Abstract
We read “Seven Challenges for the Dehumanization Hypothesis” (Over, 2021; this issue) prepared to be challenged. We are dehumanization researchers and therefore have a vested scholarly interest in the importance of this construct. Over suggests that there is reason to question the dehumanization hypothesis—which refers to two hypotheses, actually: (a) out-group members are viewed as nonhuman, and (b) this underlies the harm inflicted on them. We read her review and discussed her claims. For the most part, we do not disagree. But we also did not feel especially challenged. If social psychology has taught us one thing, it is that there are conditions that underlie the phenomena we study. Exposure to violent media has been shown to trigger aggressive behavior, but only for people who have a tendency toward aggression in the first place (Bushman, 1995). People indicate a desire to torture Muslims, but only if they very strongly identify as Christians and perceive Muslims as threatening (Viki, Osgood, & Phillips, 2013). Out-groups are not always dehumanized, and dehumanization does not always lead to harm; under some conditions, however, it does. In our opinion, it is not surprising that the dehumanization hypothesis is not always true. Rather, our goal, and challenge, is to specify the conditions under which it is.
Our comments are presented in two parts. First, we review each of the seven points and consider whether they constitute a significant challenge to what we believe to be true about dehumanization. We then consider the conditions under which the hypothesis is true—that is, when out-group members are dehumanized and when this dehumanization leads to harm. We consider some moderators defining the circumstances under which this process is likely to occur, drawing from the perspective of terror-management theory and research on the objectification of women. In doing so, we are careful to distinguish between animalistic and mechanistic dehumanization, because it makes sense that the moderating factors depend on the category of nonhumanness with which an individual is associated.
Challenges for the Dehumanization Hypothesis
Challenge 1: Comparisons to nonhuman entities are not reserved for out-groups
Over points out that comparisons to nonhumans can sometimes be used in complementary ways, such as parents referring to their toddler as a “little monkey” or an athlete being compared to a machine. We are not at all surprised by these claims, for there is no reason to expect that comparisons to nonhuman entities are reserved exclusively for out-groups. Dehumanization researchers do not make this sweeping claim. Rather, the hypothesis suggests that, compared to in-group members (and others compared to the self), out-group members are more likely to be dehumanized, and controlled experiments support this (e.g., Haslam & Bain, 2007; Leyens et al., 2001). Thus, the observation that associations are sometimes made between in-group members and animals and/or objects does not constitute a challenge for the hypothesis. Later in this commentary, we offer conditions under which dehumanizing the self and one’s in-group may even be desired (Morris, Goldenberg, & Heflick, 2014). The point is that there are conditions that affect the likelihood of dehumanization. Out-group status is one such condition. We agree with Over that “the onus is on proponents of the dehumanization hypothesis to explain when outgroups are dehumanized” (p. 7).
Challenge 2: Out-group members are often described in ways that apply only to humans
Challenge 2 argues that out-group members are appraised negatively even when described in terms of human qualities. Over highlights the example of Nazi propaganda describing Jewish people as traitors and criminals, a characterization that makes sense only with respect to humans. Again, we do not believe the dehumanization hypothesis is really undermined by this claim. Dehumanization researchers do not argue that dehumanization is the only means of marginalizing a group or individuals. There is also no reason to expect that qualities associated with humanness are entirely good. Haslam, Bain, Douge, Lee, and Bastian (2005) have shown that the humanness and desirability assigned to traits operate independently. It is certainly possible that out-group members will be described in dehumanizing ways and also in negative ways that are specific to humans. Why not describe Jews as vermin and as criminals? However, there may be conditions that make one mode of derogation particularly well suited. The challenge, as we see it, is to specify these conditions.
Challenge 3: Being associated with a nonhuman entity is not equivalent to being seen as similar to that entity
Challenge 3 argues that associating an out-group member with a nonhuman entity is not equivalent to actually perceiving the out-group member as nonhuman. This challenge seems to be one of the least substantial, with only one paragraph written to justify it. This challenge also seems to misinterpret the basis of dehumanization research, or at least misrepresents it. Of course, we do not look at an out-group member and literally see an animal (or object) in place of a human; the empirical support showing associations between certain group members and animals should not be dismissed on this basis. Research on the objectification of women shows that the bodies of women (and not men), when sexualized, are perceived using the same cognitive processes as are literal objects, rather than person perception (e.g., Bernard, Gervais, Allen, Campomizzi, & Klein, 2012). The point is that when we associate (not literally equate) a person or persons with a nonhuman entity, they are at risk of being treated in a manner similar to an animal or object.
Challenge 4: Out-group members may be denied some mental states but they are frequently attributed others
Over’s fourth challenge argues that out-group members are not always dementalized and are often denied some mental states while being attributed others. She takes the position that “it is not clear why possessing certain mental states to a lesser extent should be conceptualized as dehumanization” (p.14). But to us it is clear. Both the stereotype-content model (e.g., Fiske, 2009) and Haslam’s (e.g., 2006) conceptualization of humanness argue for specific mental-state attributions as a function of the type of dehumanization occurring. It is not claimed by dehumanization researchers that when someone is dehumanized, they are stripped of all mental ability; rather, they are attributed fewer mental abilities, especially those specific to the type of “humanness” being denied.
Challenge 5: Out-group members are granted some uniquely human attributes
The fifth challenge seems to overlap with Challenges 2 and 4, arguing that out-group members are granted some uniquely human attributes. Here, Over more specifically suggests that out-group members are often characterized with antisocial traits and that antisocial traits are not meaningfully associated with humanness when characterized as that which distinguishes humans from animals and objects. Again, we do not take issue with pathways other than dehumanization for derogating an out-group. But these points raised by Over challenge us to consider when dehumanization would be especially likely, and effective, for out-group discrimination.
In this fifth point, Over also highlights the difference a reference point can make in assigning certain antisocial qualities to someone. That is, she suggests that whereas antisocial traits are not meaningfully associated with humanness when characterized as that which distinguishes humans from animals and objects, these antisocial characteristics could potentially be represented as a third dimension of humanness—that which distinguishes humans from angels. It is difficult to conceptualize the tangible consequences of such a third dimension, given that angels do not exist.
Challenge 6: Groups are often persecuted because of their perceived humanity
Over argues that although in-group members elicit protection and care as a result of their human status, out-groups may be at risk of harm as a result of being attributed certain human mental states (e.g., cunning, spitefulness). To support this, Over turns to research arguing that certain groups have been abused and humiliated because of their humanity, arguing that “perpetrators implicitly acknowledge the difference between humans and nonhuman entities” (p. 9) when it comes to the type of harm inflicted on out-groups. This challenge means to say that the dehumanization hypothesis does not stand because punishment, harm, and derogation of out-groups can occur when out-groups are perceived as human. However, the dehumanization hypothesis does not claim dehumanization to be the sole motivating factor behind the harm directed toward out-groups—just that it may play a role in certain cases.
It is indeed an interesting point that, under specific circumstances, aggression or harm that is intended to humiliate the target requires an acknowledgment of humanness (e.g., Gopnick, 2006; Livingstone Smith, 2011). Animals, objects, plants, and buildings are incapable of experiencing humiliation. Therefore, the extent to which humans are treated in dehumanizing ways with the intention of humiliation can be dependent on recognizing the unique human ability to feel that humiliation. Nazis chose to overcrowd their victims in train cars and force them to behave in subhuman ways (i.e., being forced to relieve themselves while crammed next to others) because they knew the mental anguish that this would cause them. Humans also overcrowd pigs and chickens meant to be used as food, but that does not mean that we intentionally dehumanize animals with the intention of causing them emotional distress, especially because animals are not humans in the first place. The challenge here is, again, to understand the contextual factors that determine when humanness can offer protection to harm versus when it facilitates it.
Challenge 7: Being seen as less than human is not necessarily a risk factor for harm
In Over’s final challenge, she argues that being seen as less than human is not, on its own, enough to constitute a risk of harm. While babies lack the capacity for complex thoughts, they are treated with care. However, given that babies are most likely in-group members, it is not clear how this example really supports the argument. To explain this challenge further, Over describes how some animals, particularly pets, and some inanimate objects (e.g., Notre-Dame Cathedral) are treasured and protected. Over uses this example to bolster the claim that being perceived as less than human is not what puts someone at the risk of harm. But in this example, it is not clear how animals or buildings could be stripped of human qualities, given that they are not human in the first place. This point also contradicts itself, whereby animals and objects are being anthropomorphized and attributed qualities of mind.
Supporting the Dehumanization Hypothesis: Under What Condition?
The challenges offered by Over center around evidence that in-groups are not always humanized and out-groups are not always dehumanized. In addition, the author highlights examples of dehumanization not leading to harm, and qualitatively different kinds of harm when targets are perceived as uniquely human. While we appreciate the exercise of deconstructing what we know about dehumanization, we do not believe that these arguments dismantle our understanding. It seems perfectly appropriate that dehumanization is not the only mode of devaluing a person and that an association with nonhumans or nonhuman qualities may be more or less threatening under certain conditions. The challenge, rather, in our opinion, is to understand when out-group (and in-group) members are at risk of being dehumanized and when this is especially likely to lead to harm.
In this next section, we consider moderators that affect dehumanization on the basis of out-group status, and the implications of dehumanization on potential risk of harm. But first, we distinguish between animalistic and mechanistic dehumanization more clearly, given that they have different implications for when dehumanization will occur and motivate harm.
Dehumanization Is Not a Singular Construct
Dehumanization, as a construct, has been contemplated for centuries. In Western philosophy, Immanuel Kant (1780/1963, 1781/1988, 1963) described dehumanization as the extent to which a human may be perceived as lacking dignity or rationality, especially in the context of sexual relationships. Karl Marx (1867/1967) argued that workers were dehumanized because of an industrial focus on worth on the basis of work (along with ensuing social repercussions, such as poverty and inhumane living conditions). Like Marx, Gruenfeld, Inesi, Magee, and Galinsky (2008) found that less powerful individuals were perceived more in terms of their instrumental uses—that is, humans in less powerful social positions were perceived as tools for use and consumption. This particular approach to dehumanization is especially important when considering the objectification of women, and the resulting dehumanization (e.g., Nussbaum, 1995), and is in line with dehumanization that equated humans to objects. Other dehumanization research has also examined the extent to which humans are equated or associated with animals, especially in the years following the Holocaust (e.g., Chalk & Jonassohn, 1990). This type of dehumanization can be both explicit, wherein targets of prejudice are directly compared to animals (e.g., Hodson, Kteily, & Hoffarth, 2014), and can be an implicit association with characteristics of animals (e.g., fewer uniquely human emotions, Leyens et al., 2000). We begin to see that a variety of factors give rise to the nuances of dehumanization (for review, see Haslam & Loughnan, 2014).
Until recently, dehumanization research did not parse the difference between these comparisons of humans to nonhuman entities. This presents an important stride in the dehumanization hypothesis: a dual model of animalistic and mechanistic dehumanization (Haslam, 2006). When individuals are animalistically dehumanized, they are stripped of the qualities that make them uniquely human and separate them from other living creatures. On the other hand, individuals who are mechanistically dehumanized are stripped of qualities of human nature that signify what it means to be alive. For the most part, Over does not make a conceptual distinction between animalistic and mechanistic dehumanization in her analysis of the dehumanization hypothesis. But, these two types of dehumanization carry unique implications and occur for different reasons. Therefore, it is critical that they are differentiated and discussed in terms of their unique consequences. Below, we will use selected social-psychological phenomena to discuss when these different types of dehumanization occur and why. Specifically, we highlight the objectification of women, and the lens of terror management theory as they apply to dehumanization.
I Am Not an Animal: Animalistic Dehumanization from a Terror-Management Perspective
In The Denial of Death (1973), Ernest Becker provides an explanation for why human beings appear to be so motivated to raise themselves above, and psychologically distance themselves from, the animal kingdom. That motivation stems from the undoubted realization that animals very clearly and inevitably die. Human culture, for Becker, is built on this death denial and denial of human creatureliness. We can see this illustrated in practically every corner of human culture; the civilized manner in which people prepare and eat their food, clean and shave their bodies, have sex, and even engage in warfare. These same ideas form the basis of theorizing about disgust (e.g., Haidt, McCauley, & Rozin, 1994), in which disgust reactions are conceptualized as a protest against reminders of the continuity between humans and animals and are the manifestation of an underlying fear of death. Terror-management theory (Greenberg, Pyszczynski, & Solomon, 1986) codifies these insights to develop an empirical framework for testing these assumptions, positing that if such reactions represent a protest against death, then these tendencies should be exacerbated by conditions in which thoughts of death are likely to be salient.
In experimental research, manipulations of the salience of mortality increase disgust in response to bodily products and animals (Goldenberg et al., 2001), along with discomfort with activities that are especially physical in nature (Goldenberg et al., 2006), compared with an aversive, non-death-related condition. In addition, reminders of especially animalistic bodily functions have been found to increase the accessibility of death-related thoughts (Cox, Goldenberg, Pyszczynski, & Weise, 2007). Under conditions in which thoughts of mortality are salient, people also reported reduced appeal of the physical aspects of sex compared with the uniquely human or romantic aspects (Goldenberg, Pyszczynski, McCoy, Greenberg, & Solomon, 1999). Goldenberg and colleagues (2001) also showed that reminders of mortality led to greater preference for an essay describing humans as distinct from animals; within a mortality salience condition but not in the control condition, the essay emphasizing differences from other animals was preferred to the essay emphasizing similarities. People are also more accepting of the idea that humans did not evolve from other animals when death is salient (Tracy, Hart, & Martens, 2011). Reminders of mortality have even been found to increase people’s support for killing animals (Lifshin, Greenberg, Zestcott, & Sullivan, 2017).
Thus, there is evidence that concerns about mortality contribute to a threat associated with animality (for review, see Heflick & Goldenberg, 2013) and therefore may extend to conditions under which out-group members—who are threatening from the perspective of terror-management theory—are animalistically dehumanized. Terror-management theory suggests that the fear of death motivates humans to construct systems of beliefs that create a sense of symbolic immortality, but different cultures and groups of people differ in their systems of beliefs (e.g., religion). Consequently, out-groups and their beliefs constitute a threat, especially when concerns about mortality are salient (e.g., Greenberg et al., 1990; see also Greenberg & Kosloff, 2008). People thus have a desire to place themselves above, and psychologically distance themselves from, out-groups in the same way they distance themselves from animals. This simultaneous desire to distance the self from animals while at the same time perceiving out-groups as a threat to a way of life can lead to the equating of out-groups to animals and can work to justify in-group attitudes, practices, and belief systems, and thereby excuse cruel behavior toward those who do not follow the same set of cultural rules.
Experimental research supports this position. In three studies conducted in Italy and the United States (Vaes, Heflick, & Goldenberg, 2010), people rated their own group as more uniquely human, relative to an out-group, when reminded of death. This humanizing of one’s in-group in response to death reminders resulted in reduced accessibility of death-related thought. These results all occurred while controlling for the valence of the traits, so again, it is not the positivity or negativity of the traits that is important but their relationship to the perception of unique humanness. Note that, in this study, the effects were driven entirely by humanization of the in-group; the out-group across three studies was not significantly deprived of any humanness, only relatively so. This could be because the Vaes and colleagues (2010) studies did not include out-groups that are typically disparaged by their Italian and American participants (e.g., the Japanese and British out-groups were actually construed as possessing more desirable traits than the in-group) or that are significant threats to their belief systems, suggesting another important moderator of the dehumanization hypothesis that concerns whether the out-group is threatening. This example also highlights the point that while an out-group may be dehumanized, dehumanization on its own may not be sufficient to encourage harm.
More recently, a study provided clear evidence of out-group dehumanization as a function of mortality salience (Sanchez & Garcia, 2016). This study, conducted among university students in Madrid, assessed the attribution of secondary emotions to Muslim terrorists. This particular study employed infrahumanization, an extension of dehumanization that involves implicit, more subtle denial of human qualities (Leyens et al., 2001; Leyens, Demoulin, Vaes, Gaunt, & Paladino, 2007). The operationalization of infrahumanization is based on the idea that secondary emotions (e.g., shame and guilt) are considered to be specific to humans and are not experienced by other animals. Critically, this study identified another moderator. Only individuals who identified as being right-wing politically responded in this manner. Left-wing individuals responded to mortality salience by infrahumanizing Muslim terrorists less. Therefore, the motivation to deny human qualities to an out-group depends on the psychological mind-set of the perceiver, in conjunction with a preestablished belief system. It is that belief system, serving as a moderating factor, that can dictate whether an out-group is perceived as threatening.
The take-home point here is that moderators are important. The tendency for an out-group to be dehumanized can depend on the salience of concerns about mortality, the out-group in question, and preexisting systems of beliefs. Not all out-groups are equal—a baby out-group member is likely not as threatening as an out-group terrorist. Out-groups with more glaring cultural differences may likewise be perceived as more threatening. In addition, it is important to consider individual differences, such as political orientation, religious affiliation, and others.
Immortal Objects: Mechanistic Dehumanization from a Terror-Management Perspective
In contrast to the unambiguously negative associations with being an animal, being an object may have redeeming value when considered from the context of terror-management theory. Humanness, despite evidence that people are motivated to imbue the self and one’s in-group with more of it than others (e.g., Haslam & Bain, 2007), has one striking flaw—humans (and animals) are mortal. Objects, being that they are not alive, are not subject to the whole problem of dying. Therefore, it makes sense that under certain conditions, when concerns about mortality are especially threatening, people may be motivated, perhaps counterintuitively, to humanize themselves and their in-group less.
We see numerous examples of dehumanization in research focusing on the objectification of women. Theorizing from this area of research defines objectification as the process whereby women are reduced to instruments and seen as simply body parts meant to be used by others (Fredrickson & Roberts, 1997). From this perspective, objectification can cause women to be mechanistically dehumanized: Women are equated not with animals but with objects. This makes sense, particularly in the context of a death reminder, because as we discussed before, objects do not die. Thus, seeing the self or another human in this light has the potential to be protective rather than demeaning.
In a series of studies, Morris and colleagues (2014) demonstrated that mortality reminders lead women, but not men, to mechanistically dehumanize themselves in response to reminders of women’s role in reproduction (i.e., menstruation, pregnancy, and breastfeeding). This research also provided evidence that mechanistic dehumanization offered a sort of protection against fears associated with mortality. The more participants mechanistically dehumanized themselves (following mortality and breastfeeding priming), the less time they spent contemplating their thoughts related to their own death. The point here, once again, is that the psychological motivation underlying acknowledging humanness can be affected by contextual factors. What may appear on the surface to constitute a challenge to the dehumanization hypothesis may be, instead, a failure to consider the psychological motivations being served by humanness, or a lack thereof.
Dehumanization in response to mortality salience is not something women just do to themselves. Women are mechanistically dehumanized (i.e., objectified) by men, and this occurs more so when mortality is salient. In this case, men are dehumanizing an out-group, not necessarily or solely to demean them, but perhaps to protect themselves. Morris and Goldenberg (2015) provided evidence that mechanistic dehumanization—associating women with (nonmortal) objects—reduces the threat associated with men’s animalistic attraction to women. Men and women were exposed to advertisements that featured sexually provocative women (e.g., in a revealing swimsuit) who were either merged with an object (e.g., a Budweiser ad in which a woman was depicted in such a way that it appeared she was part of the label on a bottle of beer), or depicted as separated from the object. Men, but not women, responded to a mortality salience manipulation with lower ratings of how attractive they found the women who were not merged with objects (see also, Landau et al., 2006). In contrast, men’s attractiveness ratings for the merged (i.e., literally objectified) women increased when mortality was salient.
In another study (Goldenberg & Morris, 2016) men and women rated the attractiveness of real women and men, or female and male literal objects—that is, mannequins—after being reminded of their mortality. Heterosexual men rated the female mannequin, but not the real woman (or either male target), as more attractive after being primed with mortality. The prime did not affect the ratings that heterosexual women gave to the targets.
Here, we agree with Over’s suggestion that in-groups can also be dehumanized. Women dehumanize themselves by equating themselves with objects in a manner that serves a protective psychological function against the terror of mortality. This mechanistic dehumanization also occurs when women are the out-group, for the same psychologically defensive purpose. Again, though, we contest Over’s efforts to discredit the dehumanization hypothesis by introducing contextual factors that influence how and why dehumanization occurs.
Implications for Harm: Again, It Depends
We have laid out a framework for different contextual variables and factors that prompt animalistic and mechanistic dehumanization. Thus, we do not imply that dehumanization is the end-all, be-all explanation for harm afflicted on out-groups or that dehumanization cannot be positive in some ways. It is especially important to consider the type of humanness that is being considered and the function it serves, as well as how this might translate into a willingness or an eagerness to harm. Relying on this dual dimension of humanness framework, research suggests that when a woman is dehumanized in a mechanistic manner, she is perceived as less capable of feeling pain (e.g., Morris, Goldenberg, & Boyd, 2018) than a woman dehumanized in an animalistic manner. This could lead to a higher risk of harm, but not necessarily intentionally (e.g., using too much force without realizing it might be painful). This is not, therefore, an intentional manner of harming someone because of their dehumanization. If a woman can’t feel pain, why hurt her?
In contrast, when an individual is animalistically dehumanized, this may set the stage for more intentional harm. Research supports this, showing that a sexually objectified woman elicited more aggression than a nonobjectified woman, even in the absence of any other provocation (Vasquez, Ball, Loughnan, & Pina, 2018). Building on this, past research indicates that sexually objectified women are perceived as perfectly capable of feeling pain (Morris et al., 2018), perhaps indicating that when sexually objectified (and therefore animalistically dehumanized) women are aggressed against, it is with the intent to harm. It also may be that the model works in reverse, where dehumanization occurs as a consequence, or justification, for harm after the fact, and it is also possible that the relationship is cyclical.
Over also highlights examples in which humanness can pave the way for an even greater tendency for harm, when the goal is to humiliate the target. This is an interesting counterpoint, and one that has been recognized in other commentaries (e.g., Gopnick, 2006; Livingstone Smith, 2011). This suggests that under conditions in which the goal is to humiliate a person, there may be a motive to preserve some degree of humanness. Perhaps just enough so as to be considered “less than human,” but not “nonhuman,” for animals and objects cannot be humiliated.
Conclusion
In sum, Over highlights some important points to consider related to ideas about humanness and harm, but we are not convinced that it is time to pick another hypothesis to study. Rather, the challenges presented by Over highlight important questions that can hopefully pave the way for a more refined model of dehumanization. In short, we believe that our challenge as dehumanization researchers is to develop a better understanding of the conditions that lead to dehumanization and when this can put individuals at risk of harm.
