Hormonal response to perceived emotional distress in incarcerated men with sexual sadism

Abstract

Sexual sadists derive pleasure from humiliation, domination and infliction of pain on victims. They display increased penile arousal and activation of brain regions involved in sexual arousal and emotional states when viewing stimuli depicting individuals in physical distress. Neuroactive hormones modulate these regions, but it is unknown if sexual sadists also have endocrine responses to depictions of individuals in distress. The present study examined endocrine responses, elicited by viewing a video depicting an individual in extreme emotional distress, in incarcerated adult male sexual offenders (n = 23) with varying levels of sadistic traits. Sadism, was measured by the Severe Sexual Sadism Scale (SeSaS). Testosterone (T), adrenocorticotropic hormone (ACTH), and oxytocin (OT) were assayed before and after participants watched a video depicting an individual in emotional distress. T responses to the video were significantly and positively associated with SeSaS scores. There were no significant associations between sexual sadism and OT or ACTH. Our findings provide physiological evidence of atypical processing of distress cues in sadism consistent with the role of testosterone in sexual arousal and aggressive behaviors. These findings have implications for the evaluation and treatment of sexual sadists.

Introduction

One out of four women and 10% of men will suffer a sexual assault in their lifetimes (Watts & Zimmerman, 2002; Welch & Mason, 2007). The effects of sexual assault include the acute and long-term suffering of victims (Cutajar et al., 2010; Macdowall et al., 2013), and the social costs of investigation and incarceration of offenders (Anderson, 2017; Eher et al., 2016). The treatment of sexual offenders has traditionally focused on recidivism prevention. Yet, a meta-analysis showed that prevention efforts have limited impact on sexual reoffending (Gannon, 2019). Continued efforts are made to improve treatment strategies (Mpofu et al., 2018). Treatment interventions can be improved by identifying the physiologic mechanisms that drive sexual offenses.

Sexual offenses with sadistic features are the most violent form of sexual assault. Sexual sadism is a paraphilia that involves sexual arousal to the suffering of another. In a sadistic sexual assault, the perpetrator may act to maximize the fear, humiliation, or physical damage incurred by the victim. Sexual sadism can also be expressed consensually with a masochist who controls what happens. Non-consensual sexual sadism is overall a costly disorder, both in terms of economic costs to society as well as the individual impact on victims. The underlying mechanisms are not well understood, particularly the biological basis (Mokros et al., 2019). Self-identified male consensual sadists have shown increased penile plesthymography (PPG) responses when viewing responses to cues of violence/injury compared to non-sadist (Seto et al., 2012). Neuroimaging analyses demonstrated that male sadistic sexual offenders had greater amygdala activity and amygdala-insula connectivity compared to non-sadists when viewing people in distress (Harenski et al., 2012). Related imaging studies found that rapists and other violent offenders had volumetric reductions and altered brain connectivity in the prefrontal, anterior temporal (including amygdala), and posterior temporal cortex, regions involved in decision making and emotional processing (Bertsch et al., 2013; Chen et al., 2016; Leutgeb et al., 2016).

Testosterone modulates connectivity between the ventromedial prefrontal cortex and the amygdala (Volman, 2011) generally reducing inhibitory responses (Radke et al., 2015). This is analogous to the modulatory effect on the amygdala and the prefrontal cortex of the hypothalamo-pituitary adrenal axis (HPA) via adrenocorticotropic hormone (ACTH) and cortisol. Both HPA and hypothalamo-pituitary gonadal activity are known to influence aggressive behaviors (Rosell & Siever, 2015). Indeed, testosterone (T) and cortisol interact to influence aggressive behaviors (Montoya, 2012). Contrary to general aggression, there are no consistent associations between basal T and sexual offending (Wong & Gravel, 2018). It is possible that T is related to certain types of sexual offending, or motivations . Sexual sadism is one motivational factor, distinct from other types of motivations (Knight & Prentky, 1990). In addition, basal hormone levels may be less informative than task elicited hormonal changes. For example, Weierstall et al. (2014) showed an increase in T in male participants listening to a violent story that was associated with subsequent attraction to violent pictures (e.g., depicting physical injury, mutilation, interpersonal violence, etc.).

Other hormones influence the prosocial-antisocial axis. Oxytocin (OT) generally increases prosocial behaviors, empathy, and sexual arousal (Barraza & Zak, 2009; Behnia et al., 2014;; Terris et al., 2018). Findings regarding aggression have been mixed (Lee et al., 2009; Ne'eman et al., 2016). While OT has neuromodulatory effects throughout the brain, the amygdala is a prominent region of activity by which social cues and empathy are enhanced (Bethlehem et al., 2013; Geng et al., 2018).

The hormonal correlates of aggressive behaviors in healthy adults is well-studied, but whether hormones are associated with sexual sadism has not been investigated. The present study examined if viewing a video depicting people in distress (Barraza & Zak, 2009) elicited endocrine responses in incarcerated sadistic sexual offenders. The literature narrowed our focus to three hormones, ACTH, T and OT. Note that we measure ACTH rather than cortisol because the stimulus video was too short to expect a change in cortisol following a previously established paradigm (Lin, 2013). Indeed, all three target hormones are associated with sexual arousal (Kruger, 2003). The atypical sexual arousal by sadists to depictions of individuals in distress led us to hypothesize that T would be the primary endocrine response we would observe.

Methods

Participants

Twenty-three adult male sexual offenders were recruited from a secure sex offender treatment facility in the Midwestern United States. All had been convicted of at least one sexual offense and 96% had committed multiple sexual offenses. Inclusion criteria were: 1) native English speaker, 2) IQ of 70 or higher, 3) 4th grade reading proficiency or higher, 4) no CNS disorder or major medical illness, 5) no history of psychotic disorder in self or first-degree relative, 6) no current Axis I mood or anxiety disorders.

All sex offenders had at least one victim over the age of 17 (96%) or 15 (100%). None had comorbid paraphilia disorders (e.g., pedophilia) as determined by evaluations conducted by psychologists/psychiatrists at the treatment facility.

Written informed consent was obtained from all participants after a description of study procedures. They were financially compensated at an hourly rate comparable to institutional payment for work assignments at the facility. All procedures were approved by the Institutional Review Boards of Claremont Graduate University and the University of New Mexico and carried out in accordance with the relevant guidelines and regulations.

Measures

Sexual Sadism

Sexual sadism was assessed using the Severe Sexual Sadism Scale (SeSaS) (Nitschke et al., 2009). The SeSaS is a file-based screening instrument of behavioral indicators of sexual sadism during sexual crimes. It includes 11 dichotomous (yes/no) items whose sum indicates the total sadism score (e.g., exertion of power, control, domination; degrading, humiliating behavior towards the victim). Using institutional files, trained raters scored each participant based on the descriptions of the sexual assaults they committed. Scoring was done independently by two raters with disagreements resolved by discussion. SeSaS scores were generated for all participants and exhibited acceptable internal reliability (α = 0.71). A score of 4 or higher is considered evidence of sexual sadism (Nitschke et al., 2009), with sensitivity and specificity of .95 and .99, respectively (Nitschke et al., 2013).

The Diagnostic and Statistical Manual of Mental Disorders (DSM) has been traditionally used in the assessment of sexual sadism. However, this categorical approach to sadism diagnosis is associated with low interrater reliability (Marshall et al., 2002; Marshall, & Kennedy, 2003). A growing body of literature also suggests that sadism is better understood as a dimension than a taxon (Longpre et al., 2018; Marshall, & Kennedy, 2003; Mokros et al., 2014). In the present study, we adopted this dimensional approach using the SeSaS as a unidimensional measure of sadism (Mokros et al., 2012).

Other Measures

Past and current DSM Axis I diagnoses were assessed by psychiatrists/psychologists in the prison facilities and obtained from institutional records. Of the six participants with sexual sadism, two had an additional diagnosis of paraphilia not other specified (NOS) and three had a diagnosis of antisocial personality disorder (ASPD). Of the 13 with paraphilia NOS, three met criteria for ASPD. Two participants met criteria for all three diagnoses. Intelligence was estimated using the matrix reasoning and vocabulary subtests of the Wechsler Adult Intelligence Scale (WAIS; Ryan et al., 1999; Wechsler, 1997) and administered by trained research assistants.

Psychopathy, which is prevalent among incarcerated populations and characterized by low empathy, was assessed by trained research assistants using the Psychopathy Checklist – Revised (PCL-R; Hare, 2003). The PCL-R is a reliable instrument for the assessment of psychopathy in incarcerated populations (Hare, 1980; Hare, 1996; Hart, 1989). It uses a semi-structured interview of school experiences, employment, relationships, family, and criminal activity, in addition to a review of the participant’s institutional records. It includes 20 items, each scored on a 3-point scale, that measure the personality and behavior characteristics of psychopathy. PCL-R scores range from 0-40. The PCL-R assessments were administered and scored by trained staff.

Emotional Video

While seated alone at a partitioned computer station, participants watched a 102-second video depicting a person in distress (Barraza & Zak, 2009). The video featured a father talking emotionally about his 2-year old son who has terminal brain cancer in a hospital setting. Participants rated their affective responses to the video immediately after viewing it using a list of 12 adjectives related to empathy (sympathetic, compassion, moved, tender, warm, softhearted) or personal distress (anxious, distressed, sad, annoyed, frightened, disturbed) on a 5 point Likert scale (Barraza & Zak, 2009). The video has been used in multiple studies and shown to consistently elicit high self-reported ratings of empathic concern, and OT and ACTH release (Barraza & Zak, 2009; Procyshyn et al., 2017; Procyshyn et al., 2020). Barraza and Zak (2009) also showed that a control video showing the same father and son at the zoo did not elicit these empathic and hormonal responses.

Hormones

Before and after watching the emotional video, a phlebotomist drew 30 mL of blood from an antecubital vein from all participants using whole blood and serum seperator tubes. The tubes were placed on ice to avoid analyte degradation and centrifuged at 1500 rpm for 12 min at 4°C. Plasma and serum were extracted, transferred to 2 ml microtubes, placed on dry ice and transferred to a −80°C freezer until analysis.

Assays were performed by the Reproductive Endocrine Clinical Laboratory at the University of Southern California in Los Angeles, California. OT was assayed using a radioimmunoassay (Peninsula Laboratories, San Carlos, CA) with an extraction step using current best practices(Szeto et al., 2011). Interassay coefficients of variation (CVs) were 13% at 2.78 pg/ml and 10% at 9.94 pg/ml (17 replicates). ACTH was assayed using an Immulite 2000 (Siemens Healthcare Diagnostics, Deerfield, IL), with interassay CVs of 8.2%, 8.7%, and 9.3% at 44, 89, and 229 pg/ml, respectively. T was quantified using a radioimmunoassay (Peninsula Laboratories, San Carlos, CA), with interassay CVs of 8%, 12%, and 12% at 13, 30, and 96 ng/dL, respectively. Basal levels of all hormones for participants in the study were within the normal range (Gardner & Shoback, 2017).

Statistical analysis

We first examined the associations between potential covariates, sadism, and basal hormones using correlation analysis (Spearman’s rho). The primary covariates were age, IQ, psychopathy score and hormonal medication status. As the video includes a father and son, we also performed a Mann-Whitney test to compare hormone changes between participants with and without children.

We then performed mixed linear regression analysis with SeSaS scores and time (i.e., pre/post video) as predictors, and hormone levels (T, OT and ACTH) as dependent variables. Finally, a linear regression assessed if SeSaS scores predicted emotional ratings (empathy and personal distress). We also performed a stepwise multiple regression with emotional ratings as predictors and any identified hormonal changes associated with SeSaS scores as dependent variables. All variables were checked for outliers prior to analysis; none were present.

Results

Descriptive statistics are presented in Table 1.

Table 1.

Sociodemographic and clinical characteristics of participants

	Mean	SD
Age	47.44	11.25
SeSaS score	2.09	2.04
IQ	100.26	15.42
PCL-R score	22.53	5.23
Hormones
Basal T (ng/mL)	5.45	2.15
Basal OT (pg/mL)	2.24	4.09
Basal ACTH (pg/mL)	17.60	13.76
Race	(%)
Caucasian	52.2	/
African American	26.1	/
American Indian	8.7	/
Other	12.8	/
DSM diagnoses	(%)
Sadism	26	/
Paraphilia NOS	57	/
ASPD	57	/
Hormonal medication status
Under treatment	17	/
No treatment	83	/
Alcohol use
None	61	/
Abuse	26	/
Dependence	13	/
Drug Use
None	74	/
Abuse	23	/
Dependence	0	/
Parenthood
Parent	52.2	/
Non-Parent	47.8	/

Sadism was not significantly correlated with basal hormone values (OT: r(21)=.15, p=.50; ACTH: r(21)=−.02, p=.93; T: r(21)=.11, p=.62). Similarly, IQ was not significantly associated with basal hormone values (OT: r(21)=.31, p=.19; ACTH: r(21)=−.21, p=.38, T: r(21)=.16, p=.53). Basal hormone values were not significantly associated with the hormonal medication status (OT: U=22, p=.22; T: U=37, p=.97, ACTH: U=20, p=.16). Age was not significantly associated with basal hormones (OT: r(21) =.21, p=.34; ACTH: r(21)=−.04, p=.85; T: r(21)=.17, p=.46) and hormonal changes (OT change: r(21)=.19, p=.38; T change: r(21)=.30, p=.17; ACTH change: r(21)=.24, p=.28). The change in hormones in response to the video did not vary for parents compared to nonparents (OT change: U=46, p=0.22; ACTH change: U=50, p=0.33; T change: U=58, p=0.62). The correlation between psychopathy (PCL-R) scores and SeSaS scores was not significant (r(21)=.15; p=0.49). PCL-R scores were also unrelated to basal hormone levels or changes (OT: r(21)=.26, p=.22; OT change: r(21)=.24, p=.28; ACTH: r(21)=.07, p]=.74; ACTH change: r(21)=−.11, p=.62 ; T: r(21)=−.01, p=.96; T change: r(21)=.41, p=.05). These covariants are therefore uninformative for our analyses.

Participants experienced an average 4.36% (SD= 11.25%) increase in T after viewing the video. They also experienced an average 39.37% (SD=105.23%) change in OT and a 17.11% (SD=24.13%) change in ACTH. The mixed linear regression analysis revealed a significant time (pre/post video) effect and a time*SeSaS score interaction for T level (Table 2). Sex offenders with higher SeSaS scores showed a significant increase in T after viewing the video (Figure 1). No significant time or SeSaS effects, or time*SeSaS interactions for ACTH and OT were found (Table 2).

Table 2.

Mixed linear regression analysis for testosterone, oxytocin, and adrenocorticotropic hormone by time

	Time		SeSaS total score		Time by SeSaS total score
	F(1,16)	P	F(1,16)	P	F(6,16)	p
T	9.89	.006	1.04	.44	3.98	.01
OT	0.83	.38	1.46	.26	0.78	.60
ACTH	2.48	.14	0.42	.85	1.24	.34

Figure1.

A. Sexual offenders had an average increase of 4.36% in testosterone while watching a distressing video (p<.01). B. The change in testosterone was positively related to SeSaS scores.

Participants' self-reported empathy and personal distress after viewing the video were not significantly associated with SeSaS scores (Personal distress: F(1,20)=0.80, p=.38; Empathy: F(1,20)=0.48, p=.50) . Nor were they significantly associated with T changes (Personal distress: t=0.92, p=.37; Empathy: t=0.26, p=.78).

Discussion

Social stimuli induce endocrine responses across animal species and humans (Ma et al., 2018; Zilioli & Bird, 2017). The effects of hormones on social behaviors depend on the socio-affective characteristic of individuals (DeWall, 2014) and the social context (Luo et al., 2017). In the present study, we found that sexual sadism was positively associated with increase in T level while watching a distressing video. This contrasts with prior research using the same paradigm in healthy controls, in which the emotional video elicited an increase in OT, an increase that was not seen when participants viewed a thematically similar non-emotional video (Barraza & Zak, 2009; Zak, 2015). Our finding is consistent with a study in adult male non-offenders showing that increased T in response to violent stories was positively associated with subsequent appetitive responses to cruel and violent cues (Weierstall et al., 2014). Here we demonstrate a direct association between T change and sadism in sexual offenders.

Prior studies have reported inconsistent results regarding associations between basal hormones, including T, OT, and ACTH, and sexual offending (Lee et al., 2009; Ne'eman et al, 2016; Rosell & Siever, 2015; Studer et al., 2005; Susman, 2006; van de Wiel et al., 2004; Wong & Gravel, 2018). It is noteworthy that sadism scores were not significantly correlated with basal OT, ACTH or T, suggesting that these basal hormones are not effective to identify sexual sadists. In contrast, sexual sadism was significantly associated with T change after viewing a distressing video. The increase of T in response to emotional distress in other may therefore be a viable sadism bioassay. Such objective measures may help supplement clinical evaluations, identify treatment needs, evaluate the effectiveness of clinical treatment, and potentially predict future behavior (e.g., likelihood of sexual recidivism). This is particularly important givaen that success of treatments for sadistic sexual offenders are difficult to evaluate (Kingston, 2008). The most typical include 1) behavioral history (e.g., recidivism), a generally poor measure since treatment success should be determined before releasing someone at risk for sexual reoffending and is also confounded by incarceration that truncates offense opportunities; 2) PPG responses to depictions of sadistic acts that can be affected by age, health, recalcitrance, and clinically insignificant responses; 3) self-report that has limited value in semi-adversarial settings.

Neuroimaging studies may shed additional light on the mechanisms producing atypical T responses in sadists. In sexual offenders, studies have shown volumetric reductions and altered brain connectivity in the prefrontal cortex and the amygdala (Bertsch et al., 2013; Chen et al., 2016; Leutgeb et al., 2016). One study found that sadists showed increased amygdala responses to images depicting individuals experiencing physical pain (Harenski et al., 2012). The T change observed in the present study might explain this amygdala activity increase. Indeed, a burgeoning literature implicates impaired inhibition of amygdala activity by the prefrontal cortex affecting aggression due to testosterone (Rosell & Siever, 2015). Volman et al. (2011) found that individuals with high levels of testosterone, showed reduced inhibition from the ventromedial prefrontal cortex on amygdala responses during the voluntary control of socially motivated behaviors. Whether observed T change in sexual sadism represents reduced inhibition of arousal/behavior or increased excitation/attraction remains to be established. Indeed, our findings indicate the need for additional neurobiological studies connecting testosterone, sadism, and the processing of distress in others.

The participants in this study were all convicted sexual offenders. Those scoring high on sadism were characterized not only by deviant sadistic arousal, but also by the willingness to non-consensually act on those arousal tendencies. Compared to, for example, consensual sexual sadism, there is an additional significant antisocial element. Nevertheless, whether the current findings extend to consensual sadism is a question for future research.

A limitation of our study is that all participants were sexual offenders committed to a secure treatment facility. They were relatively older and had resided in the facility for several years due to the nature of their crimes. Although there are no similar published studies in younger, larger effects might have been found if participants were tested temporally closer to their most recent sexual offense. Additional studies using larger sample sizes and matched control subjects would also help to confirm our findings.

Pain and distress cues expressed by the victims of sadistic sexual offenders are central to the arousal they experience. The present findings add to other studies showing that enhanced arousal to pain/distress in sadism is not limited to offenders’ victims, but to general depictions of suffering in others (Harenski et al., 2012; Seto et al., 2012). Increased T in this context may represent an objective atypical response to emotional distress cues in sadism that may be useful for assessment and treatment of this high-risk population.