WHY DO individuals commit acts of extreme violence or terrorism? According to Bandura, there are cognitive strategies of reality distortion that convert a violent or transgressive behavior into an acceptable one, making violence possible. These strategies, labeled as moral disengagement mechanisms (DMS), allow individuals to maintain their transgressive actions in concordance with morality. Such strategies operate on the cognitive and emotional level and focus on different aspects of moral situations, giving rise to eight disengagement mechanisms. In Bandura's view (1999, 2001, 2016), the disengagement mechanisms are at the root of violence and terrorism because they impede the correct functioning of the moral self-regulatory system, which maintains behavior aligned with moral standards. More precisely, the DMS hinder the startup of self-sanctions (self-condemnation and self-praise), one of the components of the moral self-regulatory system along with moral standards. Arousal of self-condemnation is prevented when agents justify an immoral behavior that clashes with moral standards, turning it into a necessary or valuable action. The incongruence between moral standards and behavior is thus solved using DMS. The operation of these mechanisms may be understood in the framework of dissonance theory, as we will argue later. Knowledge about the DMS, despite their importance in the emergence and maintenance of violence, is limited because there is sparse research on how violent agents justify their acts, which mechanisms they use, and even less, how they use them in their spontaneous language. This knowledge gap exists both in the studies of violence, including political violence, as well as in the studies of religiously motivated terrorism, for example, jihadism. According to ^{Borun (2011)}, despite the increasing number of publications on terrorism since 2001 and the interest in radicalization, empirical studies are rare.

In order to analyze the DMS spontaneously used to justify political violence and terrorism, the periodic reports published by the two most important Colombian illegal armed groups are an invaluable source. These groups are the leftist Revolutionary Armed Forces of Colombia (FARC, in its Spanish acronym) and the right-wing United Self-Defense Forces of Colombia (AUC, in its Spanish acronym), both inactive now. Their communiqués presented the groups' official thinking and contained justifications that presumably accomplished the role of being internalized by the group members, serving as mechanisms for disconnecting self-sanctions and avoiding or reducing social rejection. In this line, the proposed analysis goes further by addressing the operation of a collective moral disengagement within each group, which operates beyond the individual level, on a large scale social system or sub-system (^{Bandura, 2016}). Such collective disengagement has been overlooked in previous research.

The first goal of the present investigation is, then, to explore the DMS embedded in the groups' communiqués to identify which are used and how often. Results will contribute to our understanding of the psychological mechanisms behind extreme violence and to the theory of moral disengagement, and expand the findings concerning politically motivated violence and other forms of terrorism. The frequency in the usage of mechanisms will shed light on their effectiveness to disconnect self-sanctions and reduce dissonance. The second goal is to compare the two groups to establish whether the moral mechanisms are similar in their type and frequency, regardless of differences in the political ideology. Such similarity would support the assumption that the used mechanisms respond to a psychological need of disconnecting self-sanctions and alleviating distress, rather than to political ideology. The proposed study has the advantage of analyzing two violent and contemporary actors from the same country that were enemies to each other.

In the first section, we will present the classification of the DMS, their role as facilitators of immoral behavior and reducers of cognitive dissonance. Next, we will discuss their linkage to ideology and violence. In the second section we will describe the main characteristics of the two Colombian groups.

Moral Disengagement Mechanisms and their Relationship to Other Constructs

Moral Disengagement Mechanisms and Dissonance

Moral disengagement mechanisms are cognitive strategies for distorting reality, turning violent or transgressive behavior into an acceptable conduct (^{Bandura, 1986}, 1990, 1999, 2001, 2002, 2016; Bandura, Barbaranelli, Caprara, & Pastorelli, 1996). DMS may focus on: (a) reconstructing the behavior so that it is not viewed as immoral, (b) minimizing the agent's responsibility in causing harm, (c) trivializing the consequences that follow from the action, and (d) devaluing the victims or holding them responsible for their misfortune (Bandura, 2002, 2016). The four possible foci give rise to eight different mechanisms, as shown in Table 1.

Table 1 Description of Moral Disengagement Mechanisms 

Note: Adapted from ^{Bandura (1999}, 2002).

Disengagement mechanisms allow people to act immorally without damaging their self-esteem by selectively disconnecting self-sanctions from immoral conduct. Such self-sanctions are part of the moral self-regulatory system that keeps conduct in line with personal standards, the other component of the system. Consistency between acts and the moral standards, internalized and enforced throughout life, produces satisfaction. Inconsistencies, resulting from actions against the moral standards, i.e., immoral actions, produce blame or self-condemnation. The self-regulatory system works both before and after the behavior (Bandura, 1991) and so can DMS (^{Jackson & Gaertner, 2010}). People use DMS because they know the evil of their actions and need, therefore, to alter reality and make immoral actions acceptable, not censurable. In addition to DMS, other cognitive and social influences may also account for immoral actions. As Bandura (2002) signaled: "Moral actions are the product of the reciprocal interplay of cognitive, affective and social influences" (p. 102).

The mere arising of cognitive inconsistencies between behavior and personal attitudes or beliefs causes an uncomfortable state of "dissonance" (^{Festinger, 1968}) that motivates individuals to seek strategies to reduce or alleviate it. In the moral domain, such dissonance may emerge either as anticipated dissonance, before acting wrong, or as experienced dissonance, after an incorrect action (^{Barkan, Ayal, & Ariely, 2015}). Such uncomfortable state of cognitive dissonance may overlap with the anticipation of self-condemnation, increasing the need of reducing or alleviating it. To this purpose, the mechanisms of moral disengagement seem especially appropriate, as they distort reality, thereby reducing the dissonance, and disconnecting self-sanctions from immoral conduct. By the work of these mechanisms, immoral behavior is perceived as moral, given, for example, that it was elicited by the victim, deserved by her/him or less harmful than other behaviors (cf. Table 1).

^{Festinger (1968)} proposed three main modes to reduce dissonance: (1) modification of one or more elements in the dissonance, especially, attitudes or behavior; (2) addition of new consonant elements; (3) trivialization of some of the dissonant elements. A fourth mode of dissonance reduction is a denial of responsibility (^{Gosling, Denizeau, & Oberlé, 2006}). Accordingly, a strong similarity between these modes of dissonance reduction and DMS exists, with differences only in the mode (2) from Festinger and category (d) from Bandura. Moreover, ^{Ribaud and Eisner (2010)} considered that the concepts of moral disengagement, neutralization techniques (Sykes & Matza, 1957, quoted in Ribaud & Eisner, 2010) and self-serving cognitive distortions answer the same question of what moral individuals do to minimize cognitive dissonance, threats to self-concept, and experiences of moral self-sanction when facing moral transgressions. The finding of a conceptual and empirical overlap between the three concepts led the authors to group them as moral neutralization.

It follows from the discussion above that the theory of dissonance is a comprehensive framework for better understanding the moral disengagement mechanisms. Such theory provides a set of well-defined and experimentally tested propositions that could extend to DMS.

Moral Disengagement Mechanisms, Ideology, and Extreme Violence

Ideology, despite its relevance, has been subject of multiple definitions (^{Eagleton, 1997}; ^{Gerring, 1997}). We adopted the definition proposed by ^{Seliger (1976)}, who considered it as "Set of ideas by which men posit, explain and justify ends and means of organized social action and specifically political action..." (p.11). ^{Jost, Federico, and Napier (2009)} included attitudes and values in their notion of ideology, besides ideas or beliefs. Ideology serves both as a description of the world and as a prescription of how the world should be "... specifying acceptable means of attaining social, economic, and political ideals" (p. 309). That set of beliefs, values, and attitudes has cognitive, affective, and motivational properties, so that ideological commitment is a strong predictor of a range of attitudes, preferences, judgments and behaviors. Moreover, the mentioned authors regarded ideology as a system-justifying device for explaining or rationalizing how things are or should be. From the latter, it follows that ideology entails moral beliefs insofar it is a normative system that prescribes how the world should be. Ideology, nevertheless, is a bi-dimensional construct, comprised of social and economic dimensions (^{Feldman & Johnston, 2014}), giving rise to wide individual differences in the combination of the two dimensions but narrow ones between groups. The relationship between ideology and DMS has been investigated, as far as we know, only by ^{Jackson and Gaertner (2010)}. They tested whether two groups with different ideologies-right-wing authoritarianism (RWA) and social dominance orientation (sDo)-supported war as a political intervention and whether such support was mediated by moral disengagement mechanisms. They assumed a differential use of the mechanisms according to the ideology, an assumption disconfirmed by the data. In fact, both ideologies used all DMS, mostly minimization of consequences and moral justification. These findings highlight the importance of collecting more empirical evidence and comparing groups differing in ideology, as we will do in this research.

Extreme violence and terrorism may be facilitated by the usage of DMS (^{Bandura, 1990}, 1999, 2001; 2016; ^{Tsang, 2002}) or by altering own moral standards (^{Bersoff, 1999}; Moghaddam, 2005; ^{Staub, 1989}, 1996, 2000). In turn, ^{Maikovich (2005)} considered that terrorists, starting their career as such, experience a high need of reducing their dissonance. To alleviate it, they adopt the strategies of moral disengagement, diffusion ofresponsibility, just world bias (blaming the victim in Bandura's classification), social support and the preponderance of dream imagery over external reality. With the exception of the latter two, the remaining might be part of DMS. According to Maikovich, such strategies function both to reduce dissonance and gain social support. The process of indoctrination, on the other hand, includes not only a progressive approach to terrorism and to other inhumane actions, but also to moral justifications for them. These justifications, in our view, become internalized by the group members as disengagement mechanisms. Bandura (2016) gives numerous examples of the usage of DMS by terrorists, especially, Al-Qaeda members, jihadists, and former fighters of the ira in Ireland. For the case of jihadism, ^{Cottee (2010)} documented the usage of moral justification of war and violence against non-Muslims, victim blaming and dehumanization, disregard of consequences and denial of personal responsibility or agency in the writings and public statements of some of its influential leaders.

As for the usage of the mechanisms by jihad-ist militants, our review yielded two empirical studies. The first one (^{Hafez, 2006}) was based on a content analysis of the last will and statements made by Palestinian suicide bombers before their attack on civilians. The findings showed that the most commonly used mechanisms were: moral justification of violence, advantageous comparison among Palestinian and Israeli violence, attribution of blame for the violence, and euphemistic labeling. The second study was conducted by ^{Khalil (2017)}, who analyzed 216 articles published by the isis during 2015. Results revealed the usage of all the mechanisms, with the exception of blaming the victim, not mentioned by the author, and a supremacy of moral justification. Contrary to Khalil's findings, ^{Bandura (2016)} persuasively argues that some mechanisms are suited for terrorist purposes whereas other are less useful and seldom used.

As for the politically motivated terrorism, ^{Martin-Peña and Opotow (2011)} analyzed the statements made by the Spanish terrorist group ETA after its terrorist attacks. They used categories similar to Bandura's and found a prevalence of advantageous comparison and euphemisms, Domain (a) in Table 1, but not of moral justifications. Blaming the victim and displacing responsibility, Domains (b) and (d) in Table 1, were the other two most frequent mechanisms. In contrast, ^{Sabucedo, Blanco, and de la Corte (2003)} found-in a newspaper linked to the ETA-a predominance of legitimization ofbeliefs related to the detrimental effect of the consequences and to the victim, Domains (c) and (d) in Table 1. Thus, the kidnapping and murdering of innocent people were legitimized through attribution of responsibilities to the enemy, the depersonalization of the victims, and an asymmetric evaluation of suffering. These results, although restricted to the ETA and based on small samples, evidence dispersion in the usage of DMS.

^{Sabucedo et al., (2004)} investigated the usage of delegitimization labels, corresponding to the category (d) of devaluing the victim, made by the Colombian groups-FARC and AUC-. They discovered that each group presented the adversary as violator of social norms, whereas depicting the enemy with negative personality traits was uncommon. These groups differed only in the frequency of negative political labels for the opponent, higher for the FARC than for AUC. The mentioned similarities suggest that the justificatory dynamic and categories are the same, despite sociopolitical differences (^{Borja, Barreto, Sabucedo, & López, 2008}). ^{Bolivar (2006)}, in turn, analyzed the discourse of the two Colombian groups during peace talks. Based on mass media interviews, she discovered differences in the self-portraits offered by the groups' leaders. Drawing on such differences, she concluded, nevertheless, that the two groups diverged in their emotional repertories and ideologies, and consequently in the justifications for their action. Apart from the aforementioned studies and, as far as we know, there are no other studies that compare the psychological characteristics of terrorists or violent armed groups.

As yet, it is unknown if all mechanisms are equally effective to avoid self-sanctions and reduce dissonance. ^{Bandura (2002)} believes that cognitive restructuring of harmful conduct, Domain (a) in Table 1, is the most effective for disengaging moral control. We assume that the frequency in the usage of each DM is a hint of its perceived effectiveness and it is, therefore, of interest in the present research.

Main Illegal Armed Groups in the Colombian Conflict: the faro and AUC

Colombia has suffered a long-standing internal armed conflict caused by guerrillas and paramilitaries. The Revolutionary Armed Forces of Colombia (FARC) are the oldest and most prominent guerrilla group, which emerged in 1964 as an armed organization with a Marxist-Leninist orientation, mainly composed of young and low educated peasants (^{Gutiérrez, 2008}). Their goal was to take power and change the structure of the state to bring about a fairer income distribution (^{Ferro & Uribe, 2002}). Over the years, the FARC'S ideological, political and military position weakened, due to increasing international rejection of violence and terrorism, their engagement in drug-trafficking (^{Saab & Taylor, 2009}), kidnapping and child recruitment, as well as the action of the armed forces and the massive desertion of combatants. This group signed a peace agreement with the Colombian government in 2016, after four years of negotiation.

The paramilitaries conformed the United Self-Defense Forces of Colombia (AUC), which emerged toward the end of the 80's with the support of rural elites, politicians and some members of the Colombian Armed Forces to confront the guerrillas and to defend the State and its institutions. AUC was comprised of ex-members of the Armed Forces, former common delinquents (^{Gutiérrez, 2008}), security personnel of extinct drug cartels, small or medium sized drug barons and some regional elites of landowners (^{Saab & Taylor, 2009}). AUC adopted a national political discourse of counterinsurgency and protection of the status quo (^{Duncan, 2006}; ^{Pizarro, 2004}), but in fact, they were "local responses to the guerrillas" (PNUD, 2003. p. 29). This grouping of local armies made it difficult for them to achieve unified political goals and sometimes led to actions and decisions that were not consulted with the central leaders. Over time, AUC merged with drug traffickers and emerald smugglers (^{Romero, 2003}), and an initial social tolerance to that group turned into rejection. In 2006, AUC negotiated a peace process with the Colombian government; as a result, most of their members laid down their arms and rejoined civilian life. This process has been difficult because many former paramilitaries have regrouped as armed criminal groups.

Both the paramilitaries and the FARC were involved in terrorist acts (^{Feldmannn & Hinojosa, 2009}; Observatorio de Derechos Humanos de la Vicepresidencia de la República, 2005; Report of the un High Commissioner for Human Rights in Colombia, OHCHR, 2010); hence their justifications shed light on terrorism. To reach our goals, we examined the mechanisms that these groups spontaneously used in their press releases, their frequency, and possible differences in preference and frequency between both groups.

Results

During the 55 month period, the FARC published 216 press releases, about one per week, and AUC 115¹, two per month. As for the mechanisms, their frequency was 749, for a ratio of 2.08 per press release. The average of mechanisms used by the FARC (M= 1.92; SD= 1.52) and AUC (m= 2.08; SD= 1.88) was similar, with z(329)= .08, p=.93.

Regarding the nine different mechanisms, results in Table 2 indicate that the most used were: attribution of blame (21%), euphemistic labeling (20%), moral justification (17%) and labeling with undesirable names (16%). Dehumanization represented only 4% of the total, while displacement and diffusion of responsibility, both belonging to the domain of agent's responsibility, exhibited a percentage even lower than dehumanization. Of the most used mechanisms, the first and fourth corresponded to the Domain of victim (d), while the remaining two belonged to the Domain of reconstruction of behavior (a). Overall, 74% of the used DMS corresponded to the Domains (d) and (a), each one with exactly the same proportion, whereas the remaining 26% belonged to the Domains of the agent's responsibility (b) and detrimental effects (c).

Table 2 Frequency and Percentage of the Mechanisms of Moral Disengagement and Differences by Group 

To test differences between the groups in their use of each mechanism, and to avoid Type i errors, we adjusted the required p level. Thus, after excluding the two most infrequently used mechanisms, whose percentages were too low for a z test, seven mechanisms remained to be compared. Following the Bonferroni procedure, for 7 univariate comparisons (k=7) we obtained a required p level of .007, when dividing .05 by 7. Results in Table 2 show that any comparison reached the needed significance level and consequently, the groups do not differ significantly in the use of dms in spite of their differences in ideology. Diffusion of responsibility, although used exclusively by AUC members, represents only a negligible percentage (2%) of the total of mechanisms.