A combination of combat experience, early abduction, and severe traumatization fuels appetitive aggression and violence among abductees of rebel war in Northern Uganda

Funding information Hector Fellow Academy; Studienstiftung des Deutschen Volkes Abstract Individuals who perpetrate violence may likely perceive violence as appealing and infliction of violence to derive pleasure is termed as appetitive aggression. Individuals who were abducted as children into an armed group often experience a higher number of traumatic event types, that is traumatic load and are usually socialized in a violence‐endorsing environment. This study aims to investigate the interaction between age at initial abduction with that of traumatic load, and their influence on appetitive aggression along with perpetration of violent acts by former members of an armed rebel group of both sexes. Semi‐structured interviews were conducted among a target group of formerly abducted rebel‐war survivors (including participants with and without combat experience) from Northern Uganda. Participants included 596 women and 570 men with N = 1,166 (Mage = 32.58, SDage = 9.76, range: 18–80 years). We conducted robust linear regression models to investigate the influence of age at initial abduction, traumatic load, combat experience, and biological sex on appetitive aggression as well as their perpetrated violent acts. Our study shows, appetitive aggression and the number of perpetrated violent acts were specifically increased in individuals who were abducted young, experienced several traumatic events in their lifetime, and with previous combat experience. For perpetrated violence men showed increased levels whereas for appetitive aggression the association was independent of biological sex. Therefore, early abducted individuals with a higher traumatic load, who have combat experience, need to be given special intervention to prevent any further violence.


| INTRODUCTION AND THEORETICAL BACKGROUND
For victims of conflict and terror, repeated exposure to traumatic experiences can lead to mental health disorders such as anxiety, depression, and posttraumatic stress disorder (PTSD; e.g., Ceri et al., 2016;Murthy & Lakshminarayana, 2006). Perpetrators who live in violent environments where aggression is socially accepted and even encouraged (e.g., child soldiers or combatants) often perceive perpetrating violent acts as enjoyable (Elbert, Schauer, & Moran, 2018) and this disposition is referred as appetitive aggression (Elbert, Weierstall, & Schauer, 2010;. In contrary to reactive aggression, as a spontaneous reaction to a perceived threat, and instrumental aggression, which is used to achieve a specific goal (e.g., Anderson & Bushman, 2002;Geen, 2001) whereas the purpose of appetitive aggression lies in the execution of the violent act itself and the perpetrators display feelings of fascination and positive emotional arousal . Elbert et al. (2018) recently described the development and mutual intensification of reactive aggression and appetitive aggression in a concept termed bi-cycle of violence. According to this concept, repeated exposure to threatening situations intensifies PTSD symptomatology, such as hyperarousal symptoms and angry outbursts, which in turn lowers the threshold for reactive aggressive behavior. However, for individuals who experience perpetration of violence as enjoyable, ongoing violent behavior increases their pleasure of harming others, and this intensifies their trait of appetitive aggression. These individuals' role shifts from victim to perpetrator, or to quote Karl Jaspers (1986): "Where violence is used, violence is awakened." In modern wars, male adolescents are particularly targeted for forceful recruitment as they are more suggestible than adults and physically stronger than younger children (Blattman & Annan, 2010).
Childhood and especially adolescence are critical periods for brain development and maturation-related hormonal changes (Keshavan, Giedd, Lau, Lewis, & Paus, 2014). As a result, fundamental changes in the social environment such as abduction into an armed group could cause effects that are more deleterious to individuals in critical maturation periods.
Initial evidence from Eastern Congo (Democratic Republic of the Congo) indicated that individuals recruited as children and adolescents are particularly vulnerable to perceiving violence more positively than those who joined the armed forces during adulthood (Hecker, Hermenau, Maedl, Elbert, & Schauer, 2012;Hermenau, Hecker, Maedl, Schauer, & Elbert, 2013;Köbach & Elbert, 2015;Weierstall, Haer, Banholzer, & Elbert, 2013). Weierstall, Haer, et al. (2013) assessed the age at which former Congolese combatants joined the armed group and found higher appetitive aggression among those who joined earlier in life. In accordance with the findings of Köbach and Elbert (2015), adolescents recruited at the age of 16-17 showed higher appetitive aggression than adolescents at a younger or older age.
In addition, underage abductees not only reported higher appetitive aggression, but also perpetrated more violent acts than those who joined military groups as adults Hermenau et al., 2013). Individuals who were recruited at a younger age possibly remain longer in the rebel group as they would not know a way to escape and also be fearful of the consequences of an unsuccessful escape attempt (Blattman & Annan, 2010). These individuals subsequently commit a higher number of violent acts.
Some rebel groups even prefer recruiting children, as socialization into the group is particularly successful at this age group and as individuals have not yet developed a stable identity, they can easily be transformed into loyal members (Vermeij, 2014). In such cases, socialization takes place in an environment that fosters and even explicitly encourages the use of violence . In fact, former child soldiers from Sierra Leone regarded the rebel group as a substitute family where more aggressive behavior was frequently rewarded with promotion as well as increased power over resources and people .
As initiation into the group is often accompanied by physical abuse, children and youths in armed groups are at high risk of experiencing maltreatment as well as traumatic events . Former child soldiers even experienced more violence than former adult recruits (Hermenau et al., 2013), and corporal punishment in childhood is linked to self-perpetrated aggressive acts in adulthood (King et al., 2018). Additionally, research with prison inmates showed that physical abuse during childhood predicted violent offenses (Wang et al., 2012). The number of experienced or witnessed traumatic event types (traumatic load) not only increases the risk for PTSD in a dose-dependent manner (Kolassa et al., 2010;Mollica, McInnes, Poole, & Tor, 1998;Neugebauer et al., 2009;Neuner et al., 2004), but also appears to promote appetitive aggression (Nandi, Crombach, Bambonye, Elbert, & Weierstall, 2015;Sommer et al., 2017). Similarly, the number of experienced traumatic events during military deployment had a dose-response effect on self-perpetrated violent offenses of soldiers (MacManus et al., 2013).
When it comes to armed groups, it is of importance to consider that not every member is involved in the fighting. According to the Paris Principles, the term child soldier refers to any minor who is associated with an armed group (United Nations International Children's Emergency Fund, 2007). As members can have a plethora of different duties, which can include fighting but also nonmilitary tasks like carrying looted goods, cooking, child care, or being used for sexual purposes. To date, there is a sparsity of research on mental health effects of belonging to an armed group for persons with military versus nonmilitary functions. Among the existent research is a study from Augsburger, Meyer-Parlapanis, Bambonye, Elbert, and Crombach (2015) in which women who were formerly associated with an armed group in Burundi were interviewed, to investigate group differences between individuals with combat experience and those who performed solely supportive tasks for the armed force.
The authors found significantly higher levels of traumatic load, perpetrated violence, and appetitive aggression in former combatants compared to former supporters of the armed group. However, it should be noted that the experiences for females differ substantially from that of males in rebel groups, which are characterized by a higher workload, lower ranks, sexual violence, and less implication in combat (Denov, 2008;Haer, 2019;Marks, 2017). Nevertheless, an increasing body of research describes that girl soldiers are also involved in combat actions (e.g., Haer, 2019;Vermeij, 2014) and sometimes even hold commanding positions (McKay, 2005). Nonetheless, to our knowledge research on sex differences among individuals with and without combat experiences in armed groups is sparse.
Even though studies on sex differences in relation to appetitive aggression have been conducted, findings are not yet conclusive.
While some previous investigations indicated lower appetitive aggression among females than males (Weierstall, Schaal, Schalinski, Dusingizemungu, & Elbert, 2011), other studies showed that women can perceive violent acts as equally appealing as men (Augsburger et al., 2017;Meyer-Parlapanis et al., 2016;Weierstall, Castellanos, Neuner, & Elbert, 2013). Augsburger et al. (2017) found a positive correlation between childhood maltreatment, traumatic load, and appetitive aggression in men. They also found, among women, childhood maltreatment was negatively related to appetitive aggression while traumatic load was not significantly linked to appetitive aggression and likewise, the link between perpetrated events and appetitive aggression was more pronounced among female participants in the study.
From 1986 to 2006, a brutal civil war was fought between the Ugandan governmental troops and the Lord's Resistance Army (LRA) in Northern Uganda. During this period, the LRA abducted about 50,000-75,000 civilians in Northern Uganda and more than half of them were children (Pham, Vick, & Stover, 2007). Among them, around a quarter of LRA members were girls (McKay, 2005). Due to the long duration of the LRA war against the Ugandan governmental troops and the high number of abducted civilians, a significant part of today's adult population in Northern Uganda has experienced extreme levels of traumatic events. This leaves them highly vulnerable to trauma-spectrum disorders (Ertl, Pfeiffer, Schauer-Kaiser, Elbert, & Neuner, 2014). Even though the LRA has moved to adjacent countries by now, members regularly escape and try to find their way back to the civil community. Studies from neighboring countries at war showed that particularly former child soldiers are at increased risk to abort reintegration programs and rejoin armed groups (Hermenau et al., 2013).
Based on the premises of the Northern Ugandan civil war and the aforementioned findings, we conducted the present study in a sample of former LRA abductees. In this study, we aimed to investigate the associations between traumatic load, age at abduction, and appetitive aggression/perpetrated violence in the context of biological sex and combat experience. Furthermore, with this study, we intended to contribute to a better understanding of predisposing factors which enables the development of appetitive aggression. And, this could help to identify individuals at risk as well as contribute to developing adequate therapeutic approaches to treat individuals' trauma symptoms along with addressing their appetitive aggression and violent acts. Individuals interested in participation were invited to schedule an appointment for a diagnostic interview. Before the study participation, written informed consent was obtained from the participants. In total, 1,815 persons aged 18 years and above were interviewed. Included study participants were at least once abducted by the LRA (N = 1,279). Of the excluded participants, 532 were never abducted and four had missing data regarding their history of abduction. Another participant was excluded due to doubts in the validity of the reported times of abduction. Additionally, eight participants had to be excluded for signs of current alcohol abuse as this may have affected the reliability of the person's responses. Furthermore, 104 participants were excluded from the analyses due to incomplete or nonvalid data obtained. Thus, the final study cohort comprised of N = 1,166 former LRA abductees (M age = 32.58, SD age = 9.76, range: 18-80 years), including 596 females (51.11%) and 570 males (48.89%). In total, 690 participants reported not having had previous combat experience, while 476 affirmed fighting actions. Out of these 476 individuals, 160 were women and 316 men. As displayed in Table 1, female and male participants differed significantly with respect to their age at initial abduction and the number of perpetrated violent acts. Female participants were initially abducted at a younger age and they perpetrated fewer violent acts while there were no differences in the levels of traumatic load and appetitive aggression.

| Materials and study procedure
The study procedures were approved by the Institutional Review  especially about trauma and PTSD as well as counseling and quantitative data collection. All diagnostic instruments were translated into Luo, the local language of Northern Uganda. To avoid any misinterpretation, it was back-translated and reviewed by trained interpreters before application in the interviews. Each interview lasted approximately 2 hr. Due to their remote residence location some participants were remunerated for their travel expenses.

| Sociodemographics, abduction history, and combat experience
During the interview, participants were asked for their sociodemographic information, for example, year and place of birth, age, and educational background. Also, the history of abductions by the LRA, including the year and duration of each abduction was assessed.
Participants' age at initial abduction was calculated based on their year of birth and the year of initial abduction. Furthermore, we included a question in regard of combat experience ("Have you ever been fighting in a combat?"), which was answered with either yes or no.

| Perpetrated violent acts
The number of perpetrated violent acts was assessed by a questionnaire including 17 predefined events (e.g., "Have you ever physically assaulted another person?") . The number of perpetrated violent acts (range 0-17) was used for subsequent data analyses.

| Appetitive aggression
Appetitive aggression was measured with the Appetitive Aggression Scale (AAS; . Participants responded to 15 questions (e.g., "Is it exciting for you if you make an opponent really suffer?") on a 5-point Likert scale from 0 ("Disagree") to 4 ("Agree"). For statistical analyses, the sum score of all 15 items was calculated (range 0-60; Cronbach's α = .89). The questionnaire has been validated among various populations of war-affected and combatexperienced individuals from East African and South American countries; it also showed good psychometric properties (e.g., Hecker et al., 2012;Nandi et al., 2015;.

| Traumatic load
Experience of traumatic event types was assessed with a 62-item event checklist that had been previously used in the Ugandan context (e.g., Wilker et al., 2014Wilker et al., , 2015. It assessed experiences related to war and T A B L E 1 Demographic comparison of females and males (18 items), whereas perpetrated event types (5 items) were excluded, as they were covered in more detail by another questionnaire. As previously shown, the sum score represents a precise and economic measure of trauma exposure (Conrad et al., 2017;Wilker et al., 2015).

| Data analysis
All statistical analyses were performed in R 3.6.3 (R Core

| Modeling of appetitive aggression
Bivariate correlations of all study variables are detailed in Table S1 for the total sample as well as for women and men separately. The results of the robust linear regression model for appetitive aggression by traumatic load, age at initial abduction, combat experience, biological sex, and the interaction of traumatic load and age at initial abduction are displayed in Table 2. The model explained 24% of the variance of appetitive aggression in this sample, F(5, 1160) = 227.51, p < .001. There were significant main effects for traumatic load, age at initial abduction, and combat experience in the model for appetitive aggression (see Table 2). The significant interaction effect of traumatic load with age at initial abduction further revealed that the increase in appetitive aggression related to traumatic experiences was stronger among individuals who were initially abducted at a younger age. Figure 1a and 1b display the results for individuals "with" versus "without" combat experience. There was no significant main effect of biological sex in this regression model.

| Modeling of perpetrated violent acts
A robust linear regression revealed that traumatic load, combat experience, and biological sex, as well as the interaction of traumatic load and age at initial abduction were associated with the number of perpetrated violent acts, F(5, 1160) = 1000.4, p < .001, R 2 = .44. There was no significant main effect for age at initial abduction (see Table 2). As displayed in Table 2, higher traumatic load, combat experience, and male sex were associated with a higher number of perpetrated violent acts. The significant interaction effect between traumatic load and age at initial abduction further indicated that individuals initially abducted earlier in life and who have experienced a high level of traumatic events also perpetrated a higher number of violent acts. Figures 2a and 2b display these associations separately for the subsample "with" and "without" combat experience.

| DISCUSSION
Our study investigated relationships between the age at initial abduction, traumatic load, appetitive aggression, and perpetrated violent acts, as well as sex differences, in a war-affected sample which included individuals "with" and "without" combat experience, the latter primarily involved in nonmilitary tasks for the LRA.

| Appetitive aggression
In line with our hypothesis, a combination of abduction at a younger age with more traumatic experiences was associated with higher levels of appetitive aggression. This corresponds with previous research, indicating higher appetitive aggression in individuals who joined military groups earlier in life Hermenau et al., 2013;Weierstall, Haer, et al., 2013). Entering an armed group in adolescence could play a crucial role for the development of high levels of appetitive aggression. Previous research found the ages from 16 to 17 to be a sensitive period to develop a robust trait of appetitive aggression (Köbach & Elbert, 2015). Adolescence is an important time for the development of the "social brain", for instance, to be able to handle increasingly complex social situations and relationships (Keshavan et al., 2014). The development of brain regions important for decoding mental states of others continues in late childhood until early adulthood (Keshavan et al., 2014), and this might contribute to perpetrating atrocities, assuming that children still lack the capacity to understand the adverse consequences of their actions. For adolescents, it might be difficult for them to adjust to life in a civil community after their return from an armed group due to a lack of adequate role models or the society to teach them social rules promoting peaceful coexistence. This could also contribute to their higher rate of voluntary return to armed groups than those of adult combatants (Hermenau et al., 2013).
On the systemic level, it must be noted that social norms and moral concepts develop among children in relation to their social interactions, perceptions, and interpretations of the same (Wainryb & Pasupathi, 2008). Research focusing on internally displaced children and youths in a violence-affected area of Bogota (Colombia) showed that even though the children had basic moral concepts (e.g., stealing or hurting others is wrong), this was not necessarily reflected on their expectations of others' behavior (Wainryb & Pasupathi, 2008). In this regard, the concept of in-and outgroup is crucial, F I G U R E 1 Jitterplots for robust linear regression models for appetitive aggression. The left figure (a) represents the data of participants with combat experience, whereas the right figure (b) displays the data of individuals without combat experience. Interactions were tested with continuous variables. For the purpose of displaying the results, conditional regression lines are displayed for low traumatic load (i.e., median minus 1SD; yellow) and high traumatic load (i.e., median plus 1 SD; red). Shading around the regression lines represents the 95% confidence intervals [Color figure can be viewed at wileyonlinelibrary.com] as violence against enemies was justified based on membership to an ingroup and the collective narrative of dehumanizing outgroup members (Wainryb & Pasupathi, 2008). This, in conjunction with the promotion to use violence could adversely impact adolescents as they have not yet developed a thorough commitment to norms of civil society. Their norms, values, and motivation to fight for the rebel group becomes intrinsic after socialization into armed groups (Vermeij, 2014). This combination makes them easily susceptible to overcome moral barriers of cruelty and even to enjoy the perpetration of violent acts (Elbert, Moran, & Schauer, 2017a).
Furthermore, living with the LRA, especially for recent abductees, is characterized by experiences of physical violence (Veale & Stavrou, 2003). As indicated by our results, this traumatic environment also impacts the levels of appetitive aggression. Similarly, in the context of prison inmates, it has been shown that there exists a positive correlation between appetitive aggression and physical peer violence experienced during childhood (Dudeck et al., 2016).
Our study results suggest that among the LRA abductees, individuals with combat experience showed higher levels of appetitive aggression than the ones without combat experience. This confirms the findings of a previous study with women formerly associated with combat in Burundi (Augsburger et al., 2015), however, in our study, we also included men. Interestingly, we could not detect any sex differences in our analysis for appetitive aggression. Importantly, in regard to appetitive aggression so far findings are inconclusive on sex differences (e.g., Augsburger et al., 2017;Hecker, Fetz, Ainamani, & Elbert, 2015;Meyer-Parlapanis et al., 2016;Weierstall, Castellanos, et al., 2013;. Therefore, our results are a valuable addition to the growing research body showing that there is no sex difference in experiencing positive feelings when perpetrating violence.

| Perpetrated violent acts
In addition, our results showed that the interaction of traumatic load and age at initial abduction is associated with an increase in perpetrated violent acts. Members of the LRA and particularly abductees are exposed to extreme physical and psychological adversities, these are well described in qualitative interviews. These narrations confirm that the first period after the abduction is characterized by being physically and psychologically tortured, being forced to witness murder, and sometimes even being forced to kill others (Veale & Stavrou, 2003). Especially children and adolescents may be suggestible to the commanders' constant threats of punishment (Veale & Stavrou, 2003), which makes the attempt to escape less likely.
Consequently, individuals abducted earlier in life remain longer in the rebel group and, thus, commit a higher number of violent acts. At the same time, elevated violent experiences are positively correlated with violence perpetration among former child soldiers (Hermenau et al., 2013). Building upon these findings, two studies among former Congolese (Hermenau et al., 2013) and Ugandan child soldiers (Crombach, Weierstall, Hecker, Schalinski, & Elbert, 2013) found positive correlations between the military rank, the number of selfcommitted offenses, and the fascination with violence. In contrast to adults, whose military rank depends primarily on their age and F I G U R E 2 Jitterplots for robust linear regression models for perpetrated violent acts. The left figure (a) represents the data of participants with combat experience, whereas the right figure (b) displays the data of individuals without combat experience. Interactions were tested with continuous variables. For the purpose of displaying the results, conditional regression lines are displayed for low traumatic load (i.e., median minus 1 SD; yellow) and high traumatic load (i.e., median plus 1 SD; red). Shading around the regression lines represents the 95% confidence intervals [Color figure can be viewed at wileyonlinelibrary.com] education, children were evaluated with regard to their identification with the rebel group (Crombach et al., 2013). Thus, a higher commitment, that is, a higher number of perpetrated acts of violence and fascination with violence, may help children to promote their position in the group's hierarchy, which allows them to regain at least some autonomy and control of their life .
Apart from armed groups, the association between childhood maltreatment or adverse childhood experiences and violence perpetration has been shown in several populations for instance in sexual offenders (Levenson & Socia, 2015), violent offenders (Wang et al., 2012), and perpetrators of intimate partner violence (Anda et al., 2006). Furthermore, as Elbert, Moran, and Schauer (2017b) described, there are overlapping features of appetitive aggression and psychopathy, and therefore it seems of importance to consider research of psychopathy in the scope of violent behavior. Prior research showed that psychopathy in early adolescences, along with elevated psychopathy in adulthood are associated with later offenses (Lynam, Miller, Vachon, Loeber, & Stouthamer-Loeber, 2009) and violent acts during lifetime (Lang, Af Klinteberg, & Alm, 2002).
Findings on the association between childhood victimization and psychopathy, on the other hand, are mixed, some research revealing elevated psychopathy scores for persons with a history of childhood abuse and/or neglect (Weiler & Widom, 1996) while others did not find such a relationship (Lang et al., 2002). Moreover, it has been shown that prison inmates with diagnosed psychopathy were more often sadistic than individuals not fulfilling a psychopathy diagnosis (Holt & Strack, 1999). In the forensic context, concepts of appetitive aggression and sadism have similarities. Paulhus, Curtis, and Jones (2018) in their dark tetrad framework described the sadistic personality as "unique among the Dark Tetrad in involving an appetite for cruelty". Furthermore, the sadistic facet is distinguished from the other members of the dark tetrad through the fact that individuals scoring high in the sadistic trait do inflict harm on others for pleasure (Paulhus, 2014), which it has in common with appetitive aggression.
Based on the aforementioned associations, it might be interesting to assess overlapping characteristics of sadism, psychopathy, appetitive aggression, and violence perpetration in future research.
Overall, our results indicate that individuals who were abducted earlier in life and who were exposed to a high number of traumatic  Nandi et al., 2015). Even though Augsburger et al. (2017) could not find any sex differences among former combatants in the context of committed violence, being a male did play a significant role in our regression model for perpetrated violent acts. This may be due to the fact that men are commonly to be found on the frontlines of combat in military groups (Zack-Williams, 2001), which provides more opportunities for violence, whereas women usually engage in parallel to combat also in noncombat tasks such as domestic work, child care, and they are also not deployed to combat during pregnancy (Baines, 2014).

| STRENGTHS AND LIMITATIONS
Our pioneer study shows appetitive aggression and perpetrated acts of violence were significantly modeled by the interaction of traumatic load, age at initial abduction, and combat experience. In our regression model, for appetitive aggression biological sex was not significant, whereas male sex was significant for perpetrated violent acts. To our knowledge this is the first study to investigate the interplay of traumatic load, age at initial abduction, combat experience, biological sex, along with their influence in modeling appetitive aggression and perpetrated violent acts. Thus, our study provides an indepth insight into the vicious cycle of violence perpetration in both women and men formerly associated with an armed rebel group.
However, the study is limited by its correlative-retrospective design; thus, the associations that were found do not describe causality but are of correlative nature. Longitudinal study designs can be utilized to further our knowledge of appetitive aggression and its causal relationship. In this context, it is relevant to study diverse populations with traumatic history (e.g., prisoners, refugees, war survivors, children in street situations). Furthermore, all variables of interest were assessed as retrospective self-reports, which might cause potential biases due to memory effects and social desirability.

| CONCLUSION AND FUTURE DIRECTIONS
Our study provides a unique contribution by identifying the possible links between traumatization at a young age, being forced to perform cruelties, and appetitive aggression. In future research, we recommend to include constructs of psychopathy and sadism, as well as individual risk factors such as self-control and behavioral disorders while investigating appetitive aggression. In addition, childhood maltreatment may lead to deteriorated cognitive functioning, including lower selfcontrol (e.g., Cowell, Cicchetti, Rogosch, & Toth, 2015;Kort-Butler, Tyler, & Melander, 2011), and also increases the risk for behavioral disorders (e.g., Dvir, Ford, Hill, & Frazier, 2014;Perepletchikova & Kaufman, 2010) therefore future research is needed to address those concepts in combination with appetitive aggression.
Among high-risk population and environment (e.g., post-war), it would be beneficial to implement, individual and community level aggression prevention programs for children as well as youth.
Inevitably, there is a huge need to further develop evidenced-based prevention programs.