Abstract
Scholars have long argued that dehumanization causes violence. However, others have recently argued that those who harm do so because they feel pressured or view violence as justified. Examining the Rwandan genocide, this article contends that contradictory theories of dehumanization can be reconciled through consideration of cultural and moral sociology. Research on culture and action demonstrates that when people strive to implement new practices, they often explicitly work through them cognitively and emotionally. With time, however, these conscious processes diminish until actions that were once new proceed with ease. In another vein, morality research suggests our affective responses to actions indicate their moral significance; when we do not react emotionally to actions, they are morally irrelevant. Herein, I combine these ideas with a temporal analysis of Hutus’ recollections of killing Tutsi and find cognitive, emotional, and relational transformations rendered killing mundane over time. Dehumanization was a consequence of violence, not a cause.
The opposite of love is not hate, it’s indifference.
Both scholarly and popular thinking about violence often presumes not just its connection to dehumanization but also a temporal order to that connection: Dehumanizing others leads to violence against them. Yet emerging research has begun to challenge this long-held presumption, in some cases even suggesting people may choose to harm and kill victims specifically because they perceive them as human and want to harm them as such (Bloom 2017; Manne 2016). Rather than act violently because victims no longer seem human, these scholars propose people act violently because victims are socially relevant in ways that make it feel satisfying for perpetrators 1 to harm them (Fiske and Rai 2014; Katz 1990). This has led to the critique that dehumanization holds limited power for explaining participation in violence, including the violence of genocide, to which dehumanization is so frequently linked (Lang 2020).
Examining the case of the Rwandan genocide, in this article, I seek to bridge claims that dehumanization does and does not motivate violence by considering insights from the sociology of culture and morality. Research on culture, motivation, and action shows that as people strive to adopt new practices, they often explicitly work through those practices both cognitively and emotionally. Over time, however, those deliberate processes diminish until, eventually, once “new” actions become embodied and can proceed forward unproblematically (Cerulo, Leschziner, and Shepherd 2021:64). Morality research, meanwhile, shows that our feelings about our actions indicate to us that they are morally significant; when we do not react emotionally to an action, it is morally irrelevant (Abend 2014:30; Hitlin and Vaisey 2013:55). I join together these two ideas in this article with a temporal analysis of Hutu perpetrators’ recollections about killing Tutsi and find that relational transformations are part of the process of adopting new practices as well. Furthermore, I suggest these relational transformations have moral consequences because our feelings about actions do not exist separately from our feelings about the people involved in them: A person, like an action, is not morally significant unless they have an effect.
Thus, at Time – 1, I find that Hutu with no preexisting history of violence had varying perceptions of Tutsi, but importantly, they still thought and had feelings about them—which is to say, Tutsi were socially relevant. As a result, at Time 1, participating in violence evoked strong emotional responses for Hutu, who had been socialized to view violence against other humans as wrong. At Time 2, however, later in the genocide, violence ceased to elicit an emotional response or even a second thought for Hutu killing Tutsi. The mechanism linking Time 1 to Time 2 was a shift in social perception: Participants in the Rwandan genocide reclassified Tutsi as socially meaningless, pushing them beyond the boundary of moral consideration such that killing them became irrelevant. No longer did they think or emotionally react to their violent actions, nor did they think about who they were acting toward. Dehumanization was thus a consequence of violence, not a cause.
This finding suggests the process of becoming a genocide perpetrator is not much different from processes of “becoming” more generally (see Becker 1953; Benzecry 2009; Goffman 1959; Pagis 2019; Tavory 2016; Tavory and Winchester 2012; Wacquant 2004; Winchester and Green 2019). As ordinary Rwandans habituated physically to their gruesome acts, they also habituated cognitively and emotionally, and with this “body and soul” (Wacquant 2004) habituation came a third subjective shift: a transformation in how they perceived Tutsi. Hence, despite the empirical specificity of this research on genocide perpetrators, I propose it also provides a general theoretical contribution to sociologists studying experiential—and in this case, moral—careers. After all, Goffman (1959:123) defined shifts in moral careers as the sequence of changes in a person’s self and their “framework of imagery for judging [themselves] and others.” I bracket self-judgment and return to this in the discussion, but my findings suggest, too, that relational transformations are interwoven with cognitive and emotional transformations in the process of becoming a person anew.
Theorizing Dehumanization
Dehumanization concerns the perception of a person or category of people as socially irrelevant due to their less-than-human or not-at-all human characteristics. Recent theories therefore emphasize that dehumanization operates on a spectrum: from fully human, to infrahuman, to dehuman and stripped entirely of human traits and meanings (Haslam and Loughnan 2014). These judgments connect to our perceptions of others’ emotional capacity, specifically their possession of nonuniquely human (primary) and uniquely human (secondary) emotions. Those perceived as fully human are assumed by the perceiver to hold both primary and secondary human emotions; those perceived as infrahuman are thought to hold the complete range of primary emotions—happiness, sadness, surprise, fear, anger, and disgust—but fewer of the uniquely human secondary emotions than the perceiver or their in-group (Leyens et al. 2003). Those perceived as not human are considered void of any emotions at all. In fact, social cognition (the perception of others as people with their own thoughts and feelings) fails to activate when we perceive dehumanized others.
This failure to perceive people as people is important: Dehumanization is often conflated with hatred, both in popular imagination and in academic research, but prior work shows that dislike and dehumanization are culturally learned practices interpreted differently by the brain. For example, Bruneau et al. (2018) use functional MRI scans to show that distinct parts of the brain activate when a research subject evaluates people and animals as liked or disliked versus dehumanized. Similarly, Harris and Fiske (2006, 2011) report that in the brains of research subjects evaluating dehumanized others, the areas that normally and spontaneously activate upon perceiving others do not activate (see also Cikara, Eberhardt, and Fiske 2011; Waytz et al. 2010). Accordingly, contemporary theories of dehumanization emphasize that in contrast to typical practices of social perception, whereby people are categorized into kinds and those kinds are attributed meanings of worth and value (Bowker and Starr 1999; Kunda 1999; Zerubavel 1996), dehumanization is distinct in that it entails the judgment of no worth and no value to others. Dehumanized people are socially meaningless.
Dehumanization, Morality, and Violence
Because dehumanization strips people of their worth as persons, dehumanization is said to enable total moral disengagement (Bandura 1999; Bar-Tal 1989). It denies people value as human beings (Staub 1989), divorces them from community and identity (Kelman 1973), and places them “outside the boundary in which moral values, rules, and considerations of fairness apply” (Opotow 1990:1). “Dehumanization,” writes Fein (1979:4–5), entails the construction of others as outside a “moral universe of obligation,” thus loosening the constraints that would otherwise prevent violence against them. In turn, the concept of dehumanization calls attention to the fact that humanness itself is a social characteristic that can be attributed or denied to others, with consequences for judgments and acts of harm. Not surprisingly, then, dehumanization has long been linked to genocide, the “crime of all crimes” (Schabas 2003:39).
In fact, as Adorno and Horkheimer ([1947] 2002) argued in what was then the first study to attempt to understand what the authors termed the “barbarism” of their time (Aschheim 2017:428): The “objectification of mind[s]” and transformation of “souls into things” through enlightenment obsessions with rationalism enabled Germans to act violently yet mindlessly toward Jews and others during the Holocaust (Adorno and Horkheimer ([1947] 2002:xiv, 21). Similarly, Arendt’s ([1951] 1979:475) thesis on the banality of evil asserts that central to genocide is perpetrators’ failures to think and judge—mass violence, she writes, becomes possible when reflection and evaluation are no longer part of people’s lives, an extension of how totalitarian states turn their targets into “superfluous” subgroups, “not belong[ing] to the world at all.” Finally, Bauman’s (1989:75) claim that “the dead silence of unconcern” is necessary for mass violence on the scale of the Holocaust finds kinship in contemporary theories of dehumanization and its relationship to genocide. 2 Each argues that genocide is caused by thoughtlessness, the failure to perceive victimized people as people and to think about them as humans deserving moral concern.
Yet dehumanization is not confined to genocide. In recent years, social psychologists have suggested that a number of harmful practices are caused by the failure to perceive others as human. For example, White survey respondents who unconsciously associate Black people with apes are more likely to believe police force used against them is justified (Goff et al. 2008, 2014). Concerning sexism, men who objectify women as tools and objects are more likely to be violent toward female partners (Jonnson, Langille, and Walsh 2018; Sáez et al. 2022). Men who more quickly associate women with words related to animals and objects are more likely to score high on a rape-behavioral analog that evaluates proclivity to rape (Rudman and Mescher 2012). On immigration, Americans who rate Arabs as “less evolved” are more likely to endorse policies that restrict Muslim immigration, support the use of violent counterterrorism tactics such as drone strikes and torture, and agree with vindictive statements such as, in the wake of the 2013 Boston Marathon bombing, “Muslims bombed Boston. We as a planet need to wipe them off this world. Every one of them” (Kteily et al. 2015). Dehumanization is a “fundamental sociological process” that researchers across the social sciences blame for a range of violent acts (Hagan and Rymond-Richmond 2008:875).
Dehumanization, Morality, and (Actual) Violence
Simultaneously, we still do not know how dehumanization is linked to violence from the perspectives of those who engage in it. In fact, few of the aforementioned studies actually ask respondents to imagine themselves perpetrating physical harm. 3 Moreover, although the majority of these studies find a positive correlation between dehumanization and support for harm, their reliance on implicit association tests, the Ascent of Humans Scale, and electroencephalogram and functional MRI scans coupled with batteries of attitudinal questions about social and political practices and policies, obviates their ability to identify a causal relationship between dehumanization and actual participation in violence. It is thus possible these tests are simply measuring known cultural pairings rather than implicit or explicitly held belief systems. White people, for example, may not truly perceive Black people as apes, but they may make this troubling association readily because of historical representations linking the two in U.S. culture (Over 2020:5). These studies provide no evidence that dehumanized perceptions cause violent actions.
As a result, in recent years, some scholars have suggested that much of the violence typically associated with dehumanization is in fact enacted because perpetrators perceive victims as human and want to harm them as such to assert their dominance. In so doing, meting out violence is a way to remind victims of their “rightful place” in the social order—as humans, but subordinate humans (Bloom 2017). Hence, male rapists rape because women’s full humanity makes them feel threatened; they use rape as an assertion of power over another human or group of humans, not as a morally irrelevant act against a socially irrelevant animal or object (Manne 2016). In this approach, violence is enacted not because victims are meaningless and so violence is amoral, but rather, violence is perceived as correct, even virtuous and attractive, to impassioned killers who seek to right moral wrongs (see also Fiske and Rai 2014; Katz 1990). These studies do not resolve the question of temporality but negate dehumanization’s relevance for violence altogether.
Finally, and concerning the case examined in this article, contemporary scholarly consensus on the Rwandan genocide is that within-group social ties and top-down pressures explain most Hutus’ decisions to kill Tutsi (Fujii 2009; McDoom 2013; Mironko 2004; Straus 2006). Moreover, although the Rwandan genocide is commonly considered a sui generis case of dehumanizing propaganda inciting neighbors to kill neighbors (e.g., Inskeep 2022), research tracing the relationship between media and participation in the genocide finds scant evidence that the violence can be explained by such persuasion. Straus (2007), for example, analyzed the content, timing, and exposure of Radio Télévision Libre des Milles Collines (RTLM) broadcasts. Although the International Criminal Tribunal for Rwanda convicted three RTLM journalists of fanning the flames of genocide through its depictions of Tutsi as cockroaches and encouragement of Hutu to kill them, Straus (2007) found most propaganda depicting Tutsi as such aired only after most of the violence had occurred. During the “high genocide period,” announcers varied their broadcasts, sometimes even urging Hutu not to kill Tutsi (Straus 2007:624). Prior to the genocide, RTLM’s broadcast range scarcely reached rural areas (88 percent), and fewer than 10 percent of households owned radios (Straus 2007:616). Meanwhile, Yanagizawa-Drott (2014) combined village-level data on the genocide with village-level radio reception and concluded that only 10 percent of Hutu who were prosecuted for violence could be considered mobilized by radio. Yet the mechanism linking RTLM to genocide was not persuasion but coordination and spillover effects. These findings seemingly support arguments that the dehumanization thesis is exaggerated: People act violently for reasons having little to do with perceptions of victims as dehumanized, and where it exists, dehumanizing propaganda has only indirect effects. However, everything becomes clearer when we consider insights from the sociology of culture and morality.
Becoming a Genocide Perpetrator
Cognition, Emotion, and Relationships
The relationship between attitudes and action is far from straightforward. Rather than the former necessarily causing the latter, sociologists in recent years have demonstrated that what initially appears to be a motivation for engaging in unfamiliar actions may in fact be a late narrative that helps actors adhere to and maintain new behaviors but not initiate them. For example, Winchester and Green (2019) show that individuals who are new to the stringent fasting practices of Eastern Orthodox Christianity or the intense physical exertion demanded by mixed martial arts (MMA) have to talk themselves into these practices due to their difficulty and deviation from past behavioral norms. Over time, however, and with ongoing discursive and corporeal experience, fasters and fighters adopt new subjectivities such that once-unfamiliar actions become routine and require no explicit cognitive or bodily attention for practice. Similarly, Leschinzer and Green (2013:133) show how chefs at various career stages in high-status restaurants combine automatic and deliberate cognition along with routine and nonroutine practice in their attempts to create innovative dishes. In their case, nondeclarative and declarative culture work together as people take on new practices that, with time and experience, become unconscious and embodied and allow them to move unproblematically forward (see also Surak 2017).
Yet we do not simply attend to the world through our minds and bodies (Winchester and Pagis 2022:17; cf. Merleau-Ponty 2002; Polanyi [1966] 2009). Rather, as judgment and decision-making research shows, we also attend to the world through affect, with the latter often the primary response to disruptions in practical experience (Bruch and Feinberg 2017:215). In turn, habituation is not simply about cognitive and physical mastery but also about emotional adaptation and mastery. Tavory and Winchester’s (2012) comparison of newly Orthodox Jews and converted Muslims is illustrative: As actors in each category adopted newly prescribed acts, they also experienced poignant emotional moments, which they interpreted as a connection to the divine. Over time, however, as these once difficult and disruptive actions became routine, these practitioners began to experience fewer religious feelings—a diminishment of the emotional “enchantment” that, paradoxically, led them to seek conversion in the first place. Wacquant’s (2004, 2015) enactive ethnography likewise demonstrates how the early physical pains and emotional bruises he endured as he became a boxer eventually disappeared as he came to embody the pugilistic practical sense. These and other studies on processes of becoming suggest that as people embark on new experiential careers, their actions are not driven by existing cultural habits but rather contribute to habit formation and internalization when successful. Moreover, they suggest these habits are not simply mental and physical but emotional, too.
Such findings matter for theorizing about dehumanization and violence. As is now commonly recognized, we know an action is moral when we feel a certain way about it. We feel bad when we violate moral norms, and we feel good when we uphold them. We feel nothing—that is, we have no emotional response—when we consider our actions morally irrelevant (Abend 2014:30; Hitlin and Vaisey 2013:55). As a result, as individuals cognitively and emotionally adapt to behaviors they once deemed impossible, they also transform into new moral beings. Winchester and Green (2019:265) allude to this when they describe how actors’ self-talk in the process of becoming religious or MMA practitioners helped them link their actions to a “more physically, morally, and/or spiritually advanced sense of self” when they lacked motivation to continue. Similarly, Wacquant’s (2004:15, 40, xii) description of the Woodlawn Boys Club as “a school of morality” that required fighters to adopt new “categories of judgment,” such as “an uncompromising sense of masculine honor,” suggests that adopting a new physical craft necessitates cognitive, “sensual,” and “moral” conversion. In a counterexample, Suitt (2021) demonstrates how military veterans’ inability to reconcile their actions with their mental and emotional distress leads to ongoing moral trauma. Thus, not only can people act prior to internalizing the values associated with their actions, but as their actions become routine, they no longer reflect on or require the explicit justifications that helped them endure their initial discomfort. In contrast, where individuals fail to habituate, the cognitive and emotional discomfort associated with their initial excursions into a new practice persist and generate ongoing moral conflict. Can the same processes explain how individuals with no prior history of violence adapt to their participation in genocide? And can they tell us anything about the relationship between dehumanization and violence?
To preview the argument, this article asserts yes, but with a twist. First, there is little evidence of explicit self-talk in this case, at least not during the violence, although it is possible that the motivations Hutu perpetrators provided post hoc for their participation served as contemporaneous justifications when they first began to kill. However, this article is not concerned with why they killed. It is concerned with how they thought and felt about killing, whether and how their thoughts and feelings changed over time, and whether their shifts in cognition and emotion about violence related to shifts in social perception. Accordingly, although I analyze a range of testimonies, including some where perpetrators describe their motivations for killing, I interpret these with a focus on their thoughts and feelings and find that Hutu thought about and struggled emotionally with their early acts of violence, but their cognitive and emotional discomfort diminished over time. This diminishment, I show, was tied to shifts in their perceptions of Tutsi such that as Hutu grew accustomed to killing Tutsi, they also began to perceive Tutsi as socially irrelevant. The lack of self-talk concerning their perception of Tutsi as dehumanized is not surprising when we consider that social perception is typically considered a form of nondeclarative culture and that nondeclarative culture is typically acquired from repeated activities and recurrent exposures, not conscious activity (Cerulo et al. 2021:64).
The second contribution of this article is that cognitive and emotional habituation to new practices is tied to relational changes and habituation to those changes as well. In this case, Tutsi whom Hutu previously regarded as neighbors, friends, or at the very least socially relevant were no longer relevant to them at all. In fact, they became socially meaningless and thus dehumanized. As a result, Hutu perpetrators’ mental and emotional responses to killing changed concurrently with shifts to their social cognition. The more they killed, the less they thought about killing, emotionally reacted to their own acts of violence, and considered those they were harming.
This finding need not be limited to perpetrators of genocide: Green (2011), for example, discusses relational transformations in his emphasis on the new communities MMA fighters form over time, but he does not discuss how those new ties bore on his respondents’ preexisting ties or on their habituation to becoming MMA fighters. Similarly, Winchester and Green (2019:265) note how “engaging in Eastern Orthodox and MMA careers made [practitioners] appear ‘strange,’ ‘weird,’ or ‘crazy’ to significant others,” but they do not explore the consequences of their subjects’ ongoing participation in conversion and fighting for their preexisting relationships. It is possible, however, that the more actors became fighters or converts, the more their old ties transformed. One can imagine strained and splintered friendships, ruptured relationships, and even divorce as an outcome of these new commitments, with the deterioration of old ties becoming irrelevant over time. This article thus suggests that as we become new people—killers, fighters, or converts—we relate to people from our pasts anew. These relational transformations matter because they affect our ability to proceed unproblematically in our new “careers.” 4
Finally, a third contribution of this article is that the shifts in social relationships that accompany processes of becoming have moral implications. This is because morality does not simply concern how we feel about actions and values in the abstract or even only in relation to ourselves; rather, morality contrasts our interests with those of others. Whether we feel it is right or wrong to steal, lie, or cheat, those feelings are likely to change whether the victim is a charity, school, store, relative, friend, or stranger (Bloom 2011; Earp et al. 2021; Hester and Gray 2020). Our moral decisions are thus informed by culturally acquired moral “oughts” and ideals and culturally informed relational expectations and demands (Cikara, Martinez, and Lewis 2022; Luft 2020a). Hence, much as we can alter how we feel mentally and emotionally about unfamiliar or difficult actions through explicit self-talk and practical routinization, we can also alter how we feel about difficult actions by altering, even unconsciously, how we evaluate our relationships. Our judgments of people do not have to guide our actions toward them; they can change as a result of our actions, including our participation in actions we find abhorrent, and these subjective shifts in how we perceive others can alter how we morally evaluate these actions. Embodied cognitive, emotional, and relational experiences are interwoven.
Data and Methods
I focus on the cognitive and emotional experiences of genocide perpetrators and how these experiences relate to their perceptions of victims. To do so, I combine three qualitative data sources, which I interpret by examining how Hutu thought and felt about killing and by asking if and whether those thoughts and feelings changed over three time periods and in relation to their perceptions of Tutsi. It is, however, important to note two additional features of these data that complicate yet also complement my analytic approach: Much of these data were produced at a fourth point in time, between 7 and 20 years after the genocide, and respondents were not explicitly asked about their thoughts and feelings. Rather, they were asked by different interviewers about why they killed and how they killed, and it is in the process of narrating their experiences that I identify patterns in how their thoughts and feelings about killing and about their victims changed. I also include data from observers who were in Rwanda before and during the genocide and from survivors and rescuers, who similarly testify to witnessing perpetrators’ shifts in cognition and emotion despite not being asked explicitly about it. This allows me to generate a theory of the moral careers of genocide perpetrators—that is, a “standard sequence of changes in [an individual’s] conceiving of selves”—that can be tested with future work (Goffman 1959:141).
Deception and Data Set Variation
Interviews with Rwandan genocide perpetrators from Bugesera in Hatzfeld’s (2005) Machete Season: The Killers in Rwanda Speak form the core of my analysis. The total number of interviews is 10; however, after the initial round of coding, I eliminated 2. One interview was removed because the respondent was a municipal commander of the Interahamwe, an extremist Hutu militia that helped prepare for and organize the genocide before it began. This participant’s status as a “violent specialist” (Tilly 2003:35) cannot be compared to the others, who were all civilians prior to the genocide. The other interview was eliminated owing to the respondent’s prior experience with violence. This respondent, a former member of the military, joined the police department but was discharged for killing two civilians in 1992. 5 Given this article’s emphasis, it was essential to include only interviews with respondents who had never killed another human. 6 The remaining participants were farmers’ sons and men with a median age of 29, reflective of the average Rwandan genocide participant (Straus 2006). All were incarcerated at the time of their initial interview in 2001, having already undergone a judicial investigation and confessed to their crimes.
Nevertheless, this is a small sample from a single point in time; therefore, I rely on two more sources to strengthen my analysis. According to Small and Cook (2021:29), questions on morally charged actions make deception more likely. As noted earlier, I am not interested in why respondents acted as they did; nonetheless, I do rely on interviews in which respondents discuss questions of motive, and I interpret these interviews with a lens toward what I am interested in—their changing thoughts and feelings about victims and violence. Thus, the potential problem of deception remains relevant. To counter this, I took up Small and Cook’s (2021:16) suggestion of triangulating data and Tavory and Timmermans’s (2013:690) strategy of ensuring data set variance. I read the totality of Human Rights Watch’s (HRW) 7 reports on violence in Rwanda from 1990 to 1999 (n = 9) and the USC Shoah Foundation Visual History Archive’s 25 oral testimonies from Rwanda (5 are from Bugesera: 3 survivors and 2 rescuers). The narratives in HRW reports, oral testimonies, and interviews conducted by Hatzfeld (2005) represent different groups and types of actors and were, as mentioned previously, collected at different times (i.e., some HRW reports predate the genocide, and the Shoah Foundation’s oral testimonies were collected between 14 and 20 years later). 8 This data set variation helps me corroborate Hatzfeld’s (2005) respondents’ claims.
Finally, in examining the Rwandan genocide, data set variation is especially important because since 2003, the Kagame regime has strictly controlled conversation about the genocide, punishing those who depart from the state’s narrative (Ribara Uwariraye 2022). Central to this account is the contention that Belgian colonial constructions of Hutu and Tutsi as two distinct racial categories instilled in all Hutu a racist hatred of Tutsi, which was exacerbated by Hutu extremists’ dehumanizing propaganda in the years preceding the genocide and led to the outbreak of mass violence in 1994 (Gatebuke 2023). Consequently, the current RPF government, through its National Unity and Reconciliation Project, has institutionalized a new national identity category of “Rwandan” that aims to erase these divisions and return Rwanda to its supposed precolonial harmony (Chakravarty 2016; Luft and Thomson 2021; Purdeková 2015; Russell and Carter 2019; Thomson 2013). Nevertheless, Kagame’s regime promotes racialized distinctions by prohibiting Rwandans from discussing the genocide in all its complexity, including Hutu as resisters, rescuers, bystanders, and victims (Fox and Brehm 2018; Fujii 2009; Jessee 2019; Luft 2015; Luft and Thomson 2021; Thomson 2013; Ribara Uwariraye 2022). Public references to Hutu and Tutsi that deviate from the narrative of all Hutu as racist killers and all Tutsi as only victims are currently punishable by three to seven years imprisonment and a fine of 500,000 to 1 million Rwandan Francs, with some convictions resulting in 10- to 25-year sentences (Amnesty International 2010:1). Forced disappearance is also prevalent (Gatebuke 2023). Hence, at each time data have been collected since the genocide, the physical, material, and symbolic interests of Rwandans have been at risk for different reasons, making it all the more important to rely on evidence gathered over various time periods; this also makes it ethically problematic to collect new data as well (see online Appendix A). Data set variation enables me to recognize and acknowledge that the present influences the past in how Rwandans recall the genocide, and given how tightly the RPF currently controls the postgenocide narrative, it is all the more remarkable that similarities across respondents’ recollections from before and during the genocide (HRW), shortly after (HRW and Hatzfeld), and up to two decades later (Shoah Foundation) exist.
Interpretation and Within- and Across-Case Variation
In addition to data set variation, two further sources of variation guide my methodological approach: within-case temporal variation and across-case comparison. First, I treated each interview within the total population as a case, which, in following Beach and Pedersen’s (2016:5) comparative case method, I examined as “an instance of a [hypothesized] causal process playing out, linking a cause (or set of causes) to an outcome.” Then, for each case, I followed a temporal process-tracing pattern whereby I organized all interviews by person speaking. I hand-coded each interview for information describing how participants thought and felt about the violence they participated in; if those thoughts and feelings changed, then how and why; and anything that spoke to how respondents perceived and related to Tutsi (or in the case of Tutsi survivors and Hutu rescuers, how they remembered relationships with each other). I then arranged the results according to perpetrators’ recollections of timing and the precise details they provided about killings in which they participated (or that survivors survived and rescuers helped them avoid). This permitted me to map the descriptions of respondents’ thoughts, feelings, and perceptions onto a rough estimate of the genocide trajectory in Bugesera: T–1: before the genocide (any time prior to April 6, 1994); T1: first few killings (April 11–23, 1994); T2: high point of genocide/ongoing participation in violence (April 23–May 14, 1994); and T3: end of genocide (May 14 in Bugesera; July 4 in Rwanda overall).
Second, after completing my within-case process tracing, I compared the results organized by within-case temporal variation across all cases. I then further evaluated them in light of HRW’s reports from before, during, and after the genocide. The following is a summary of my findings, and I propose they be considered retroductive (Katz 2001:461) for future research: Instead of making a positivistic claim that T1 will always lead to T2, I argue that when we see evidence of dehumanization, T2, people have already participated in violence, T1.
At T–1, respondents thought about and perceived Tutsi as socially relevant.
At T1, respondents still perceived Tutsi as socially relevant and thought killing them was wrong, resulting in feelings of shock or horror when they participated in violence.
At T2, respondents stopped perceiving Tutsi as socially relevant, and they did not perceive killing them was wrong. Respondents had no cognitive or emotional response to violence.
Overall, increasing participation in violence led previously nonviolent individuals to kill unproblematically and to no longer perceive their neighbors as human.
Theorizing and Considering Similar Kinds of Studies
Given the broader theories of culture and morality this study draws on and the broader connections to theories of becoming I seek to make, I consider my findings in relation to the shared communities of inquiry discussed earlier (Timmermans and Tavory 2022). For example, drawing on studies that attend to the emotional transformations involved in processes of becoming (e.g., Tavory and Winchester 2012; Wacquant 2004), I looked for when and how respondents described problems they needed to solve from their own perspectives—what Katz (2001) calls “luminous” or “poignant moments” that cause a rupture in recalled experiences—and I compared these with moments when they described experiences or actions as not emotional at all, or in fact could not remember them because they had become so insignificant. Similarly, I looked for evidence of explicit self-talk (justificatory deliberation), particularly concerning the classification of victims as dehumanized, in keeping with past work on the temporal dynamics of experiential careers that suggests people have to “talk themselves into” new and difficult practices (Winchester and Green 2019), to examine whether shifts in social perception and relationships follow a similar process.
Toggling between my results and other research in this way allowed me to identify the surprising finding that whereas cognitive and emotional habituation has, in past work, been linked to temporal changes in practical experience, both are also linked to relational shifts such that precisely the relationship originally experienced as meaningful in the moment of rupture—here, Hutus’ relationships with Tutsi—in fact diminishes in significance the more people embody their new practices. Yet this shift in social perception and relationships is not a result of explicit deliberation. Rather, it is a consequence of routinization, much like how social perception is learned more generally. This has moral consequences because, as explained earlier, our ability to think about others when interacting with them and to perceive them as socially relevant is central to how we evaluate actions against them. I therefore build on and extend theories of the temporal dynamics of experience by engaging with similar kinds of studies and attending explicitly to the relational changes involved in habituating to once-novel or difficult actions. As violent practices become routinized, mental and emotional reactions to participating in violence diminish along with the social significance, for perpetrators, of their victims. Violence and victims become meaningless together.
Analysis
T–1: Perceptions of Tutsi before the Genocide
Hutu in Bugesera held varying perceptions of Tutsi before the genocide. Some disliked or feared Tutsi; others had Tutsi spouses, friends, and colleagues whom they liked and cared for. Across the cases, respondents held diverse views of Tutsi, yet none described them as dehumanized—that is, as people to whom they attributed no value and whom they did not perceive as socially relevant. As research reviewed previously suggests, even disliking or fearing people implies perceiving in them some degree of humanness, and sometimes people harm others for this reason, although I find minimal evidence of that here. Dehumanization, in contrast, implies total social irrelevance and moral disengagement. Respondents perceived their Tutsi neighbors in different and complex ways before the genocide, but not one described Tutsi as irrelevant, a marked contrast with how they would come to perceive them over time and with repeated participation in violence.
For example, Ignace was the only participant who expressed a strong dislike for Tutsis, and he attributed this dislike to land conflicts prior to the genocide: “[H]atred flourished in the fields because the plots of land were not large enough for two ethnic groups” (Hatzfeld 2005:217). Fulgence, however, disagreed: “Actually, Hutus did not detest Tutsi as much as that. Not enough to kill them all, anyway” (Hatzfeld 2005:217). According to Fulgence, conflicts over land engendered a fear among some Hutu that was taught to them by “the old folks,” but he distinguished this fear from what he later described as a numbness toward Tutsi during the genocide and an “evil spell much worse than hatred” (Hatzfeld 2005:217). Alphonse similarly attributed his feelings about Tutsi before the genocide to concerns over land and livestock but added that this “natural jealousy” varied depending on the strength of the harvest in any given year. In times of scarcity, Hutu children would hear adults claiming Tutsis had too much land and “we cannot fight poverty in this situation.” Yet after abundant harvests, “the words are forgotten” (Hatzfeld 2005:219). Alphonse later added, “I do not believe the cows presented a truly hateful problem, or else we could have just slaughtered cows. I do not believe our hearts detested the Tutsis” (Hatzfeld 2005:221). These expressions of hatred and fear of Tutsi alongside the recognition that such feelings were triggered by conflicts over scarcity suggest that although these men did not like Tutsi, Tutsi were nevertheless socially relevant, and they recognized their feelings about them as dependent on variable economic grievances and resentment. None thought killing was the solution to their problems.
Other respondents liked Tutsi and even loved them. Many had Tutsi friends and spouses. By one account, one-third of all Hutu in Ntarama, Bugesera’s capital, were married to Tutsi before the genocide (Mamdani 2001:4), and Bugesera had one of the highest percentages of Tutsi out of any Rwandan district because of the expulsion of Tutsi from the north with Rwandan independence in 1962 (1993 HRW report), followed by an influx of Hutu who migrated over the next two decades in a government land resettlement scheme (Boone 2014). Most Hutu interacted with Tutsi at work, play, school, church, and bars, and these relationships were largely peaceful. Pio, for example, described having close relationships with Tutsis before the genocide and explained how Hutus and Tutsis “lived in neighborly harmony,” so much so that “even pushing and shoving or trading harsh words” among friends “didn’t seem right to me” (Hatzfeld 2005:218). Silas, a Hutu rescuer, similarly detailed how he had Tutsi friends before the genocide, and although he overheard people using the word “Tutsi” as a slur when he was young, his father encouraged him and his siblings “not to believe anti-Tutsi ideology” (Shoah Foundation, 55532). Valerie, also a Hutu rescuer, and Yvette, a Tutsi, likewise described hearing rumors about Tutsi and negative stereotypes as children, but they said their families “never talked about it” (Shoah Foundation, 52062; 57984). Emmanuel M., also Tutsi, said “most [children] did not even know what they were” and that political, not ethnic, conflicts marked when “people really started to divide” (Shoah Foundation, 52066). None deny there was a history of violence between Hutu and Tutsi in their district, but all attest it was irrelevant for their own relationships before the genocide.
As for the influence of dehumanizing propaganda, Pancrace was the only perpetrator to mention the radio before explaining how it influenced his feelings about Tutsi with the start of the Rwandan civil war in 1990. Still, at that time, he felt “suspicion, not hatred” (Hatzfeld 2005:219). For him, hatred only entered his heart after the president was assassinated. Pancrace believed the shift was due to direct mobilization by authorities: Even though “the radios were yammering at us since 1992 to kill all the Tutsis,” he said, this did not transform his suspicion into hatred until the militia, with district leader support, entered his commune and required all men to assemble in the village center at risk of noncompliance (Hatzfeld 2005:72). There, he explained, “The intimidators shouted ‘Just look at these cockroaches—we told you so!’ And we yelled, ‘Right, let’s go hunting!’” (Hatzfeld 2005:219). The radio, he claimed, “prepared” him to obey, but without this face-to-face mobilization, his life would likely have continued as usual after the president’s assassination (Hatzfeld 2005:72). Adalbert’s interview bolsters Pancrace’s, as he recounted how “the Saturday after [the president’s] plane crash was the usual choir rehearsal day at the Church in Kibungo. We sang hymns in good feeling with our Tutsi compatriots, our voices still blending in chorus” (Hatzfeld 2005:140). Only after authorities entered his commune, Adalbert explained, did he join in the violence. Léopord said that despite the president’s assassination, “it wouldn’t have occurred to the farmers to do the work [of killing]” had the organizers not shown up: “[Th]e decision to kill, that was definitely from the organizers” (Hatzfeld 2005:182). Propaganda could only go so far.
These statements may seem self-serving. However, studies that trace the timing of the genocide support them because they link the onset of killings to the entrance of military and Interahamwe in local communes (Nyseth Brehm 2017) and to the strength of state power mediated by social cohesion at the local level (McDoom 2012). Moreover, HRW reports and oral testimonies validate these recollections. Des Forges’s (1999:141) on-the-ground reporting for HRW during the genocide explains that because initial violence targeted prominent Hutu moderates alongside Tutsi, both Hutu and Tutsi feared for their lives:
They saw the killings as broader than a genocide . . . with victims chosen on partisan, regional, or economic grounds. Both in [the capital] Kigali and elsewhere, Hutu cooperated with Tutsi in fighting off militia attacks or they fled together to places of refuge. . . . Often, Hutu made such decisions not just because of their political beliefs but because of ties of family or friendships with Tutsi.
For many, it was not clear that genocide was expected or even anticipated. The Hutu/Tutsi social cleavage was, as Emmanuel M. noted in the aforementioned, not always the most salient cleavage in Rwanda (see also Fujii 2021:20–24). Additionally, the civil war was broadly perceived as a conflict between the Rwandan Armed Forces and the RPF. It affected civilians but not clearly along an ethnoracial divide. Hence, Valerie recalled how Hutu and Tutsi initially went on patrols together “hand in hand,” and Silas and Emmanuel M. both recounted how Tutsi fled with Hutu to Kigali because they thought the government and military would protect them (Shoah Foundation, 52062; 55532; 52066). Another respondent declared that at the start of the genocide, “we didn’t know who was attacking who” (1999 HRW report: 255). Dehumanization of Tutsi did not presage the genocide in Bugesera: Extremist authorities had to work to frame their violence as meant to target them all. As a result, Valerie explained, “No one expected what happened” (Shoah Foundation, 52062).
T1: The First Kills, and Then —
Three findings stand out among participants’ initial recollections of killing another human: (1) It was easier to not feel affected by their actions when they were undertaken in a large, chaotic group; (2) more intimate, face-to-face violence triggered feelings of shock, disgust, and horror; (3) their first kills were memorable, even years later. These findings suggest Collins’s (2008) interaction ritual theory of violence has merit in genocidal contexts: Large-group dynamics can produce emotional states that override the fear and difficulty associated with killing, propelling people into actions they might not have participated in otherwise. Yet not everyone involved described the same feelings, nor did these feelings last beyond the immediate situation in which they were produced (for similar critiques of Collins, see Ermakoff 2017; Malešević 2021a). Most respondents described feeling sickened by their more intimate acts of violence—violence committed in small groups of 2 to 10 people—and these murders were memorable, suggesting not only that they evoked intense negative reactions but also that they were traumatic for those who killed (McNally 2003). At first, violence was emotionally difficult for ordinary Rwandans: They did not foresee themselves becoming murderers, and because they still perceived Tutsi as people, they reacted strongly, and often negatively, to their first few times killing.
Pancrace first murdered a Tutsi at the Ntarama Church in a massacre organized by the Interahamwe. He does not remember this act of violence because “he did not identify that one person in the crowd.” The chaos of the situation was such that he “happened to start by killing several without seeing their faces. . . . I was striking, and there was screaming, but it was on all sides, so it was a mixture of blows and cries coming in a tangle from everyone.” Nevertheless, Pancrace does remember “the first person who looked at me at the moment of the deadly blow”—it was someone he clearly recognized as a person, in contrast to the chaos of hacking and screaming in the church. Pancrace said, “The eyes of someone you kill are immortal . . . they have a terrible black color. They shake you more than the streams of blood and the death rattles, even in the great turmoil of dying. The eyes of the killed, for the killer, are his calamity if he looks into them” (Hatzfeld 2005:21–22). For Pancrace, not all Tutsi were dehumanized before he killed; some retained their individuality and hence their humanness, with lasting consequences. Even though he claims to have hated Tutsi, this hatred was not the same as dehumanization: Early in the genocide, he still perceived at least some Tutsi as people and felt awful, years later, about his murderous actions. I show later, however, that in time and with experience, both who he was killing and the act of killing became meaningless to him.
Adalbert also described how the chaos in the church meant he felt “no personal pain in the commotion” because he “struck . . . on all sides” and so did not feel individually responsible for a single person’s death. In contrast, he believed he really killed for the first time—what he called “the first time worth telling from a lasting memory”—when he murdered two children who he found hiding in the corner of a house “keeping quiet as mice.” Adalbert had access to a gun because he was put in charge of a killing group and wanted “to try [the gun] out.” He put the children “side by side 20 meters away” and “shot twice at their backs.” Having never used a gun before, he was surprised at how easily they fell. It was “strange to see the children drop without a sound (Hatzfeld 2005:13).” The weapon and the distance made Adalbert’s physical task easier but emotionally more difficult—not unlike recent findings that military drone operators are more likely to experience emotional distress and posttraumatic stress disorder than are noncombatant airmen (Chappelle et al. 2014) or the hypothesis that drone operators experience higher rates of moral injury than combatants who face direct harm when they kill (Press 2021). Reflecting on how he felt about this action years later, Adalbert said, “Now, I am seized by the memory of those children, shot straight out, like a joke” (Hatzfeld 2005:25). For many interviewees, the large massacres in which they participated early on were too chaotic to think about killing any particular person and too fast to emotionally react. But in their subsequent, more intimate experiences with violence, they felt stunned by their actions, which were emotionally meaningful to them precisely because they recognized the personhood in the Tutsis they killed.
Somewhat similarly, Fulgence admitted his first time killing someone was when he struck an elderly woman, but in his words, “she was already lying almost dead on the ground, so I did not feel death at the end of my arm. I went home that evening without even thinking about it.” Like the others, he was able to emotionally distance himself from personal responsibility in this instance. The next day, however, he “cut down someone alive and on their feet” at Ntarama Church. He described what happened next:
At one point I saw a gush of blood begin before my eyes, soaking the skin and clothes of a person about to fall—even in the dim light I saw it streaming down. I sensed it from my machete. I looked at the blade, and it was wet. I took fright and wormed my way along [the others at the church] to get out, not looking at the person anymore. I found myself outside, anxious to go home. I had done enough. That person I had just struck—it was a mama, and I felt too sick even in the poor light to finish her off. (Hatzfeld 2005:21)
For Fulgence, several things stood out about what he considered his first “real” kill: The Tutsi he murdered was a “person,” fully “alive.” Moreover, she was a “a mama,” which presumably refers to a young and possibly pregnant woman. He remembers the way her blood looked and how it felt; in his words, he “sensed it” from his machete. He remembers feeling afraid and desperate to get out of there. He remembers feeling anxious and sick. He could not continue. And he wanted to go home.
Jean-Baptiste also described feeling sick and wanting to run from the horrors of his actions after his first time killing a Tutsi. According to Jean-Baptiste, he was pulled into participation by others who threatened to kill his Tutsi wife: “Jean-Baptiste, if you want to save the life of your wife . . . you have to cut this man right now. He is a cheater! Show us that you’re not that kind.” The “cheater” was another Hutu rumored to have helped some Tutsis hide. They pinned him down and brought Jean-Baptiste a blade. He recalled:
The crowd had grown. I seized the machete. I struck a first blow. When I saw the blood bubble up, I jumped back a step. Someone blocked me from behind and showed me forward by both elbows. I closed my eyes in the brouhaha and delivered a second blow like the first. It was done, people approved, they were satisfied and moved away. I drew back. . . . I never looked back in that unhappy direction. (Hatzfeld 2005:23)
Jean-Baptiste felt disturbed by his actions. He did not want to participate in the violence, but he felt he had to. In his memory, he recoils when he sees blood for the first time. He is pushed forward and then closes his eyes for the second strike to protect himself from the horror. He recoils again and describes being unable to face the consequences of his actions. Like the others, he remembers these details years later, remembers how it felt, remembers not wanting to kill—all, I show later, in marked contrast to how it felt later on.
Finally, respondents’ descriptions of how others felt about killing also reveal this task was emotionally difficult. Fulgence explained how initially, “most appeared uneasy with the awful suffering” (Hatzfeld 2005:129), and Pancrace described how the Interahamwe had to “initiate those who seemed uneasy with this work of killing” (Hatzfeld 2005:36). Alphonse, Pio, Jean-Baptiste, and Adalbert described people who tried to avoid killing neighbors by only cutting them slightly, in ways that seemed unlikely to kill the victims, forcing “specialist[s] . . . to intervene, catch up with the target, and dispatch it” (Hatzfeld 2005:36–37). One foreign witness (1999 HRW report: 156) explained how in Kigali, “soldiers taught hesitant young people to kill on the streets. . . . When the young people balked at striking Tutsi, soldiers stoned the victims until the novices were ready to attack.”
These accounts indicate murder was initially difficult for ordinary Rwandans. To be sure, it was easier for some in the large-group setting of church massacres where collective emotions overrode individual moral commitments and victims were depersonalized, making it easier to distance oneself from a sense of direct responsibility—a finding Anderson (2017:46) characterizes as a “genocidal technique of neutralization.” Yet this was not true in all cases, as Fulgence’s vivid example of killing a young mother demonstrates. For Hutu civilians, murdering Tutsi was hard and traumatic because they recognized they were killing people and killing people was wrong. Their victims were not yet dehumanized, murder remained a moral violation, and as a result, these abhorrent acts of violence were memorable—at least initially.
T2: Shifts in Perception and Adaptation to Violence
It is impossible to separate the shifts these men experienced in how they perceived Tutsi from their cognitive and emotional adaptation to participating in violence. Even to them, the two are linked. The more they killed, the easier killing became; from their perspectives, they no longer thought about what they were doing or to whom they were doing it. As a result, respondents recalled few details about their victims as murder became routine; no longer did they experience the memorable trauma of their first kills. With time, experience, and the dehumanization of their neighbors, both victimized Tutsi and the act of murder became irrelevant.
Consider Ignace, who like Pancrace, Léopord, and Adalbert described how early in the violence, the energy of the crowd took hold and facilitated his participation in killings. A shift in how he perceived his neighbors allowed him to continue: “At the beginning,” he said, “we were too fired up to think. Later on we were too used to it. In our condition, it meant nothing to us to think we were busy cutting our neighbors down to the last one. It became a goes-without-saying.” In other words, murder stopped eliciting a cognitive or emotional response. They literally did not think about it, and they felt nothing when they killed. Describing a shift in how he perceived Tutsi, Ignace said, “They had already stopping being good neighbors of long standing. . . . They had become people to throw away, so to speak. They no longer were what they had been, and neither were we. They did not bother us, and the past did not bother us, because nothing bothered us” (Hatzfeld 2005:47).
Similarly, Alphonse stated, “Man can get used to killing, if he kills on and on. He can even become a beast without noticing it. . . . Me, I was not scared of death. In a way, I forgot I was killing live people. I no longer thought about either life or death” (Hatzfeld 2005:49). Later, he added, “We killed everything we tracked down in the papyrus. . . . We were cutters of acquaintances, cutters of neighbors, just plain cutters” (Hatzfeld 2005:20). Who they were cutting became irrelevant. Pio, who played soccer with Tutsis before the genocide and “never noticed any unease in their company,” explained how “we no longer saw a human being when we turned up a Tutsi in the swamps. I mean a person like us, sharing similar thoughts and feelings.” Pio recalled:
It is as if I had let another individual take on my own living experience, and the habits of my heart, without a single pang in my soul. This killer was indeed me, as to the offense he committed and the blood he shed, but he is a stranger to me in his ferocity. I admit and recognize my obedience at that time, my victims, my fault, but I fail to recognize the wickedness of the one who raced through the marshes on my legs, carrying my machete. . . . The most serious changes in my body were my invisible parts, such as the soul or the feelings that go with it. (Hatzfeld 2005:218, 47–48)
Pio connects the shift in his thinking about Tutsi to the shifts in his body and soul as he admits but struggles to comprehend his own participation in the genocide. Similarly, Jean-Baptiste, who had a Tutsi wife and felt horrified the first time he killed, described how “[a]t first, killing was obligatory; afterward, we got used to it. . . . You could feel uneasy about the activity waiting for you in the marshes . . . [but] after a while you got used to it and joked around like before” (Hatzfeld 2005:74, 231–32). Killing Tutsi became mundane. Silas, who saw this transition firsthand as a Hutu rescuer, observed how “people were changing by the minute—the more time went by, the more people would change” (Shoah Foundation, 55532). An anonymous survivor interviewed by HRW (1999 report: 329) recalled how, among the early assailants, “Very few of them seemed convinced of what they were doing. . . . The people of July were very different. They made me really afraid.” Another remarked how, among “the [Hutu] I knew, I couldn’t recognize them. They had transformed themselves into animals. They were like lions.” Even for those who resisted and survived the perpetrators’ violence, the change to perpetrators was striking.
Léopord, who earlier suggested violence would not have happened had organizers refrained from entering his commune, described how he first killed a man at the marketplace and said, “I can tell you the exact memory of it because it was the first” (Hatzfeld 2005:27). Later, however, he said “since I was killing often, I began to feel it did not mean anything to me” (Hatzfeld 2005:51). Léopord summarized the process of adjusting to violence and learning to perceive Tutsi as dehumanized:
We no longer looked at [our Tutsi neighbors] one by one; we no longer stopped to recognize them as they had been, not even as colleagues. . . . When you ask me to tell you my thoughts during those awkward moments . . . I don’t know what to reply. . . . We started, we got used to it, we were satisfied. . . . During the killings I no longer considered anything in particular in the Tutsi except that the person had to be done away with. (Hatzfeld 2005:121, 218, 51)
Tutsi had lost all relevance to Léopord as individuals–they only mattered to him insofar as they were dead. Léopord provided an analogy to articulate how truly mindless killing became as his perceptions of his neighbors shifted:
The teacher going to his school debates which lesson he will pull out for his class. The mechanic chooses the engine part he is going to clean. But the killer in the marshes is not bothered by personal questions. . . . Many of our thoughts were empty, and they left no memories behind. (Hatzfeld 2005:230–31)
Léopord’s correlation of this mindlessness and his inability to recall details about his behaviors years later is mirrored in other respondents’ testimonies. Pio explained, “I don’t forget the terrible things I did [but] I do forget names, days, situations” (Hatzfeld 2005:160). All his violent actions and all his victims blurred together. Tutsi became meaningless as killing became meaningless, too.
Adalbert was the only respondent to mention the influence of dehumanizing propaganda in shaping how it felt to kill over time. Adalbert had sung with Tutsi peers in church after the president’s assassination, before extremists took charge. He described how Tutsi had been called cockroaches in the past but that these slurs had little influence on his relationships. By contrast, early in the genocide, Tutsi were called “snakes,” “zeros,” and “dogs,” which he equated to “less-than-nothings.” He explained that these “taunts . . . made the job easier” and “more comfortable” because Tutsi “seemed less like us in this position” (Hatzfeld 2005:13). Thus, although dehumanizing characterizations of Tutsi preexisted the genocide and some Hutu, like Adalbert, were aware of it, there is no evidence Hutu perceived Tutsi this way, nor, as previous research has shown, did it cause them to kill. Rather, killing was difficult until these words helped ease their discomfort with their gruesome task. For Adalbert, these taunts instilled in him a “patriotic” feeling, which helped him overcome his initial discomfort, but that, too, dissipated as he adjusted to killing his neighbors (Hatzfeld 2005:220). “Later on,” he explained, “even those kinds of feelings deserted us.” Adalbert summarized this transition with reference to his memories since the genocide: “[A]s to daydreams or persistent memories, what comes to mind are only those from the first few days, when it was still new for me to kill Tutsis. The other memories, the daily expeditions that followed, have been worn away by habit” (Hatzfeld 2005:157). Tutsi became irrelevant over time. Adalbert recalled that the dehumanizing slurs helped him kill at first, but with more experience killing, the slurs were rendered insignificant as he habituated to participating in violence.
Participants in the Rwandan genocide described, over and over again, how ongoing involvement in violence rendered the act of killing and those they were killing irrelevant. Murder became an insignificant feature of daily life, no more cognitively or emotionally taxing than any other chore. Participants simply killed as they went about their day, never stopping to think about their actions or their victims. Habituation to murder was coterminous with a shift in how ordinary Hutu perceived Tutsi. Killing became meaningless because Tutsi became meaningless. Tutsi were no longer perceived as human, liked or disliked, and so murdering Tutsi was no longer perceived as wrong. This transition manifested in participants’ memories—or lack thereof—years later. As their victims and actions were rendered insignificant, they became blurry in their memories. Details fell away. Léopord was blunt: “It’s murky—I cannot keep track anymore in my memory. I considered [the people I killed] unimportant; at the time of those murders I didn’t even notice the tiny thing that would change me into a killer” (Hatzfeld 2005:26–27). What was this tiny thing? “We no longer considered the Tutsis as humans or even as creatures of God. . . . That is why it was easy for us to wipe them out” (Hatzfeld 2005:144–45).
Comparisons and Discussion
The Rwandan genocide is not the only case where cognitive, emotional, and relational transformations joined to render killing easier over time. The Holocaust and the Second Sino-Japanese War, to provide two examples, also offer evidence suggesting dehumanization is a consequence of violence rather than its cause.
In the Holocaust, the “ordinary men” of Police Battalion 101 who killed Jews in Józefów, Poland, initially struggled to murder their victims and they vividly remembered both who they killed and their early acts of violence. Yet by the time of their second massacre, only one policeman recalled details about his victims, suggesting they had become meaningless to the majority of murderers (Browning [1992] 2017). Similarly, Wehrmacht soldiers who facilitated Einsatzgruppen massacres of Jews in Belarus were initially shocked by the violence, expressed discomfort in written complaints, and told their superiors “the acts of violence . . . demonstrate a quite incredible lack of human and moral feeling” (Beorn 2014:48). Others wrote to family and friends that they “would never forget what [they] saw,” were affected “so much that [they] couldn’t eat,” and “this has been the most terrible day of my life . . . an evil seed [has] been sown” (Beorn 2014:124). Nazi intelligence, in response, recorded, “Not all soldiers have the proper attitude toward the Jews. They do not approach [them] with the desirable ruthlessness and the distance that should be self-evident for National Socialist soldiers” (Beorn 2014:57). Still, within several months, Wehrmacht soldiers began volunteering for mass executions, creating “Jew games” to torture and shoot Jews indiscriminately, and going on “Jew hunts” to find and kill those who had escaped (Beorn 2014: chapter 8). They also ceased mentioning their participation in atrocities in letters home. 9
In another conflict on another continent, men in the Imperial Japanese Army who killed, raped, and tortured Chinese civilians recalled their experiences in eerily similar ways. Sakakura-san recounted:
The [leaders] made us just watch and learn. They would stab at bodies until you couldn’t make out the body—the form anymore. Maybe thirty, twenty people would stab, all together. And we would observe. And when we went back to the troop, about half the people didn’t eat. And our—me too—we didn’t feel great, you know? And when you’d remember that—while you were remembering the sight of the body, you couldn’t eat your food. This is the truth. And then, you know, you’d get used to it, the first time, second time, third time—and by then, you’re used to it. (Dawes 2013:68)
Similarly, Yuasa-san, who performed a range of medical experiments on Chinese farmers, said:
To open that man’s [coughs] throat, there was a thing called a “field mach—surgery machine”—a machine for slicing open [organs], and we used that. But when we pu—put it into the throat, blood—whoosh, all of this RED [emphasized] blood spurted out—it came out with the air. I remember that. . . . This sort of surgery practice was conducted four, five [coughs] times. At first, despite that, I felt disgusting—I was timid. The second time, I—the second time, I felt just fine. Around the third time, I took the initiative and planned everything out. . . . I chose vivisection, and of my own volition performed them. . . . I was never really conscious of the wrongness of the fact that I was killing people. (Dawes 2013:25–28)
X—, who also dissected Chinese civilians while they were alive, recalled how “[h]e and his colleagues did not refer to the prisoners as human beings. They referred to them as ‘logs’” (Dawes 2013:40).
These poignant quotes mirror many made by the Rwandans. Hence, although a full comparative analysis lies beyond the scope of this article, the latter two examples, worlds apart from each other and from Rwanda, suggest verisimilitude across cases (Katz 2022) and the potential portability of the theory proposed here: Violence is difficult and memorable at first for those who participate in it, in part because they still recognize the humanness in those they are killing and that violence is wrong. With experience and the dehumanization of victims, however, killing is rendered mundane. Future work should probe this theory through comparative research with similar data to examine dehumanization and violence over time.
Simultaneously, this theory, and the broader argument to which it is connected, does not concern only dehumanization and violence. Rather, it raises the question of whether violence is similar to any behavior that is initially new and difficult for people participating in it. In this article, I suggest similarities exist. Looking to past research on culture and morality, for example, we see how people’s relationships can change, suggesting new practitioners may shift their perceptions of others as they progress in their new “careers” (e.g., Green 2011:388–91; Winchester and Green 2019:265). And although the violence of genocide offends our senses more than religious conversion or recreational fighting, there is no a priori reason to think the cognitive, emotional, and relational processes involved in habituating to violence differ from how humans habituate to any once-difficult practice. All involve “turning point[s] in the way in which [a] person views the world”—including how we perceive and conceive of others (Goffman 1959:141). Still, future comparative work should explore if differences exist.
Additionally, although this theory suggests that anyone can learn to become comfortable with violence, it does not imply that all who participate in violence, or other once-difficult actions, will inevitably habituate to these actions and begin to perceive them and their past relationships (including, where relevant, those they are acting against) as meaningless. After all, Hutu in the Rwandan genocide and Nazis in the Holocaust occasionally desisted from violence to save Tutsis and Jews (Luft 2015, 2020b). Future work thus ought to examine what can disrupt cognitive, emotional, and social habituation processes once they have begun. Likewise, it would be helpful to examine the decision-making of those who, after experiencing the horrors of their first kills, abstain from further violence and those who continue to kill despite ongoing distress (e.g., Suitt 2021).
It would be remiss not to also mention the findings here suggest that in many cases, killers stop perceiving themselves as human the more they kill, too. It is, however, unclear, given that these data are retrospective, whether this dehumanization of self is another post hoc “technique of neutralization” that allows perpetrators to distance themselves from responsibility for their violent actions (Anderson 2017:46) or whether, as Bastian, Denson, and Haslam (2013) suggest, it is a way for perpetrators to accept responsibility and express remorse by describing themselves and their past conduct as inhuman. It is also possible that perpetrators undergo a subjective transformation in the process of becoming killers as they stop caring, as Alphonse said, about life or death. Reflecting on this shift, Alphonse described himself as having “become a beast without noticing it,” Ignace remarked that “[we] no longer were what [we] had been,” and Léopord and Pio, while admitting their agency in killing, struggled to recognize who they became the more they killed (Hatzfeld 2005: 49, 47). In another set of interviews with men who killed in Rwanda, one described how it felt like “the enemy Satan . . . moved into people and changed their hearts, and their hearts became like animals” (Lyons and Straus 2006:86).
Similar quotes exist in evidence from the Holocaust and the Second Sino-Japanese War. For example, Nazi SS officer and commandant of Auschwitz, Rudolf Höss (2001:147–48), who described in his memoirs his relief at the establishment of gas chambers to kill Jews after having observed “many members of the Einsatzkommandos [go mad and commit suicide],” remarked to a friend in 1944 how he himself, after initially struggling to monitor the murders, had “ceased to have any human feeling” (Fest 1970:278–79). Kubotera-san, who killed a Chinese mother and her infant, lamented similarly: “I lost my humanity” (Dawes 2013:70). These perpetrators, like the Rwandans, described themselves as having fundamentally changed in some way connected to their humanness as a result of their ongoing participation in violence. Yet only recently has research begun to investigate self-dehumanization (Kouchaki et al. 2018:1235), and this work does not consider how ongoing experience perpetrating harm may matter. In contrast, the results of this study suggest there is likely a correlation between routinization and habituation to violence, when people believe their core values regarding people and practices have changed, and when they feel they have lost their own sense of humanity. Indeed, research indicates that our values are what we consider to be most central to our selves (Stets and Carter 2012; Strohminger and Nichols 2014; Tavory 2011), and Goffman (1959:123, 141) asserts that changes in how individuals perceive themselves are part of the changes people experience as they adopt new moral careers. Future work would benefit from more systematic examination of how and whether ongoing participation in violence and the dehumanization of victims relates to dehumanization of self.
This leads to a final suggestion for research. Dehumanization scholars have long observed that dehumanizing rhetoric can take many forms; for instance, it may be mechanistic (objectification is one such example) or animalistic, in which a person or social group is portrayed as lacking human essence or uniquely human secondary emotions (Haslam 2006:255–60). In Rwanda, Tutsis were described as cockroaches and snakes. Nazi propaganda depicted Jews as vermin, rodents, parasites, and infectious diseases (Landry, Ihm, and Schooler 2022; Landry, Orr, and Mere 2022; Musolff 2007; Smith 2020). During the Second Sino-Japanese War, Chinese victims were referred to as logs (“maruta”; Gold 2019). Existing research on dehumanization and its connection to violence has done little to examine the meanings of these various rhetorical forms in their respective cultural contexts and how they may influence the cognitive, emotional, and relational processes involved in perpetrating harm. Future research would therefore benefit from examining how macro-level cultural conceptions of humanness influence the micro-level cognitive, emotional, and relational processes involved in killing when different forms of dehumanization are invoked (Leader Maynard and Luft 2023). Similarly, future research would benefit from examining how dehumanizing rhetoric matters at the micro-level of decision-making about violence. Such rhetoric is likely still relevant in genocide, but the assumption that civilians hear dehumanizing discourse, believe it, and become motivated to kill friends, family, and neighbors, is itself dehumanizing—in this case, it strips ordinary Rwandans of moral agency and the ability to make choices about violence, even if those choices during the genocide were horrific.
In turn, if we treat perpetrators as mindless machines or non-human animals incapable of moral reasoning and decision-making before violence has begun, then all hope for intervention is lost. After all, in spite of how pervasive dehumanizing rhetoric is in our world, genocide is not. Once we have a greater understanding of how dehumanizing rhetoric contributes to violence, we can better comprehend this paradox and determine where to direct our intellectual and practical efforts.
Conclusion
Genocide research and research on other forms of violence have long argued that dehumanization causes people to harm others. Yet in recent years, some have argued that there is little evidence that dehumanization motivates violence and that those who harm do so for different reasons: because they view violence as socially and morally justifiable or because they feel pressured to kill. This article contributes to the discussion by drawing on and extending sociological theories of culture and morality.
Research on culture and processes of becoming suggests that what initially appears to be a motivation may be a late narrative that helps actors cognitively adapt to new and challenging practices, which tend to also elicit strong emotional responses, although both dissipate over time as new practices become habit. Morality research suggests our emotions indicate evaluation: An action is considered good or bad when we have feelings about it; otherwise, it is morally irrelevant. I join together these insights in a temporal analysis of Rwandan genocide perpetrators’ recalled thoughts and feelings about killing Tutsi, and I find that as people adopt new practices—here, violent practices—their cognitive, emotional, and relational habits change.
Specifically, I examine how perpetrators recall perceiving Tutsi prior to, at the onset of, and as their participation in the Rwandan genocide became routine, and I trace how continued participation in violence led to a dulling of their initial cognitive and emotional discomforts with violence and a shift in their social perceptions of Tutsi. The more they killed, the more meaningless both killing and their victims became to them. Perpetrators dehumanized their victims over time.
These findings indicate that our evaluations of others need not dictate our behavior toward them. Rather, our social perceptions can change as a result of our actions and influence how we morally evaluate those same actions, in turn. Moreover, the shift in social perceptions that accompanies behavioral routinization and cognitive and emotional habituation to new acts is not due, at least in this case, to justificatory self-talk regarding social perception specifically or to explicit efforts to resolve the emotional difficulties associated with early forays into new practices. As a form of nondeclarative culture acquired through experience and exposure, social perceptions and the meanings of relationships transform unconsciously.
Consequently, this article contributes not only to research on dehumanization and violence but also to research on the temporal dynamics of experience and the sociology of morality: The process of acquiring new practices, even practices once thought and felt to be abhorrent, is inextricably, yet implicitly, tied to shifts in our social perceptions and relationships. The process of becoming a new person—a new moral self—involves the interweaving of cognitive, emotional, and relational transformations.
Supplemental Material
sj-docx-1-stx-10.1177_07352751231203716 – Supplemental material for The Moral Career of the Genocide Perpetrator: Cognition, Emotions, and Dehumanization as a Consequence, Not a Cause, of Violence
Supplemental material, sj-docx-1-stx-10.1177_07352751231203716 for The Moral Career of the Genocide Perpetrator: Cognition, Emotions, and Dehumanization as a Consequence, Not a Cause, of Violence by Aliza Luft in Sociological Theory
Footnotes
Acknowledgements
For their comments on revisions of this article, I am grateful to Rogers Brubaker, Clayton Childress, Jeff Guhin, Steve Hitlin, Jack Katz, Omar Lizardo, Jared McBride, Laura K. Nelson, Letta Page, Andrew Perrin, Gabriel Rossman, Abigail Saguy, Stefan Timmermans, and Ezra Zuckerman. I would also like to thank Paul DiMaggio, Carly Knight, Steven Lukes, and others at the NYU Sociology of Culture Workshop, Duke Political Science, Elisabeth Jean Wood, Louisa Lombard, Elizabeth Nugent, Jonathan Wyrtzen, and others at the Yale Political Violence and Its Legacies Workshop, IGR, Nina Eliasoph, Paul Lichterman, Josh Seim, Hajar Yazdiha, and participants in the USC Sociology Colloquium, the UC-Berkeley Sociology Colloquium, University of Minnesota Sociology, Ann Mische and participants at the Kroc Institute of Notre Dame, Monica Prasad, Ji-won Lee, and participants at the Northwestern Problem-Solving Workshop, and participants in The Allure of Violence conference at the Neubauer Collegium for Culture and Society at University of Chicago. Finally, this article is indebted to three anonymous reviewers who encouraged its connection between violence research and the broader theoretical arguments made within. All errors are my own.
Supplemental Material
Supplemental material for this article is available online.
Notes
Author Biography
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
