Abstract
Over (2021) presents seven challenges to a representation of the prevalent theoretical position about social dehumanization in social psychology. The article concludes with positive suggestions for understanding contemporary research and historical examples related to dehumanization. We, as authors, had different initial reactions to these arguments. Still, we ended up agreeing that although the positive suggestions are useful and lead to interesting questions for research, the objections are exaggerated.
Specifically, Over characterizes the target “dehumanization hypothesis” as follows: “(a) Victims of intergroup harm are perceived as being similar to nonhuman entities, and, as a result, (b) natural inhibitions against causing them harm are eroded, leading, in extreme cases, to genocide and torture” (p. proof 1). However, this is not the statement that the seven challenges actually aimed to refute, by and large. Here is our reconstruction of that statement, with the differences italicized: (a) Victims of intergroup harm, and nobody else, are explicitly affirmed as being identical to nonhuman entities, and (b) as an inevitable result of dehumanization, natural inhibitions against causing them harm are eroded, and all aspects of genocide and torture can be completely explained via this mechanism.
We know of no research on intergroup dehumanization that has endorsed such an extreme statement. Over’s objections target the extreme version but miss many nuances of the more reasonable one. We outline the seven challenges below.
Examples of in-groups likening themselves to well-regarded animals challenge the extreme claim that “victims of intergroup harm are perceived as being similar to nonhuman entities . . .” (p. proof 1) but not the actual hypothesis that focuses only on victims of harm. Indeed, it is possible for an in-group to ennoble (or brutalize) itself by comparing itself to a specific animal and to lower and brutalize the enemy group by seeing this group as less than human. The examples also do not refute findings about the relative prevalence of in-group and out-group dehumanization—most studies on this topic compare the humanity of the in-group with that of the out-group and find a relative difference without requiring the in-group to be seen as fully human. And finally, although some individual species may be seen as admirable, the general category “animal” and indeed most animal names are not, we suspect, compliments when applied to people. An exception proves the rule: The generic “animal” becomes grudgingly positive when traits seen as uniquely human, such as inhibition or mercy, must be cast off to achieve some goal (e.g., “fight like beasts,” “party animals”).
Showing that victims of intergroup violence are sometimes, even often, described in human (Challenge 2) and mentalizing (Challenge 4) terms refutes the extreme claim that they are seen as identical to animals but not the actual claim that they are seen as “similar.” In fact, all of the research and theories reviewed in the article specify only one or two dimensions of perceived difference as crucial, be they essence, mental development, mental traits, or emotional experience. Here, too, the argument becomes perilously unfalsifiable when explicit descriptions of humans as animals are taken as evidence that humans are actually not seen as animals—“There is no sense in consistently reminding a rat that it is, in fact, a rat” (p. proof 4-5).
Rejecting implicit group-animal/human associations as evidence for dehumanization refutes only the phantom claim that dehumanization must be explicit. First, some evidence for everyday intergroup dehumanization is explicit or “blatant” in nature (Kteily & Bruneau, 2017). In addition, as Over acknowledges, the function of an associative metaphor is to bring to mind information in line with the metaphor, whether or not its proposition is actually endorsed. If a group of people are associated with apes, this association facilitates the same kind of judgments of them as mentally and culturally subordinate, more violent, and so on, as if the similarity were explicitly believed. Research on implicit associations (e.g., Goff et al., 2008) supports this point by connecting such associations to more explicit forms of discrimination. The absence of an explicit endorsement, whether sincere or motivated by social presentation, does not undermine the reality of the phenomenon.
See Challenge 2 above.
Again, the specific nature of most actual theories of dehumanization allows for people to be seen both as belonging to human categories “criminal” or “avaricious” and as being similar to animals in other ways. In fact, there is no other way to interpret the majority of studies on intergroup dehumanization that start from the premise of group labels that apply only to humans (e.g., pedophiles and terrorists in Giner-Sorolla & Russell, 2019, or the national and racial labels in many other published studies), making it absurd to hold total dehumanization as a standard. In addition, dehumanization has seldom been claimed to be the motivation behind every act of cruelty. In fact, Bandura (1999) outlines numerous other mechanisms of moral disengagement that can both facilitate and justify post hoc such acts, such as denying responsibility or using euphemistic language. These mechanisms disengage the conscience in many more ways than simply putting victims outside of moral concern by viewing them as less than human.
To the extent that this objection is distinct from Challenge 5, it points out that some forms of cruelty and discrimination are aimed at harming people through their human faculties of imagination, self-regard, and so on. This objection, too, aims for the phantom point that total dehumanization is necessary to explain all facets of crimes against humanity. But unlike the previous ones, this challenge does successfully hit an assumption of many theories that is not explicitly tested: the assumption that human-like mentalization is a sufficient, active ingredient for activating the special moral concern that is attached to the human species. It questions, in other words, whether all dimensions of perceived humanity bear equally on moral treatment.
As in Challenge 6, Over argues only for a plurality of grounds for human status, as indeed the theories reviewed provide, rather than against the importance of human status per se. However, Over does succeed in questioning whether human mental traits are necessary to the moral consideration that is attached to the human category. Going even beyond the provided example of a baby, the moral status of a human fetus or embryo critically depends on whether it is seen as a human life. It is this essential interpretation of human “similarity” that often intervenes in morality to arouse concern for individuals whose mental status is not at all prototypical.
Although these seven lines of attack mostly miss the point of actual dehumanization theory and research, the positive mechanisms proposed near the end of the article are welcome. They demand clarity in a field of research that often confuses several concepts under the label dehumanization. We now share our thoughts on two more specific implications for research going forward.
Dehumanization as a Matter of Prototype Categorization or Decategorization
The idea that human beings are dehumanized by being completely excluded from the human category or by being treated in every respect like animals has never been plausible. Rather, a more useful concept is that groups of humans might be judged as nonprototypical, metaphorically likened to objects or animals less worthy of moral consideration (rather than to angels or lions) and therefore subject to inferior treatment. This corresponds to the dominant prototype models of categorization in cognitive psychology (e.g., Rosch, 1975), in which decisions operate via similarity and centrality of exemplars to a set of prototypical traits rather than via an all-or-nothing categorization.
For example, although antisocial characteristics such as cunning and villainy may be seen as typical of humans more so than animals, they may also indicate that their possessors are not prototypical humans and therefore lack a number of other typical human characteristics that Haslam (2006) and Leyens and colleagues (2000) elaborated in their respective research, such as conscience or capacity for reflection. The prototype model leads to caution about equivalence between these human-defining traits because a person can be seen as human on one of the following dimensions yet dehumanized on others:
Human-essential category (vs. all nonhumans; as measured by blatant identification as human and by associations to the human concept)
Human mental complexity (vs. animals; as measured by human-uniqueness traits and emotions)
Human (and animal) capacity for spontaneity and creativity (vs. inanimate objects and machines; as measured by human-nature traits)
Human (and animal) capacity for suffering (vs. insensate things)
Human prosociality (vs. villains or “demons”)
Whether any given nonprototypical status entails degradation along other dimensions of humanity is an empirical question. That is, a “criminal,” although acknowledged as having some amount of mind and intentionality, may or may not be seen as similar to nonhuman entities such as animals (e.g., a wild dog), intelligent but emotionless machines (e.g., the Terminator), or more fanciful examples of antisocial nonhumans such as demons, zombies, or monsters. This also calls for empirical tests, rather than assumptions, of whether each departure from prototypical humanity undermines the granting of positive moral rights (e.g., citizenship) and negative ones (e.g., freedom from harm). In this regard, studies that link lower levels of different prototypical human traits to differential diminutions of moral responsibility and suffering (e.g., Bastian et al., 2011) are admirable if by no means completely resolved.
Recall, however, that for all of these dimensions to be meaningfully called dehumanization, they should be demonstrably linked to a reluctance to include target individuals and groups in the human-essential category, the first category that we have listed. Otherwise, in line with Over’s suggestions, it would be more parsimonious simply to speak of the denial of emotional experience, mental capacities, and so on, rather than to use the blanket term dehumanization.
If the different elements of the human prototype were shown to be inherently connected to each other, then even applying positive animal metaphors to the in-group would carry added drawbacks. Perhaps the purpose of these metaphors is to dehumanize the in-group in selected, very specific ways—to promote and advertise the freedom from remorse and inhibition ascribed to animals in the specific context of winning a sports game or a war. All the same, these seemingly complementary metaphors could create associations of the in-group with inhumanity that have paradoxical effects, such as lowering the impulse to cooperate with other in-group members.
Dehumanization Can Only Be Partial in a “Humanized” World
Under a prototype model of the human category, the multiplicity of mental dimensions makes less likely the extreme case of dehumanization in which perceivers deny victims their minds (or “human essence”) altogether and view them only as bodies or meat, such as when consumers perceive animals that they eat to possess a less complex mental life than other domestic animals (Bastian et al., 2012). There are more examples for such denials of minds during wartime (e.g., militaries referring to a “body count” to keep track of the number of enemy soldiers killed). However, even a focus on the physical body does not often lead to full dementalization but rather to a differential perception of certain mental states (Gray et al., 2011). Even in seemingly extreme cases of dehumanization, people still do perceive minds and mental characteristics in victims.
Indeed, most cases of dehumanization probably result from some kind of flexible mind perception and ascription, as Over points out. After all, humans are animals. As Over rightly states, comparisons to nonhuman entities are not reserved for out-groups (Challenge 1, p. proof 3-4). Consequently, dehumanization seems to rest on the more fundamental phenomenon of ascribing minds to nonhuman entities in the first place, a phenomenon referred to as anthropomorphism (Epley et al., 2007). Children (Hood et al., 2012) and adults (Forstmann & Burgmer, 2015) view minds (both of animals and humans) as distinct from the physical world, and we cannot help but see them everywhere.
People therefore seem rather inclined to impose at least some dimensions of humanity rather than neglect it. However, Over writes that “It would be extremely unusual, and most likely inappropriate, to describe an animal or a machine as disloyal or spiteful, for example” (p. proof 6). But who has never yelled at their computer or another piece of malfunctioning equipment that was (presumably intentionally) “messing with us”? Likewise, who has not laughed at one of the countless videos on YouTube showing how loyal dogs are and that cats could not be any more disloyal and indifferent? People are apparently more likely to rely on the idea of lesser minds, or lesser persons, than to completely abandon mind and person concepts when dealing with nonhuman—or dehumanized—beings.
To conclude, we think the field of dehumanization research has been overdue for a focus on its fundamental terms and assumptions. Although the challenges presented by Over do not destroy the evidence for dehumanization, they do force its proponents to be more clear about what they mean and to do more work to test the coherence of the human category’s many facets. A dimension that forms part of the human prototype may or may not connect directly to categorizing those who lack it as “less human” in essence. And given the ubiquitous occurrence of mind perception, it may be hard to see any sufficiently complex agent as completely lacking in all human mental qualities.
