both self- and observer-rated child modes and coping modes were related to aggression in a sample of 59 male forensic psychiatric inpatients. Clercx et al. (2021) examined whether schema modes predict short-term violence risk in a sample of 103 male forensic patients in mandated clinical care. Short-term violence risk, as assessed by the Short-Term Assessment of Risk and Treatability (START; Webster et al., 2009), was not influenced by either healthy or maladaptive schema modes. Healthy modes but not maladaptive schema modes were predictive for protective factors (i.e., START strengths; Clercx et al., 2021). These studies show tentative support for schema therapy’s crime theory but they do not distinguish between types of violent offenses, such as violent versus sexual crimes although research show differences in characteristic between types of offenders and their offense trajectories (Cale et al., 2016; Garofalo et al., 2018; Gillespie et al., 2018).
The Current Study
Few studies have compared emotional states prior and during the crime while understanding that potential differences may have implications for prevention, assessment, and intervention. In the current study, we retrospectively examined schema modes prior and during sexual and violent crimes committed by Dutch male 2 patients who received mandated inpatient care. We chose to operationalize emotional states as “schema modes” because of schema therapy’s crime theory and because schema therapy is a common therapeutic approach in forensic inpatient settings in the Netherlands. We aimed to replicate the study by Keulen-de Vos et al. (2017) who examined both schema modes related to criminal behavior and the events preceding criminal behavior, and whether these modes were predictive for institutional violence. Therefore, the aims of our study were to (1) compare emotional states in sexual and violent offenses, observed in the events leading up to the crime and those observed during the crime, rated from crime descriptions in patients’ charts; and (2) examine the relationship between these emotional states and institutional violence, assessed prospectively using an incident classification system in the first year of mandated treatment as institutional transgressions are common in forensic inpatients (Dexter & Vitacco, 2020; Jeandarme et al., 2019). Also, behaviors, beliefs, and emotional states in the lead up of the offense may be similar to current behavior (i.e., institutional behavior) because it may be driven by the same process (offense paralleling behavior framework; Daffern et al., 2010).
Based on the literature reviewed, we hypothesized, with regard to events leading up to the crime, no group differences with regard to vulnerable, angry, and impulsive child but different levels of lonely child between offenders. With regard to the crimes itself and in line with previous studies (Dunne et al., 2018; Keulen-de Vos et al., 2016) and schema therapy’s crime theory, we hypothesized no group differences in angry child mode, self-aggrandizer, and predator mode but more bully and attack and overcontroller mode in violent offenders. For example, perceived hostile intentions of others has been linked to increased rates violent crimes (Bratton et al., 2017; Coid et al., 2016; Wood & Dennard, 2017). Both in the events leading up to crimes and during the crimes, we expected a difference in conning and manipulation mode in sex offenders as narcissism or entitlement and instrumental aggression is associated with sexual coercion (Arbanas et al., 2022; Blinkhorn et al., 2016; Hepper et al., 2014; Krusemark et al., 2018) and because of the process of grooming that some sex offenders employ to entrap or access victims (e.g., relationship forming, flattery, and pretending to be under age when using online chatrooms; Black et al., 2015; Burgess & Hartman, 2018). Both in events leading up to violent and sexual crimes and during the crimes itself, we expected no difference in intensity of detached self-soother mode because substance abuse has often been associated with committing several types of crimes (Kraanen & Emmelkamp, 2011; Novaco, 2016; Ogloff et al., 2015).
Finally, with regard to institutional violence, we hypothesized that schema modes rated during the events leading up to the violent and sexual crime and during the crimes itself would predict institutional violence in the first year of mandatory inpatient treatment as we expected more transgression in the first year due to adjustment difficulties We did not anticipate differences between offenses because, according to schema therapy, schema modes represent risk factors for future antisocial behavior in general (Keulen-de Vos et al., 2017), and is consistent with research indicating that prior criminal behavior and emotional deficits are predictors of institutional misconduct (Douglas et al., 2014; Keulen-de Vos et al., 2016).
Method
Design
This study had a retrospective archival research design. We included male patients who had exclusive sexual contact (non-fatal) offenses and exclusive non-sexual violent offenses. We excluded patients who were convicted for:
- non-contact sexual offences (e.g., possession/distribution of child sexual exploitation materials, exhibitionistic behavior, and voyeurism; N = 10);
- combination of sexual contact and non-contact offenses (N = 33);
- sexual offenses against non-human victims (i.e., animals; N = 3);
- sexual and violent offenses (N = 55);
- violent offenses with a sexual component (e.g., assault and threatening to rape the victim; N = 29).
Procedure
The study was approved by the Ethical Committee of Maastricht University’s Faculty of Psychology and Neuroscience in The Netherlands (ECP-110). For each forensic patient, information on the events leading up to the crime, and a description of the crime itself were extracted from the hospital record. In the first year of admission to the forensic hospital, a clinical psychologist drafts a document describing the scenario of the crime and the events preceding them. The scenario is based on statements given by the patient as well as police reports that included victims’ and witnesses’ statements. This scenario document was used for our assessments.
Setting and Participants
This study was conducted at forensic psychiatric center de Rooyse Wissel in Venray, the Netherlands. The hospital admits male patients who receive treatment on behalf of the state (in Dutch referred to as “TBS”; van Marle, 2002). There are three conditions for instating a TBS measure in the Netherlands: the crime committed carries a maximum prison sentence of a minimum of 4 years, criminal responsibility was diminished due to mental illness or personality disorder, and there is a high risk of reoffending—assessed through the use of a validated risk assessment tool (van Marle, 2002). The length of mandated treatment is renewed or terminated by the criminal court every 1 or 2 years based on the violence risk assessment and the proportionality of the duration of treatment related to the crime. The average length of treatment in the hospital is 7.8 years.
An a priori power analysis (G*Power 3.1.9.7; Faul et al., 2007) indicated a required total sample of 86 cases per group to detect medium effects (d = .50) with 90% power with a two-tailed alpha set at .05. In this study, case files of 200 forensic male patients were assessed: 97 patients were admitted because of a sexual offense (e.g., rape and child sexual abuse), and 103 patients because of a non-sexual violent offense (e.g., intimate partner violence, manslaughter, assault, and murder).
The mean age of the participants at the time of admission to hospital was 36.9 years (SD = 9.5; range 20–64 years). The most frequent mental disorders, as assessed with the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I; First, et al., 1997) and the Structured Interview for DSM Personality-IV (SIDPIV; Pfohl et al., 1995), were substance related disorders (63.5%) and mood disorders (17%), which were more prevalent in the violent offense group than the sexual offense group (χ2(1) = 5.976, p = .015 and (χ2(1) = .15.963, p < .001, respectively). The most common personality disorders were antisocial PD (40%), followed by narcissistic PD (17%) and borderline PD (15.5%). Antisocial PD was more prevalent in the violent offense group (χ2(1) = 10.218, p = .006); there were no group differences for borderline and narcissistic PD. The average psychopathy score, as assessed with the Psychopathy Checklist-Revised (Hare, 2003) was 22.7 (SD = 7.1); there were no differences between offense groups.
Materials
Detailed diagnoses and demographic information were extracted from patients’ file charts. Diagnoses were based on clinical and/or semi-structured interview diagnoses made by certified psychiatrists and clinical psychologists. Scores in our dataset were already part of the patient file and were copied to a data file.
Emotional states
We assessed emotional states with the Mode Observation Scale (MOS; Bernstein et al., 2009), which is an observation-based instrument that measures the intensity of emotional states in forensic patients. It assesses 18 schema modes on a 5-point scale (i.e., from “1” = absent to “5” = extremely intense). The MOS is typically used in clinical situations, such as therapy sessions (van den Broek et al., 2011, 2021) but it can also be used for file review (Keulen-de Vos et al., 2016). We rated both the file descriptions of the events leading up to the crime and the description of the crime itself for the intensity of emotional states. We examined 10 emotional states that are common in forensic patients (Keulen-de Vos et al., 2016; Keulen-de Vos & Bernstein, 2017). See Table 1 for a description of these schema modes. Previous studies show adequate to good inter-rater reliability based on file review, with inter-rater agreement (ICC) ranging from .65 to .99 (Keulen-de Vos et al., 2016). In our study, records of 77 patients (38.5%) were rated by two raters for the purpose of interrater reliability analysis; all other 123 records were rated by a single rater. For the events leading up to the crimes, the ICCs ranged from .79 to .99 for the schema modes, whereas for the crime descriptions, the ICC ranged from .74 to .99.
Institutional violence
Information on institutional violence was obtained from clinical data, such as daily bulletins and incident reports that were registered by hospital staff in the first year after admission. These incidents were observed by staff and registered in the patients’ files but no criminal charges were filed. Institutional violence was registered according to four incident categories: verbal aggression, threats, physical aggression, and violation of hospital rules, which is similar as the scheme used by de Vogel (2005). Earlier research has shown excellent inter-rater agreement in a sample of a hundred incidents (observed agreement = 92%; Hildebrand et al., 2004). In our study, the data were based on clinical data, therefore no inter-rater agreement information was available. We created a total score per category and an overall total score (i.e., total number of incidents).
Analyses
All analyses were conducted using the Statistical Package for the Social Sciences, version 28. The analyses were specified prior to data collection and no cases were excluded from our analyses. For the first study aim, we used Mann-Whitney U tests with a two-tailed alpha of .05 to compare emotional states preceding and during the crime in the group of patients with sexual offenses and the group of patients with violent offenses. We corrected our alpha for multiple comparisons according to the false discovery rate correction for 20 tests (10 Emotions × 2 Situations), using a p-value of p < .01390 (see Narum, 2006, p. 787). To test the second aim to examine the predictive value of emotional states on institutional violence, we used separate multivariate hierarchical linear regression analyses, with age as a controlling variable (block 1) and the schema modes as independent variables (block 2) and the total number of the incident types as dependent variable (e.g., total number of verbal aggression, verbal threat, physical violence, violation of hospital rules, and total number of incidents). We controlled for age as several studies show a trend that as the age at admission decreases, the severity of incidents and violence increases (Verstegen et al., 2017). These regression analyses were repeated for both the emotional states that relate to the events leading up to the crimes and the crime-related schema modes for both the group of patients with sexual offenses and the group of patients with violent offenses. Because our study variables had a non-normal distribution, as assessed by Shapiro–Wilk’s test (p < .05), we performed bootstrapping (10,000 samples) when conducting regression analyses (Russell & Dean, 2000).
Results
Differences in Emotional States
Table 2 presents the mean scores and standard deviations of the schema modes rating in the events leading up to the crime and during the crime, as rated to the group of patients with sexual offenses and the group of patients with violent offenses. Mann-Whitney U tests showed that during the events leading up to the crime, there were no differences on vulnerable child (U = 4,289, z = −.940, p = .347), angry child (U = 3,915, z = −1.955, p = .051), impulsive child mode (U = 4,882, z = .716, p = .474), and lonely child mode (U = 4,924, z = 1.080, p = .280) between offender groups. Patients with violent offenses showed significant more detached self-soother mode in the events leading up to the crime compared to patients with sexual offenses (U = 3,764, z = −2.550, p = .011). In the group of patients with sexual offenses, the crimes were preceded by more conning and manipulation (U = 5,942, z = 3.886, p < .001) than in the group of patients with violent offenses.
Sample Characteristics Related to Schemas Modes Prior to and During the Crime.
Note. n = 97 sex offenses and n = 103 violent offenses.
In the violent offender group, the crime itself showed higher intensity of angry child mode than the crime of the sexual offense group (U = 3,322, z = −4.200, p < .001) whereas sexual offenses in the sexual offense group were characterized by higher scores on self-aggrandizer (U = 6,047, z = 3.078, p = .002). There were no differences between patient with violent and sexual offenses with regard to detached self-soother mode (U = 5,114, z = 0.429, p = .668). In both offender groups, sexual and volent offenses were equally characterized by bully and attack (U = 4,916, z = −0.201, p = .841) and predator mode (U = 5,242, z = 0.795, p = .474). In the patients with sexual offenses, the crimes were characterized by higher scores on conning and manipulation (U = 6,620, z = 5.180, p < .001) compared to the crimes of the patients with violent offenses, whereas in the latter patient group, violent offenses were characterized by higher intensity of paranoid overcontrol (U = 4,130, z = −2.560, p = .010).
Emotional states and institutional violence
In the group of patients with violent offenses, the results showed that emotional states prior to the crime were not predictive for verbal aggression (F (11, 81) = 1.855, p = .053), verbal threats (F (11, 81) = 0.728, p = .708), nor violation of hospital rules (F (11, 81) = 0.511, p = .891), or total number of incidents (F (11, 81) = 0.533, p = .875). Vulnerable child (B = −0.095, p < .001), angry child (B = 0.057, p = .043), and self-aggrandizer (B = −0.065, p = .043) were significant predictors, explaining 24% of the variance (R 2 unadjusted) in physical violence, F (11, 81) = 2.345, p = .015. However, the bootstrap for these coefficients indicated that these modes were no longer significant after bootstrapping. In the group of patients with violent offenses, the self-aggrandizer mode during the violent crime was a significant predictor (B = 0.418, p = .025), explaining 21.6% of the variance (R 2 unadjusted) in verbal aggression, F (11, 81) = 2.028, p = .036. However, after bootstrapping, self-aggrandizer was no longer a significant predictor (p = .054). Schema modes rated for the crimes of the patients in the violent offense groups were not significant predictors for verbal threat (F (11, 81) = 0.901, p = .543), violation of hospital rules (F (11, 81) = 0.796, p = .644), physical violence (F (11, 81) = 0.949, p = .499), or total number of incidents (F (11, 81) = 1.304, p = .237).
In the group of patients with sexual offenses, the results showed that emotional states prior to the sexual crime were not predictive for verbal aggression (F (11, 66) = 0.401, p = .951), violation of hospital rules (F (11, 61) = 1.165, p = .328), physical violence (F (11, 66) = 0.882, p = .562), or the total number of incidents (F (11, 66) = 1.075, p = .394). Angry child was a significant predictor (B = 0.117, p < .001), explaining 25% of the variance (R 2 unadjusted) in verbal threats, F (11, 66) = 2.012, p = .041, however, after bootstrapping, angry child was no longer a significant predictor (p = .098). Schema modes rated for the crimes of the patients in the sexual offense groups were not predictive for verbal aggression (F (11, 70) = 0.515, p = .887), violation of hospital rules (F (11, 70) = 0.360, p = .967), physical violence (F (11, 70) = 0.587, p = .833), or the total number of incidents (F (11, 70) = 0.268, p = .990). Self-aggrandizer (B = 0.096, p < .001) and conning and manipulation (B = −0.055, p = .014) were significant predictors, explaining 27% of the variance (R 2 unadjusted) in verbal threats, F (11, 70) = 2.371, p = .015. However, after bootstrapping, these modes were no longer a significant predictor (p = .195 and .173, respectively).
Discussion
In line with our first hypothesis and schema therapy’s crime theory, both sexual and violent crimes were preceded by painful emotions, such as sense of abandonment (“vulnerable child”), anger (“angry child”), and frustration (“impulsive child”). These findings are also in line with attachment theory. Attachment refers to the emotional bond between an infant and their caregiver. This bond is of critical importance because it affects social and emotional development; it promotes the ability to regulate and cope with painful emotions (Cassidy, 1994). For example, soothing behavior leads to a secure attachment, whereas separation and emotional deprivation will result in insecure attachment. Research suggests a link between insecure attachment and (sexual) violent behavior (Miller & Klockner, 2019; Yang & Perkins, 2021). For example, Simane-Vigante et al. (2018) reported that anxious-ambivalent attachment styles are common in a sample of 77 males with violent convictions. Also, Garofalo & Bogaerts (2019) showed that avoidant and anxious attachment dimensions were dominant in a sample of 84 perpetrators of child sexual abuse.
We hypothesized that a sense of loneliness (i.e., lonely child) would be more prominent in the events preceding sexual offenses but our study showed that the level of loneliness is similar in the onset of sexual and violent behavior. Loneliness is related to negative feelings during social interactions, less intimacy, problematic relationship skills and simply having relatively little social contact at all (Wielinga et al., 2021). Most studies report that the experience of loneliness is greater among men convicted for sexual offenses than men convicted for violent offenses (e.g., Martin & Tardiff, 2015; Nunes et al., 2012; Schulz et al., 2017). Our contradictory finding might be explained by the fact that the (perceived) lack of supportive social network is also considered a general risk factor for violence (Douglas & Skeem, 2005). Risk increases if they have few or no pro-social supports in their life and if they have little involvement in pro-social leisure and recreational pursuits. Furthermore, violent behavior in general may be elicited by social functions (Matson & Kozlowski, 2012). For example, individuals may be exhibited to gain attention or to ask for help, or may be an attempt to (temporarily) avoid or escape undesirable social demands, situations, or people.
Violent offenses were more often preceded by self-soothing behavior, such as drug and alcohol misuse (“detached self-soother mode”), than sexual crimes. During both types of crimes, substance use played an important role. These findings are largely in line with the literature. The link between substance use and involvement in criminal behavior has been well-established (Falk et al., 2014; Håkansson & Jesionowska, 2018). For example, Kopak et al. (2014) reported significant associations between drug and alcohol use and involvement in violent crimes and Torok et al. (2012) showed that offenders with a history of violence and consequent drug use tended to adopt more severe criminal conduct. Also, substance use is included as a central risk factor in the RNR model (Andrews & Bonta, 2017). We had expected no differences in the onset of both types of offenses. Results also showed high levels of conning and manipulation mode during the events leading up to sexual offenses. Perhaps our sample of patients with sexual convictions engaged in deceitful and manipulative behavior (i.e., overcompensatory coping strategy) after painful emotions are triggered whereas patients with violent crimes engaged more in avoidant coping such as self-soothing behavior. The higher levels of manipulation and deceit prior to and during sexual offenses compared to violent offenses fit with typical grooming that some sex offenders employ to get close to a (underaged) victim. In fact, sexual grooming has been deemed an integral part of the child sexual abuse process (Elliott, 2017; Winters et al., 2020).
As hypothesized, violent offenses were characterized by intense overcontrolling mode whereas sexual crimes were more often described by a sense of grandiosity (“self-aggrandizer mode”) whereas both types of crimes were characterized by instrumental aggression (“predator mode”). These findings are in line with previous studies that show that narcissism or entitlement and instrumental aggression is associated with sexual coercion and offending in general (Arbanas et al., 2022; Krusemark et al., 2018). Also, paranoid states(“overcontroller mode”) are more typical for violent offenders than for those convicted for sexual offenses (Coid et al., 2016; Wood & Dennard, 2017). Unexpectedly, angry child mode was more common during violent offenses compared to sexual offenses. While anger is an antecedent for offending in general it is only a dominant emotional state during violent offenses. Indeed, the majority of treatment approaches with violent offenders are aimed at improving anger control and regulation (Roberton et al., 2014). Although emotion dysregulation is also considered an important risk factor for sexual aggression (Gillespie et al., 2018), perhaps sexual arousal instead of anger is a more dominant emotion during most sexual aggressive acts (Smid & Wever, 2019). Sexual offenses are an act of aggression but not necessarily an act driven by anger. Intimidation and aggression to get what one wants, however, does seem to be characteristic for both sexual and violent crimes, as both offenses showed similar levels of bully and attack mode. These findings are in line with Raghavan et al. (2014) who specified that threats of physical force, humiliation/intimidation, pressure, and bullying are common tactics of sexual coercion. Similarly, victims of domestic violence or other non-sexual violence typically experience intimidation and coercive control (Hart & Ostrov, 2020; Policastro & Finn, 2021).
Unexpectedly, none of the emotional states prior or during the violent or sexual offenses were predictive for institutional violence. Verbal abuse, verbal threats, physical aggression, and violation of hospital rules in the first year of mandated inpatient treatment was not predicted by crime-related emotional states. There may be dissimilarities between the context of the (events leading up to) criminal behavior and the institutional context. For example, within an institutional context, interpersonal violence may be triggered by rules that are imposed by staff. Also, because of the involuntary nature of their admission in secure hospitals, patients may perceive their treatment as coercive. Coercive treatment can arouse strong negative feelings. Patients may feel constraint of freedom and attacked in their feelings of autonomy and individual dignity, which may result in irritability and defensive aggressive behavior (Kurtz & Zavala, 2017; Levi et al., 2010). Criminal behavior, however, is often triggered by painful emotions (Keulen-de Vos et al., 2016), as was also the case in our study.
Limitations and Future Directions
The findings of this study should be considered in the light of certain limitations. First, our study has a retrospective design for assessing emotional states and we solely examined between group differences. Results might have been different if we had used other measures to assess emotional states, such as self-reports, physiological measures, or interviews with patients. A related factor is that we examined emotional states in a sample of Dutch male forensic inpatients. In the future, multiple kinds of measures should be used to evaluate patients’ emotional states and examine potential cross-cultural and gender differences across various settings (e.g., prison and community service). Furthermore, perpetrators of violent and sexual crimes are heterogeneous groups so future studies should examine within group differences (i.e., different types of violent or sexual offenses) and focus on difference between diagnostic categories (e.g., personality disorders, psychotic disorders, and intellectual disabilities). Moreover, our study did not distinguish between adult and child victims of sexual offenses, whereas all violent offenses were aimed at adult victims. Schema modes related to sexual violence may differ as a function of age of the victim. Future studies should examine this. Second, our study only showed which schema modes were characteristic for sexual and violent crimes. Our study design hampers the ability to conclude whether emotional states cause criminal behavior and whether these states are amendable to change during treatment. Emotional states are only one contributing factor to interpersonal violence. Other factors, such as cognitive distortions, neuropsychological factors (e.g., social information processing), and external factors (e.g., social support and employment) may also play a role. Future research should examine these factors and establish whether some factors mediate or moderate the relation with criminal behavior. Relatedly, other frameworks and approaches for understanding criminal behavior ought to be examined and studies should examine whether crime-related schema modes are similar to schema modes relating to institutional violence as this was not tested in our study.
Our study examined emotional states prior and during sexual and violent offenses in a sample of Dutch male forensic inpatients and examined the relationship between these emotional states and institutional violence. From a theoretical perspective, our findings add to the literature on schema therapy concepts in offending populations. From a clinical perspective, the study leads to a better understanding of which emotional states play a role in criminal behavior. Our findings inform therapists to explicitly address manipulative coping strategies in (pre)treatment programs for forensic patients with sexual offense histories. For example, prior to the start of treatment, forensic inpatients often have to undergo a program first that focuses on all aspects related to the crimes that they have committed. These pre-treatment offender programs usually explore why a particular patient committed a crime at a particular point in time, how and under which circumstances the patient committed this crime, and why the patient chose this particular victim. Based on our study, deceit and manipulation plays a central role in sexual offenses. There are no immediate clinical implications for dealing with institutional violence. Future studies should disentangle possible relationships between emotional states during treatment and institutional violence and possible mediating factors.
Footnotes
Acknowledgements
We thank the board of directors of Forensic Psychiatric Centre De Rooyse Wissel for permission for this study to be conducted.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
