Dehumanization is widely thought to occur when someone is treated orregarded as less than human. However, there is an ongoing debate abouthow to develop this basic characterization. Proponents oftheharms-based approach focus on the idea that to dehumanize someoneis totreat them in a way that harms their humanity; whereasproponents ofthe psychological approach focus on the ideathat to dehumanize someone is tothink of them as less thanhuman. Other theorists adopt a pluralistic view that combines elementsof both approaches.
In addition to explaining different views on what it means todehumanize someone, this article focuses on related issues, such ashow to resolve the so-called “paradox of dehumanization”;the causes and consequences of dehumanization; the sorts of contextsin which dehumanization typically occurs; and the relation betweendehumanization and objectification.
Unsurprisingly, ideas about what it means to dehumanize someone haveco-evolved with ideas about what it means to be human (Kronfeldner2018, 2021b). Contemporary ideas about what it means to be human areoften influenced by the varieties of humanism that emerged during theRenaissance and Enlightenment. During this period, some humanistsemphasized the idea that to be human is to possess a certain sort of“human dignity”. Others focused on the idea that to behuman is to possess certain “natural” rights, such asliberty and equality, where these rights were thought to be groundedin distinctively human capacities, such as autonomy and rationality(Copson 2015). If one adopts these ideas about what it means to behuman, then it is natural to think that dehumanization occurs wheneversomeone’s dignity or natural rights are violated. This way ofthinking about dehumanization is reflected in various versions of theharms-based approach (see§2); as well as those versions of the psychological approach according towhich dehumanization involves thinking of someone as lacking agency orrationality (see§3.2).
Another way in which ideas about humanness and dehumanization haveco-evolved concerns modern ideas about biological humanness.Darwin’s discovery that human beings and nonhuman animals have ashared ancestry led to spurious ideas about certain groups of peoplebeing less than fully human in a biological-evolutionary sense: theseideas are a key element of the pseudoscientific view known as“scientific (or biological) racism” (Steizinger 2021).Some proponents of the psychological approach have developed ways tomeasure this form of dehumanization and have found that it remainsalarmingly prevalent (see§3.3).
Academic research into dehumanization has been thriving for decades.It has become highly interdisciplinary, encompassing fields such asphilosophy, social and developmental psychology, neuroscience,political science, sociology, peace and conflict studies, genocidestudies, and gender studies.
The models of dehumanization developed by social psychologists havebeen especially influential. An initial wave of models emerged duringthe latter part of the twentieth century (Kelman 1973; Bandura,Underwood, & Fromson 1975; Staub 1989; Struch & Schwartz 1989;Bar-Tal 1989; Opotow 1990; Bandura, Barbaranelli, et al. 1996).Proponents of these models often drew on qualitative analyses ofextreme violence, such as the Holocaust and the Mai Lai massacre.During the early 2000s, a “new look” emerged, with modelssupported by lab-based studies: these models were also geared moretowards capturing the “subtle” and “implicit”forms of dehumanization that purportedly occur in everyday contexts(see§3.2–§3.3).
Within philosophy, influential discussions of dehumanization began toappear during the mid-twentieth century. Arendt (1951, 1963, 1977)discussed dehumanization’s role in war, genocide, totalitarianregimes, and colonialization (see also Levi 1947 [1959], 1986 [1988]).Other influential works during this period focused on dehumanizationin the context of gender oppression (de Beauvoir 1949; Dworkin 1981,1987; LeMoncheck 1985, 1997; MacKinnon 1987, 1989, 1993; Nussbaum1995a, 1995b). Today, dehumanization is studied in various subfieldsof philosophy. This has led to different views on what dehumanizationis, and what a theory of dehumanization can explain.
Proponents of the harms-based approach argue that to dehumanizesomeone is to treat them in a way that harms or diminishes theirhumanity.
The harms-based approach often figures in feminist accounts of sexualobjectification (see entry onfeminist perspectives on objectification). MacKinnon (1987, 2006) and Dworkin (1981, 1985, 1987) defend theKantian view that certain capacities, such as autonomy andsubjectivity, are central to being human, and that to underminesomeone’s ability to exercise these capacities is to“objectify” or “dehumanize” them. Forinstance, Dworkin (1985 [2000: 30–31]) remarks that when a humanbeing is treated in a way that does not respect their autonomy, theyare “turned into a thing or commodity”, and that“those who can be used as if they are not fully human are nolonger fully human in social terms; their humanity is hurt by beingdiminished” (see also de Beauvoir 1949; LeMoncheck 1985; Dworkin1987 [1997: 140–141]; Gardner & Shute 2000; Cudd 2006).
Nussbaum (1995a) defends a similar view. She distinguishes betweenpermissible and impermissible forms of objectification, identifyingdehumanization with the latter, but not the former (e.g., see 1995a:276). Permissible cases of objectification involve treating someone“as an object”, while respecting their humanity.Impermissible cases involve treating someone “as anobject”, but without even tacitly respecting their humanity.Nussbaum takes treating someone “as an object” to involveany, or all, of the following: instrumentality; denial of autonomy;inertness; fungibility; violability; ownership; and denial ofsubjectivity. Even though Nussbaum repeatedly emphasizes thatobjectification is a way oftreating someone, in certainplaces she explicitly endorses the disjunctive thesis thatobjectification involvestreating and/orseeingsomeone as an object (e.g., see Nussbaum 1995a: 251). This suggeststhat Nussbaum holds a pluralistic view that includes elements of boththe harms-based approach and the psychological approach (see also Cudd2006: 165–166).
Elsewhere, Nussbaum (1995b, 2000) develops a version of humanistfeminism according to which there are certain capacities that arecentral for “truly human functioning” (2000: 6). Sheargues that exercising these capacities to a certain degree isrequired for leading a “fully human life” (2000: 74). Someof these capacities include bodily integrity, bodily health, practicalreason, emotions, and affiliation (2000: 78–90). On thisaccount, then, to dehumanize someone is to inhibit their ability torealize these capacities because doing so undermines their ability tolead a “truly human life”.
According to Mikkola’s version of the harms-based account, anaction is dehumanizing if and only if “it is an indefensiblesetback to some of our legitimate human interests, where this setbackconstitutes a moral injury” (2016: 145; see also Mikkola 2021).“Legitimate human interests” include those that (i) arewidely shared by members of our biological species (Homosapiens), and (ii) contribute to the basic well-being of membersof our biological species, such as “the ability to setone’s own life plan” (2016: 165); “the continuanceof life” (2016: 168); and the ability “to enjoy andmaintain friendships” (2016: 168). A setback to one of theselegitimate human interests constitutes a “moral injury”when it damages the “realization and acknowledgement of theperson’s value” (2016: 170). For example, Mikkola suggeststhat telling a white lie need not count as dehumanization, becauseeven though it may count as morally wrong on other grounds, it may notset back any legitimate human interest, or it may count as a morallypermissible setback (2016: 174).
Killmister (2024) argues that dehumanization is constituted by“actions and practices that function to erode people’smembership in the socially constructed category of the human”(2024: 2). Killmister (2024: 3) draws on Searle’s (1995) viewthat social categories are constructed and maintained by thecollective acceptance of two kinds of rules:constitutiverules anddeontological rules (see also Rust 2021 andthe entry onsocial ontology). Constitutive rules specify the conditions an entity must satisfy tocount as a member of the given social kind, while deontological rulesspecify the powers the entity must possess. Killmister suggests thatin contemporary liberal societies, being aHomo sapienscounts as a constitutive rule for membership in the social kindHUMAN, while having human rights counts as a deontologicalrule for membership (2024: 3). Killmister argues the social categoryHUMAN is also constructed and maintained by two other sortsof factors: (i) broad “norms of treatment”, such as havingone’s death marked in some way; and (ii) “the socialimaginary” which encompasses “the stories that we tellabout our shared humanity and our self-understanding as human”(2024: 4). Given this framework, Killmister argues that there arethree forms of dehumanization. “Constitutivedehumanization” involves actions that are intended to underminea constitutive rule for being human (Killmister 2024: 5–6);“corrosive dehumanization” involves attempting to modifythe constitutive rules indirectly “by challenging who thedeontological rules and norms of treatment apply to” (2024:6–9); and “hermeneutic dehumanization” occurs when“the social imaginary forecloses certain ways of beinghuman”, such as being a human with a non-binary gender (2024:9).
There is a key difference between Killmister’s version of theharms-based approach and those outlined above. For Killmister,“dehumanization” refers to attempts—successful orotherwise—to undermine someone’s status as“human” by influencing the criteria for membership in thesocial kindHUMAN. For other proponents of the harms-basedapproach, dehumanization does not necessarily involve an attempt toreconstruct the social kind,HUMAN: rather, it involves asuccessful attempt to undermine someone’s humanity by inhibitingsome of their distinctively human capacities or interests.
Some philosophers have raised general concerns with the harms-basedapproach (for some critiques of specific versions, see Guenther 2012;Mikkola 2016, 2021; Killmister 2024). Smith (2021, 9–29) arguesthat an adequate conception of dehumanization should distinguishbetween dehumanization itself and the sorts of harms that it causes.Given that proponents of the harms-based approachidentifydehumanization with certain sorts of harms, Smith argues that they donot leave adequate room for this distinction.
Insofar as proponents of the harms-based approach hold that certaincapacities are “essentially human”, they are vulnerable toanti-essentialist arguments concerning human nature (see the entry onhuman nature). Consider the claim that some capacities are “essentiallyhuman” in the sense that they are unique to and shared by allhuman beings. Various theorists have argued that there are no suchcapacities: instead, there is a high degree of genetic and phenotypicvariation within our species, which is required for evolution bynatural selection to occur (Hull 1978, 1986; Kronfeldner, Roughley,& Toepfer 2014; Kronfeldner 2018). Anti-essentialism thereby castsdoubt on the claim that dehumanization involves inhibiting“essentially human” capacities or interests.
Some proponents of the harms-based approach address this sort ofconcern head-on by explicitly rejecting the type of essentialismoutlined above. For instance, Nussbaum (1995c: 88) argues that whichcapacities are essential to being human is determined by our norms andvalues, as opposed to “matters of natural scientific fact”(for a critique of Nussbaum’s essentialism, see Antony 2000).Mikkola (2016: 165–170) avoids essentialist claims about humannature altogether. She holds that dehumanization involves underminingsomeone’s “legitimate human interests” in certainways; however, for a legitimate interest to count as“human”, it need only be widely, as opposed touniversally, shared by members of our species.
Other arguments against the harms-based approach stem from the sortsof views defended by sentientists and posthumanists, who tend toresist the idea that what makes certain ways of treating other peoplewrong is that theirhumanity is being harmed or diminished(Crary 2021: 162–163). For instance, according to Singer (1975,2010), how one ought to treat another individual does not depend onwhether they are a human being per se, but rather on the intereststhey happen to possess, such as the interest in not suffering. On thisview, purportedly “dehumanizing” forms of mistreatment arenot wrong because they harm the victim’s humanity: they arewrong because they frustrate the victim’s interests in certainways, irrespective of species membership.
According to the psychological approach, to dehumanize someone is tothink of them as less than human. The literature on thepsychological approach contains a maze of overlapping ideas. Ingeneral, though, versions of this approach can be roughly sorted intotwo families: those that focus mainly oncategoricaldehumanization, and those that focus mainly ongradeddehumanization. Categorical dehumanization occurs when someone isjudged as not belonging to the categoryHUMAN; whereas gradeddehumanization occurs when someone is merely judged aslesshumanlike than certain other members of the categoryHUMAN.
Smith (2011, 2014, 2016, 2021) argues that to categorically dehumanizesomeone is to believe that they possess a “subhumanessence”. Drawing on research into psychological essentialism(see Neufeld 2022), Smith argues that the dehumanizer thinks of thesehuman and subhuman essences as having the following features: (1)Possession of them does not come in degrees; (2) they are inherited;(3) they are immutable (one cannot go from being subhuman to human orvice versa); and they ground certain generalizations (e.g.,“Subhumans are dangerous”). Importantly, Smith also arguesthat in paradigmatic cases of dehumanization, such as Naziantisemitism, dehumanizers often hold contradictory beliefs abouttheir victims: on the one hand, they think of them as subhuman; but onthe other hand, they cannot help but think of them as human due totheir humanlike appearance (Smith 2016; 2021: 206–255). Smithargues that these conflicting thoughts give rise to a feeling of“uncanniness” in the dehumanizer (see also§4.2).
Other theories of categorical dehumanization focus on the idea thatwhile dehumanizers often categorize their victims as human in abiological-species sense, they categorically deny that they are humanin a different, non-biological, sense. In what follows, these theoriesof categorical dehumanization will be referred to as “ambiguitytheories”.
For instance, Steizinger (2018) observes that some Nazi ideologuesdeployed two concepts of humanness: a biological-species concept, anda metaphysical concept of the human “race-soul”, whichthey saw as grounding the ability to form culture and a“collective identity”. Proponents of this Nazi ideologybelieved that Jewish people are categorically human in thebiological-species sense, but categorically nonhuman in themetaphysical sense.
Drawing on lab-based studies, Phillips (2022a, 2023) argues thatpeople tend to have two concepts of humanness: a descriptive concept,and a normative concept of “true” (or “real”)humanness. To be human in the descriptive sense is to be a member ofthe biological species,Homo sapiens, whereas to be human inthe normative sense is to possess a deep-seated, emotionally-laden,commitment to good moral values. For example, Phillips (2022a)presents evidence that people tend to agree that serial killers arenot “true humans”, even if they are human in a“biological sense”.
According to de Ruiter (2023a, 2023b), denying that someone is humanin the non-biological sense is a matter of failing to recognize thattheir capacity for subjective experiences is “a factor thatcounts against his or her mistreatment” (2023b: 82): in otherwords, it is to regard them as outside the sphere of moralconcern.
In contrast to the theories of dehumanization outlined above, thetheories devised by social psychologists tend to focus more ongraded dehumanization.
According toinfrahumanization theory (Leyens, Paladino, etal. 2000; Leyens, Rodriguez-Perez, et al. 2001), to dehumanize someoneis to perceive them as having an “incomplete humanessence”, where having an incomplete human essence involveslacking uniquely human capacities to some extent. In developing thisidea, Leyens and colleagues focused on the distinction between“primary” and “secondary” emotions. An emotioncounts as “secondary” if people tend to think of it asuniquely human: examples include guilt and nostalgia. In contrast, anemotion counts as “primary” if people tend to think of itas experienced by human beings and nonhuman animals alike: examplesinclude fear and pleasure. Thus, according to Leyens and colleagues,when we think of someone as highly capable of experiencing primaryemotions, but as relatively incapable of experiencing secondaryemotions, we have thereby “infrahumanized” them—thatis to say, we have perceived them as having an incomplete humanessence.
Thedual model builds on infrahumanization theory by drawinga distinction between two kinds of dehumanization:animalistic andmechanistic (Haslam et al. 2005). Toanimalistically dehumanize someone is to regard them as lacking, tosome extent, attributes that are considered “uniquelyhuman”. Haslam and colleagues found that people tend to regardattributes such as rationality, civility, refinement, maturity, andmoral sensibility as uniquely human. Thus, according to the dualmodel, when we regard someone as lacking these sorts of attributes tosome extent, we perceive them as relatively animal-like. Tomechanistically dehumanize someone is to regard them, to some extent,as lacking attributes that are considered essential aspects of“human nature” (even if these attributes are shared withnonhuman animals). Haslam and colleagues found that people tend toregard attributes such as emotional responsiveness, openness, warmth,agency, and emotional depth, as aspects of human nature. Thus,according to the dual model, when we regard someone as lacking thesesorts of attributes to some extent, we perceive them as relativelyobject-like or robotic.
Some researchers have developed models of dehumanization that focusexclusively on the denial of mental capacities (H. Gray et al. 2007;Epley, Waytz, & Cacioppo 2007; Waytz et al. 2010; Waytz &Epley 2012). H. Gray and colleagues (2007) found that people’smental-state attributions tend to vary along two dimensions:agency andexperience. To attribute agency tosomeone is to regard them as possessing the sorts of capacities thatwe associate with goal setting and planning, whereas to attributeexperience to someone is to regard them as possessing feelings,emotions, and other subjective experiences. Gray and colleagues alsofound that people tend to regard humans as more agentic than nonhumananimals, and as more capable of experience than inanimate objects.Thus, according to Gray and colleagues’agency-experiencemodel, this engenders two forms of dehumanization: one occurswhen we regard someone as relatively non-agentic; and the other occurswhen we regard someone as relatively incapable of subjectiveexperiences.
According to thestereotype content model, we tend to judgefellow humans along two primary dimensions: warmth and competence(Fiske et al. 2002). This differentiates outgroups into four clusters:(1) high warmth and high competence (high-high); (2) highwarmth and low competence (high-low); (3) low warmth and highcompetence (low-high); and (4) low warmth and low competence(low-low).
Harris and Fiske (2006) argue that only the members of low-low groups,such as homeless people, are dehumanized. They defend this claim bypresenting evidence that the medial prefrontal cortex (mPFC) isinactive when people view the low-low, but active in all other cases.Given that mPFC activation “reliably covaries with socialcognition”, Harris and Fiske interpret this as evidence that thelow-low are “not processed primarily as human beings”(2006: 849; see also Harris & Fiske 2011).
More recently, Fiske (2021) has developed a variant of the stereotypecontent model according to which there arethree forms ofdehumanization. As was claimed in the original version of the theory,one form occurs when someone is perceived as low-low. According toFiske, this form of dehumanization resembles animalisticdehumanization and involves feelings of disgust (2021: 248). Anotherform occurs when someone is perceived as competent, but cold. Fiskesuggests that this form resembles mechanistic dehumanization to someextent, and typically involves feelings of envy (2021: 248). Forexample, Fiske notes that high-achieving women are vulnerable to thisform of dehumanization. Finally, a third form of dehumanization occurswhen someone is perceived as warm but incompetent. Fiske argues thatthe elderly and the disabled are often subjected to this“pitying dehumanization” (2021: 249).
Some theorists have developed models ofperceptualdehumanization, with an emphasis on the idea that we sometimesvisually perceive others as less than human. For instance, Fincher andTetlock (2016) present evidence that people tend to process the facesof criminals in a relatively non-configural manner, meaning that theeyes, nose, mouth, etc., are not perceived together as a gestalt, butas separate features. Fincher and Tetlock found that the faces ofcriminals thereforelook less humanlike than the faces ofnon-criminals, which are processed in a more configural manner (seealso Fincher & Tetlock 2016; Hugenberg et al. 2016; Fincher,Tetlock, & Morris 2017).
Vaes et al. (2019) presented participants with images of scarcelydressed men and women, interspersed with doll-like avatars. Neural andbehavioral indicators suggested that participants were visuallyperceiving the scarcely dressed women and the doll-like avatars in asimilar manner. Vaes and colleagues conclude that objectified womenare visually perceived, not just thought of, as object-like.
Varga (2021) argues that we can visually perceive bothanimacy andmentality: the former involvesperceiving something as an animate being, while the latter involvesperceiving something as having a mind. Thus, he argues that whensomeone is visually perceived as lacking either animacy or mentality,they have been perceptually dehumanized.
Psychologists have also developed models ofimplicitdehumanization. Some have presented evidence that peopleassociate certain social groups with animals or animal-related terms(Viki, Winchester, et al. 2006; Goff, Eberhardt, et al. 2008;Saminaden et al. 2010; Rudman & Mescher 2012). For example, in aseries of studies, Goff and colleagues (2008) found some evidence thatin the United States, people tend to implicitly associate Black peoplewith apes. Other researchers have found that people tend to associatesecondary emotions more with their ingroup than with an outgroup(Paladino et al. 2002).
It is important to recognize that the term “implicit” isused in various ways by psychologists. The term “implicitmeasure” is often used to refer to those instruments, such asthe Implicit Association Test (Greenwald et al. 1998), that assesspeople’s thoughts or feelings in a way that is“indirect” (e.g., they do not rely on subjects’self-reports). On the other hand, some psychologists use the term“implicit” to refer to a distinctive type of psychologicalstate or process. Most notably, psychologists often claim that“implicit” states and processes are associative as opposedto propositional; and unconscious as opposed to conscious (see theentry onimplicit bias). In presenting evidence for “implicit dehumanization”,some researchers explicitly adopt both claims. For instance, Goff andcolleagues interpret their findings as evidence for “abidirectional association between Blacks and apes that can operatebeneath conscious awareness” (Goff, Eberhardt, et al. 2008: 304;see also Saminaden et al. 2010). However, in presenting evidence for“implicit dehumanization”, other researchers do not makeany claims about the associations in question being unconscious (e.g.,see Paladino et al. 2002; Viki, Winchester, et al. 2006; Rudman &Mescher 2012).
In developing the psychological approach, researchers have drawn onmultiple measures of dehumanization.
As was explained above, some researchers deploy measures ofdehumanization that are “implicit” in the sense that theyassess dehumanizing mental states in a way that is“indirect”. For instance, in assessing whetherparticipants in their studies associate Black people with apes, Goffand colleagues (2008) did not explicitly ask them whether theyassociate Black people with apes. Instead, participants completedvarious tasks that involved reacting to images and words pertaining toeither Black people or apes. Performance on these tasks was then usedto assess whether they implicitly associate Black people with apes. Incontrast, other measures are more “explicit” in the sensethat participants are asked to assess someone’s humanity in arelatively direct way.
Another influential distinction concerns “subtle” versus“blatant” measures of dehumanization. According to Kteilyand Landry (2022a: 225), a measure of dehumanization counts as“blatant” to the extent that an ordinary person would“readily recognize” that it is designed to assessdehumanization; whereas a measure of dehumanization counts as“subtle” to the extent that they would not (for a similarexplanation, see Haslam 2021: 134).
The most influential measure of blatant dehumanization has beendeveloped by Kteily and colleagues (2015). Their measure uses the“Ascent of Hu(man)” image, which depicts a series of“evolving” individuals, starting with some stereotypicallyape-like individuals on the left, through to a modernHomosapiens on the right. Using the image as a guide, participantsare asked to rate the extent to which the members of certain groupsseem “evolved” and “humanlike” on a scale from0 to 100. Given that it would be readily apparent to an ordinaryparticipant that they are being assessed on dehumanization, Kteily andcolleagues’ Ascent scale qualifies as a blatant measure.
To illustrate why some measures of dehumanization count as“subtle”, reconsider infrahumanization. In developinginfrahumanization theory, Leyens and colleagues (Leyens, Paladino, etal. 2000; Leyens, Rodriguez-Perez 2001) began by asking people to ratethe extent to which various emotions are uniquely human. As wasexplained above in§3.2.1, participants in these studies tended to regard certain emotions, suchas jealousy and nostalgia, as uniquely human. Then, inindependent studies, participants were asked to rate theextent to which the members of certain social groups exhibit thesesecondary emotions. Presumably, participants in the latter studies didnot readily recognize that they were being assessed fordehumanization: from their perspective, they were just being asked torate the extent to which various people experience emotions such asjealousy and nostalgia. The measures in question therefore qualify asmore subtle than blatant.
The distinction betweenimplicit versusexplicitmeasures is often regarded as orthogonal to the distinction betweensubtle versusblatant measures (e.g., see Kteily,Bruneau, et al. 2015; Kteily & Landry 2022a). The two distinctionsare also commonly regarded as spectrums, rather than strictdichotomies (e.g., see Haslam 2013, 77; Kteily, Bruneau, et al. 2015;Kteily & Landry 2022a).
Some theorists have raised concerns about theconstructvalidity of certain measures of dehumanization: an instrument isconstruct valid if and only if it measures the theoreticalconstruct that it is intended to measure. For example, if an item in aquestionnaire is designed to measure narcissism, but it actuallymeasures self-confidence, then this item lacks construct validity.
Enock and colleagues argue that the measures used by proponents ofboth infrahumanization theory and the dual model lack constructvalidity because they fail to distinguish between dehumanizingattitudes and the mere attribution of antisocial traits (Enock,Tipper, & Over 2021; Enock, Flavell, et al. 2021; see also Over2021; Enock & Over 2022). More specifically, they argue thatpeople do not, in fact, attribute uniquely human traits to ingroupmembers more than outgroup members. Instead, they present evidencethat people merely have a bias towards attributingprosocialtraits to ingroup members, andantisocial traits to outgroupmembers, regardless of whether the traits in question are perceived asuniquely human (for a direct reply, see Vaes 2023; and for somedefenses of the construct validity of other measures, see Kteily,Bruneau, et al. 2015; Bruneau, Jacoby, et al. 2018; Phillips2023).
Some researchers have addressed ethical concerns about Kteily andcolleagues’ Ascent of (Hu)man scale. Izydorczak, Grzyb, andDolinski (2022) assessed the hypothesis that engaging with the scalemightcause participants to see certain outgroup members asrelatively unevolved and ape-like: more so than they would haveotherwise. However, their study found no evidence of this.
Some theorists have raised relatively general arguments against thepsychological approach to dehumanization.
Smith (2023) critiques what he regards as the three major accounts ofdehumanization within social psychology: the dual model, the mindperception thesis, and the stereotype content model. First, he arguesthat they do not fit with paradigmatic cases of dehumanization, suchas Nazi antisemitism (2023: 2). For example, he argues that they areincompatible with the fact that Nazi ideologues regarded Jewish peopleas intelligent, rational, and highly competent. Second, he argues thatthese models neglect “the most egregious cases of blatantdehumanization” in which the victims are seen as“monstrous, demonic, evil beings” with “superhumanpowers” (2023: 2–3). Finally, he argues that denials ofhumanness are always categorical, whereas according to the models inquestion, they come in degrees (2023: 3). Smith (2014, 2023) arguesthat dehumanization is always categorical because it involvesattributing a subhuman essence to someone, and attributions ofessences do not come in degrees (for an extended argument thatdehumanization need not involve attributing a subhuman essence tosomeone, see Kronfeldner 2021c; see also Killmister 2024).
Some critics argue that proponents of the psychological approachconflate dehumanization with other negative attitudes (Over 2021;Bloom 2022; see also Rai, Valdesolo, & Graham 2018; Enock,Flavell, et al. 2021; Enock, Tipper, & Over 2021). For instance,Bloom (2022: 539) argues that in much of the psychological literature“the notion of dehumanization extends to seemingly any instancein which people think badly about another individual or group”.He goes on to suggest that this approach “stretches the notionbeyond recognition” (2022: 539; for some direct replies to thisargument, see Kteily & Landry 2022b; Phillips 2023).
Probably, the most influential critique of the psychological approachis that in purportedly central cases of dehumanization, such as Naziantisemitism, the victims were not persecuted because they were seenas less than human: instead, they were persecuted preciselybecause they were seen as human (Appiah 2008; Lang 2010,2020; Johnson 2017; Manne 2016, 2017: chapter 5; Bloom 2017, 2022;Rai, Valdesolo, & Graham 2017; Enock, Flavell, et al. 2021; Over2021; Enock, Tipper, & Over 2021; Enock & Over 2022; see alsoKuper 1989). For instance, Manne argues that dehumanization is not acentral feature of misogyny because misogynists see women as“potentiallydangerous andthreatening in waysonly a human being can be” (2016: 148). More generally, Mannesuggests that “only a human being can sensiblybeconceived as an enemy, a rival, a usurper, a subordinate, a traitor,and the like” (2016: 153). Relatedly, some theorists haveemphasized that the perpetrators in purportedly central cases ofdehumanization often see the victims as morally responsible for theirperceived crimes. For example, the lynching of African Americans wasfrequently cast as bringing the guilty to justice; and the Nazissubjected Jewish people to humiliating forms of punishment. This mightsuggest that the perpetrators in these cases saw their victims ashuman beings after all, for it would not make sense to humiliate andpunish a creature that is not human, such as a rat (Appiah 2008: 144;Lang 2010; Manne 2016, 2017; Bloom 2017; Over 2021).
The fact that perpetrators in purportedly central cases ofdehumanization think of their victims as having distinctively humanattributes is known as “the paradox of dehumanization”(Smith 2016).
As was discussed in§3.4, some theorists try to resolve the paradox by concluding that thepsychological approach to dehumanization has little explanatory power(Appiah 2008; Lang 2010; Manne 2016, 2017: chapter 5; Bloom 2017,2022; Over 2021; see also Jordan 1968). For example, Bloom (2022)remarks that “Nobody could doubt that we sometimes think of oneanother as less than human”. However, he goes on to argue thatthinking of someone as less than human is rarer than proponents of thepsychological approach tend to think (see also Bloom 2017; Over 2021;Leader Maynard 2022: 114). In developing this solution to the paradox,some theorists argue that dehumanizing language is usuallymetaphorical, and that it is merely intended to convey negative ideasabout the victims, such as the idea that they are disgusting anddangeroushumans (Over 2021; Manne 2016: 163–165).
Another way to resolve the paradox is to claim that dehumanizers oftenpossess contradictory beliefs about their victims, simultaneouslythinking of them as both humanand subhuman. In defendingthis solution, Smith (2016; 2021: 206–255) argues thatdehumanizers normally acquire the belief that their victims aresubhuman via propaganda and other ideological influences; whereas theyacquire the belief that their victims are human via the mechanismsresponsible for the visual detection of human faces (2016: 435).According to Smith, the co-occurrence of these inconsistent beliefsgives rise to a feeling of “uncanniness” in thedehumanizer, because the victim is perceived as a creature whotransgresses “culturally sanctioned metaphysicalcategories” (2016: 430). Smith goes on to suggest that when thevictim is seen as threatening, they are also seen as horrifying andmonstrous (he calls this “demonizing dehumanization”); butwhen they are not perceived as threatening, they are merely regardedas uncanny (he calls this “enfeeblingdehumanization”).
Ambiguity theorists attempt to resolve the paradox by arguing thatdehumanizers possess multiple concepts of humanity, simultaneouslyviewing their victims as “human” in one sense, but as lessthan “human”in a distinct sense (see§3.1.2). For instance, Phillips (2022a, 2023) argues that the Nazis regardedJewish people as biological humans, yet they denied that they are“real” or “true” humans in a distinctivelynormative sense precisely because they regarded them as essentiallyevil, possessing traits such as criminality and deceitfulness (seealso Steizinger 2018). Similarly, de Ruiter (2023a, 2023b) argues thatdehumanizers often see their victims asHomo sapiens withspecies-typical capacities for subjective experiences; however, theydeny that their victims are “human” in the sense that theyview them as lacking moral status.
Instead of resolving the paradox by holding that dehumanizers havemultipleconcepts of humanity, some theorists attempt toresolve it by arguing that dehumanizers can simultaneously recognizecertainaspects of an individual’s humanity, whilstalso denying others (Fincher, Kteily, & Bruneau 2018; Kteily &Landry 2022a; Leader Maynard & Luft 2023: 3). For example, thereis evidence that intelligence and good moral character are central topeople’s concept of humanness (see§3). Thus, a dehumanizer might simultaneously recognize an outgroupmember’s intelligence, but deny that they possess good moralcharacter—plausibly, this was how Nazi ideologues tended toperceive Jewish people.
In developing this approach, Kteily and Landry (2022a: 231–234)suggest that people have a Platonic concept of the “idealhuman”, and that the degree to which someone is dehumanized isdetermined by their perceived distance from this Platonic ideal.Perceived distance from the ideal human is determined by two factors:(1) which traits the dehumanizer perceives the target as possessing orlacking; and (2) how central each trait is to the dehumanizer’sconcept of ideal humanness. For example, suppose morality is morecentral to a dehumanizer’s concept of ideal humanness thanintelligence. Moreover, suppose that this dehumanizer perceives personA as intelligent, but immoral, and personB asmoral, but unintelligent. According to Kteily and Landry, then, thisdehumanizer will perceive both individuals as less than fully human,but they will perceiveA as less human thanB. Moreto the point, the dehumanizer will perceive both individuals as lessthan fully human, even though they perceive bothas humanlike incertain respects.
Finally, some theorists address the paradox by adopting a pluralisticview of dehumanization: that is to say, a view that combines elementsof both the psychological approach and the harms-based approach(Nussbaum 1995a, Kronfeldner forthcoming).
For instance, Nussbaum (1995a) argues that to dehumanize someone is toeither treat or see them as an object, without even tacitly respectingtheir humanity (see§2.1). Nussbaum does not address the paradox of dehumanization explicitly;however, she makes the following remark about a case of dehumanizingsexual objectification:
What is made sexy … is precisely the act of turning a creaturewhom in one dim corner of one’s mind one knows to be human intoa thing, a something rather than a someone. (1995a: 281)
Here, Nussbaum appears to be claiming that the victim is subjected todehumanizingtreatment, even though they are tacitlythought of as human. This suggests a view according to whichdehumanization sometimes occurs even when the dehumanizer recognizesthe victim as human, because they are still subjecting them todehumanizing treatment.
Kronfeldner (forthcoming) adopts a pluralist account according towhich there are multiple “forms” and “levels”of dehumanization. “Forms” of dehumanization includecategorical and graded denials of humanness. “Levels”include “behavioral dehumanization” (i.e., the type ofbehavior regarded as dehumanizing by proponents of the harms-basedapproach); “cognitive dehumanization” (i.e., thinking ofsomeone as less than human); and “rhetoricaldehumanization” (i.e., depicting someone as less than human).Kronfeldner’s framework also includes the claim that there arethree different senses of being human: a “group sense”according to which being human means being part of a group (e.g., thespeciesHomo sapiens); a “property sense”according to which being human requires having certain properties(e.g., rationality); and a “moral sense” according towhich being human means having moral standing. According toKronfeldner (forthcoming), once we adopt this pluralist framework, theparadox of dehumanization becomes a “mirage arising frommisplaced abstraction”. This is because in the sorts of casesthat allegedly generate the paradox, dehumanization is likelyoccurring with respect to at least one form, one level, and one senseof being human. For example, on Kronfeldner’s account, even if aperpetrator thinks of a victim as humanin all three senses,so long as they subject them to the relevant sorts of harms, they willcount as dehumanizing them at “the behavioral level”.
In addition to the debate about what it means to dehumanize someone,there are ongoing debates about the causes and consequences ofdehumanization.
In seeking to explain how and why dehumanization occurs, it is usefulto distinguish between “proximal” and“ultimate” explanations. A proximal explanation of somephenomenon aims to identify the mechanisms that tend to produce it inany given moment; whereas an ultimate explanation seeks to identifythe relatively distal factors that explainwhy it exists inthe first place. In evolutionary biology, ultimate explanations invokethe fitness consequences of a given trait (Scott-Phillips, Dickins,& West 2011). However, sometimes the phrase “ultimateexplanation” is used more broadly to refer to the relativelydistal causes of some trait or phenomenon, but without thisnecessarily involving an appeal to fitness consequences (Smith 2021:176). In this section, the term “ultimate explanation”will be used in this broad sense.
In attempting to provide ultimate explanations of dehumanization,proponents of the psychological approach often argue that itfacilitates equanimity on the part of the dehumanizer. The most commonversion of this idea is that dehumanization helps the dehumanizer toovercome their inhibitions against participating in violence (Bandura,Underwood, & Fromson 1975; Kelman 1973, 1976; Opotow 1990;Bandura, Barbaranelli, et al. 1996; Goldhagen 2009; Smith 2011, 2021;Savage 2013; Kronfeldner 2021b: 11–12; Machery 2021). Forinstance, Smith (2021: 206–224) argues that humans have anautomatic inhibition against killing fellow humans and that thinkingof certain people as subhuman is one way for mass killers to disableit. To illustrate, Smith (2021: 222) suggests that Heinrich Himmlerdevised the gas chambers, in part, because they minimized the sort offace-to-face encounters that might cause Nazi camp guards to perceivetheir victims as human, thereby activating the inhibition againstkilling fellow humans.
Other proponents of the psychological approach emphasize that byharboring dehumanizing thoughts, dehumanizers are not just able tooverride inhibitions against committing extreme violence in thefuture: it also enables them to rationalize and cope with acts ofextreme violenceafter they have already committed them(Bandura 1999; Castano & Giner-Sorolla 2006; Luft 2020, 2023; seealso Lelieveld et al. 2024).
Some theorists argue that the functions of dehumanization go beyondfacilitating hostile acts. For instance, one idea is that peopledehumanize outgroup members to cope their own mortality (de Beauvoir1949; Nussbaum 2004, 2013; Heflick & Goldenberg 2013). Anotheridea is that we dehumanize people to minimize the emotional costs ofempathizing with them, even in contexts in which we are helping them(Cameron, Harris, & Payne 2016).
Finally, in providing ultimate explanations of dehumanization, someproponents of the psychological approach have focused on the role ofdehumanizing ideologies (Hagan & Rymond-Richmond 2008; Steizinger2018, 2021; Smith 2021; Leader Maynard & Luft 2023). For instance,Smith (2021: 192–194) explains how certain racializingideologies paved the way for the belief that Jewish people are demonicand subhuman.
Like some of the theorists mentioned above, proponents of theharms-based approach often argue that people dehumanize others tomaintain hostile and oppressive relationships with them. For example,Mikkola remarks that “dehumanization works by utilizing thesubjects’ human agency, so that they are put back to their placeshould they venture to rebel” (2021: 338). Similarly, MacKinnon(1987, 1989, 2006) and Dworkin (1981, 1985, 1987) both argue thatdehumanizing objectification exists, in large part, because of the waythat social hierarchies frame the objectifier’s sexual desires(see also Nussbaum 1995a: 290).
Dehumanization researchers have also defended numerous ideas about themoreproximate causes of dehumanization. Proponents of thepsychological approach have presented evidence that individuals withcertain character traits are more likely to harbor dehumanizingthoughts: examples include narcissism (Locke 2009); nationalism (Viki& Calitri 2008); right-wing authoritarianism (Maoz & McCauley2008; Kteily, Bruneau, et al. 2015); and social dominance orientation(Hodson & Costello 2007). Some proponents of the psychologicalapproach have also presented evidence that transitory mental states,such as disgust (Harris & Fiske 2011; Landry, Ihm, & Schooler2022a) and perceived threat (Viki, Osgood, & Phillips 2013; Maoz& McCauley 2008) tend to trigger dehumanizing thoughts. Relatedly,some have presented evidence that we tend to form dehumanizingthoughts about others when we believe that they harbor dehumanizingthoughts aboutus: this has become known as“metadehumanization” (Kteily, Hodson, & Bruneau 2016;Landry, Ihm, & Schooler 2022b).
Another cluster of ideas concerning the proximate causes ofdehumanization centers on people’s beliefs about humanness. Oneidea is that dehumanization is driven by the belief that to be humanis to possess an underlying human essence (Smith 2011, 2014, 2016,2021; Landry, Ihm, Protzko, & Schooler 2022; although, seeKronfeldner 2021c). Another idea is that dehumanizing thoughts aredriven by the belief that to be human is to possess a morally good“true self” (Phillips 2022a, 2023). Finally, there is alsoevidence that dehumanization is driven by the belief that humans arefundamentally superior to nonhuman animals (Costello & Hodson2010, 2014; see also Smith 2011, 2021).
Relatedly, some researchers have stressed that dehumanizing beliefsare directly influenced by the conception of humanness that prevailsin the given socio-cultural context (Kronfeldner 2017, 2018;Steizinger 2018; Leader Maynard & Luft 2023; see also Smith 2013).For example, Leader Maynard and Luft (2023, 2) argue that theprevailing ideology in Nazi Germany fueled the belief that variousethnic outgroups were “biologically inferior forms of the humanspecies”; whereas in Stalin’s Soviet Union, animalisticdehumanization took on a different guise because Marxist-Leninistideology precluded this “biological hierarchization” ofhumanness.
In discussing the means through which dehumanizing ideologies aretransmitted, some researchers have focused on how propaganda andderogatory language can cause people to have dehumanizing thoughts(Hagan & Rymond-Richmond 2008; Tirrell 2012; Steizinger 2018;Smith 2011, 2021; Leader Maynard & Luft 2023). For instance, usingthe Rwandan genocide as a case study, Tirrell argues that certainderogatory speech acts “train potential participants todehumanize their targets” (2012: 211). Similarly, Smith arguesthat propaganda can “motivate men and women to slaughter oneanother” by “inducing men to regard their enemies assubhuman creatures” (2011: 71).
Proponents of the harms-based account have also theorized about theproximate causes of dehumanization. In examining the proximate causesof objectifying dehumanization, some feminists have focused onpornography (see the entry onfeminist perspectives on objectification). For example, Langton (1995: 178) remarks that “As a matter ofhuman psychology, when men sexually use objects, pornographicartifacts, as women, they tend to use real women as objects”(see also Assiter 1988: 68; Langton 2009: 38–39). However, otherdefenders of the harms-based approach have resisted, or at leastheavily qualified, the claim that pornography is among theprimary causes of objectifying dehumanization (see MacKinnon1993: 109 and 25; Nussbaum 1995a: 286, 290).
Researchers have also defended various ideas about the consequences ofdehumanization. Some proponents of the psychological approach arguethat when someone is dehumanized, this often results in them beingperceived as having little or no moral standing (e.g., see Kelman1973; Machery 2021). Importantly, though, not all proponents of thepsychological approach hold this view. For instance, some defend theconverse view that dehumanization is a consequence of perceivingsomeone as having little or no moral standing (Opotow 1990); whileothers hold that to perceive someone as having little or no moralstandingjust is to dehumanize them (de Ruiter 2023a, 2023b;see also Bandura 2002).
Some proponents of the psychological approach argue thatdehumanization is a catalyst for genocide. For instance, according toStanton’s (1998) influential Genocide Watch model,dehumanization prepares genocidaires to overcome the “normalhuman revulsion against murder” (see§6.1 for more details).
Social psychologists have conducted numerous lab-based studies toexamine the consequences of dehumanization. These studies suggest thatdehumanizing thoughts can lead to various forms of hostility, such assupport for wartime aggression (Jackson & Gaertner 2010; Viki,Osgood, & Phillips 2013; Bruneau & Kteily 2017; Slovic et al.2020; Rousseau, Gorman, & Baranik 2023; Landry, Fincher, et al.2024); support for vengeance against perceived enemies (Kteily,Bruneau, et al. 2015); the endorsement of harsher sentences incriminal cases (Bastian, Denson, & Haslam 2013; Fincher &Tetlock 2016); an unwillingness to support rehabilitation for criminaloffenders (Viki, Fullerton, et al. 2012); proclivity towards rape andsexual harassment (Rudman & Mescher 2012); and a reduced tendencyto help others (Cuddy, Rock, & Norton 2007).
In addressing these sorts of lab-based studies, some theorists haveargued the relation between dehumanization and hostility is morecomplex than social psychologists tend to think. For instance, LeaderMaynard and Luft (2023) argue that the link between dehumanization andviolence is mediated by various factors that go beyond the psychologyof individuals, such as the ideological, social, and institutionalcontext in which the violence occurs. Similarly, Steizinger arguesthat in studying the relation between dehumanization and massviolence, theorists should go beyond the psychology of individuals andconsider “ideological contexts and social practices” aswell (2018: 157).
Other theorists have argued that having dehumanizing thoughts aboutsomeone only increases support forcertain types of violenceagainst them. Most notably, Rai, Valdesolo, and Graham (2017) presentevidence that dehumanization increases support for “instrumentalviolence” but not “moral violence”. Theycharacterize violence as “instrumental” when it is used toachieve some nonviolent goal, such as pain inflicted on animals inmedical research. In contrast, they characterize violence as“moral” when it is driven by moral judgments of thevictims. In a series of studies, Rai and colleagues found thatdehumanizing attitudes increased support for instrumental, but notmoral, violence. They argue that dehumanization does not lead to moralviolence because it “removes the very qualities that make moralviolence meaningful” (2017: 8514), and that moral violence isintentionally targeted towards “complete human beings capable ofdeserving, suffering, and understanding the harm done to them”(2017: 8515). Importantly, though, Rai and colleagues only tested theagency-experience model of dehumanization (Fincher, Kteily &Bruneau 2018; see also§3.2.2). Moreover, other studies, deploying different models and measures,suggest that dehumanization predicts support for both moral andinstrumental violence (e.g., see Bastian, Denson, & Haslam 2013;Kteily, Bruneau, et al. 2015; Slovic et al. 2020; Phillips 2023).
Proponents of the harms-based approach have also theorized about theconsequences of dehumanization. One core idea is that dehumanizingactions perpetuate systems of social hierarchy, oppression, andcontrol (MacKinnon 1987, 1989, 2006; Dworkin 1981, 1985, 1987;LeMoncheck 1985, 1997; Nussbaum 1995a; Cudd 2006; Mikkola 2016). Forexample, Mackinnon (1989: 127–128) and Dworkin (1985 [2000:30–32]) both argue that by shaping men’s sexualpreferences, the dehumanizing objectification of women results insocial inequality and the normalization of a male-dominated socialorder. Similarly, Mikkola argues that dehumanizing actions oftenresult in the victim being “put back to their place should theyventure to rebel and challenge their lot” (2021: 338).
Researchers have invoked dehumanization in attempting to understand arange of societal problems. This section outlines some commonexamples.
Dehumanization is widely seen as a common occurrence during war andgenocide (Kelman 1973; Staub 1989; Chalk & Jonassohn 1990; Fein1990; Goldhagen 1996, 2009; Stanton 1998; Hagan & Rymond-Richmond2008; Smith 2011, 2020, 2021; Savage 2013; Haslam 2020; Haagensen& Croes 2012; Landry, Orr, & Mere 2022).
According to the influential Genocide Watch model, dehumanizingthoughts facilitate genocidal violence by preparing participants toovercome “the normal human revulsion against murder”(Stanton 1998; see also Goldhagen 2009; Smith 2021). Others emphasizethe idea that dehumanization also facilitates genocidal violence as itunfolds. For example, one idea is thatduring periods ofgenocidal violence, the killers need to maintain dehumanizing thoughtsabout their victims to minimize distress and feelings of guilt (Smith2011: 132–162; Haagensen & Croes 2012: 225; Haslam 2020:131). Other theorists argue that participants in genocide often beginby seeing their victims as fully human, but that dehumanizationbecomes increasingly common as they adapt to the experience ofcommitting extreme violence (Luft 2020, 2023).
Evidence from lab-based studies suggests that the more civiliansdehumanize their military rivals, the more likely they are to supportvarious forms of military violence. For instance, some studies suggestthat dehumanizing thoughts predict support for torturing prisoners ofwar (Viki, Osgood, & Phillips 2013); while others suggest thatdehumanizing thoughts predict support for the indiscriminate bombingof both military personnel and civilians (Kteily, Bruneau, et al.2015; see also Jackson & Gaertner 2010; Leidner, Castano, &Ginges 2013; Bruneau & Kteily 2017; Slovic et al. 2020; Rousseau,Gorman, & Baranik 2023; Landry, Fincher, et al. 2024).
Some theorists have questioned whether dehumanization facilitates warand genocide (see also§3.4 on general arguments against the psychological approach). Forinstance, Lang (2010) starts by defining “dehumanization”as the denial of others’ “subjectivity”—inother words, he appears to adopt a version of the mind perceptionthesis (see§3.2.2). He then argues that concentration camp guards in Nazi Germanyrecognized Jewish prisoners as humans with subjective experiences,because one of their primary goals was to exert power over them bysubordinating their subjectivity to their own (for a response, seeSteizinger 2018). Leader Maynard (2022) agrees that dehumanization canfacilitate violence, however, he argues that its role in mass killingis often exaggerated. For example, he argues that during World War II,policymakers in Britain and the US perceived Germans as“contemptible”, but “rarely denied theirhumanity” (2022: 114).
Lab-based studies suggest that the members of racial and ethnicoutgroups are especially vulnerable to dehumanization. Some of thesestudies suggest that dehumanizing beliefs about the members of racialand ethnic minorities predict reduced support for immigration (Hodson& Costello 2007; Esses et al. 2008; Kteily, Bruneau, et al. 2015:913; Bruneau, Kteily, & Laustsen 2018; Utych 2018; Markowitz &Slovic 2020; Landry, Ihm, & Schooler 2022a). Other studies focuson the dehumanization of racial and ethnic minorities in various partsof the criminal justice system (see§6.4 for more details).
Some theorists have examinedwhy racial and ethnic outgroupsare especially vulnerable to dehumanization (Smith 2011, 2021;Costello & Hodson 2014; Hund 2021; Jardina & Piston 2021,2023). For instance, Smith (2011: 163–201; 2021: 83–99)argues that the dehumanization of whole groups of people is“typically preceded and facilitated by theirracialization” (2021: 98; see also Hagan & Rymond-Richmond2008). In developing this thesis, Smith draws on studies suggestingthat people, including children, often have essentialist beliefs aboutrace. People with essentialist beliefs about race tend to think thatall and only members of the same racial group possess an underlying“racial essence”, where this essence has certain keyproperties. For example, there is evidence that racial essentialiststend to think that racial essences are immutable; transmitted viadescent; and inductively rich (Hirschfeld 1996; see also Machery &Faucher 2017; Phillips 2022b). Smith argues that racial essentialismand dehumanization are thus structurally similar, and that racialessentialism often leads to the belief that members of the givenracial group possess asubhuman racial essence (2011:200–201; 2021: 99; see also Landry, Ihm, Protzko, & Schooler2022).
Various theorists have examined how dehumanizing ideologies are usedin attempts to legitimize slavery and colonialism (Patterson 1982;Fredrickson 1971 [1987]; Jahoda 1999; Smith 2011: 112–123; 2021:39–49, 145–146, and 74–81; Machery 2021; Jardina& Piston 2023).
As Fredrickson observes, during the nineteenth century, WhiteAmericans’ ideology concerning the enslavement of African peopleoften gravitated around two spurious ideas. The first idea was thatslaves are akin to children whose underdeveloped cognitive capacitiesmean that they need parental-like guidance by their masters(Fredrickson 1971 [1987: 55–58]; see also Jahoda 1999; Rollo2018). The second idea was that slaves are akin to subhuman beasts whoneed to be tamed by their masters (Fredrickson 1971 [1987:55–58]; see also Jahoda 1999; Smith 2011: 115–119).
Both ideas figure in other defenses of slavery (Stuurman 2021;Sebastiani 2021; Machery 2021; Hund 2021). For instance, Aristotleargued that certain people are “natural slaves” becausethey are not naturally able to engage in rational deliberation.According to Aristotle, it is therefore just and beneficial for thesepeople to receive direction from a master who can rationallydeliberate on their behalf (Politics I.5, 1254b21–23,and I.13, 1260a12–13). Similarly, Sepúlveda (1547) arguesthat the Spaniards’ enslavement of indigenous people wasjustified because they “are as inferior to the Spaniards as arechildren to adults and women to men” (1547 [1960: 526]). He alsoinvokes the idea that indigenous people exhibit“barbarity” and “natural slavery” (1547 [1960:528]).
In discussing how these ideologies about slavery are dehumanizing,Machery (2021) invokes the agency-experience model (see§3.2.2). He argues that enslaved people are seen as lacking in agency wheneverthey are regarded as child-like or beast-like (2021: 152–154).He also argues that enslaved people are often regarded as lackingcapacities for subjective experiences. For instance, he points outthat during the nineteenth century, defenders of slavery often claimedthat slaves are relatively insensitive to pain, and that they have anunderdeveloped capacity to experience moral emotions (2021:154–155; see also Fredrickson 1971 [1987: 57–58]).
Proponents of the harms-based approach have also examined the ways inwhich slavery and colonialism are dehumanizing (see Nussbaum 1995a:264; Mikkola 2016; Kara 2017; Sangiovanni 2017). Nussbaum (1995a: 264)argues that slavery is dehumanizing because it involves using a humanbeing “as a tool of one’s own purposes”. Kara arguesthat slavery is dehumanizing because it produces conditions“that strip the person of his or her dignity and humanity”(2017: 35). For example, Kara argues that coerced labor often involvesrestricting someone’s “freedom of movement, which is acore freedom that makes us human” (2017: 35).
Not all theorists agree that dehumanization is a common feature ofslavery (see Jordan 1968; Cavell 1979 [1999: chapter 13]; Rinehart2016; Johnson 2017). Historian Winthrop Jordan argues that despitereferring to Black slaves as “subhuman”, Americancolonists must have thought of them as fully human because they had“human relationships” with them. For example, Jordan notesthat the colonists frequently talked with their slaves, and they oftenhad (nonconsensual) sex with them (1968 [1969: 234]). This is anexample of the “paradox of dehumanization” (see§4). Smith (2011: 112–123; 2021: 39–49 and 226–255)addresses this instance of the paradox by arguing that slaveownerstended to have contradictory beliefs about slaves, simultaneouslythinking of them as both human and subhuman (see§4.2).
Researchers in several disciplines have argued that dehumanizationoccurs at multiple levels of the criminal justice system. Some studiesaddress dehumanization in the context of policing (Goff, Eberhardt, etal. 2008; Goff, Atiba, et al. 2014; Zlobina & Andujar 2021;Perillo et al. 2023; see also Mekawi, Bresin, & Hunter 2016,2019). Other studies suggest that dehumanization affects beliefs aboutpunishment type and severity. For instance, there is evidence that theanimalistic and mechanistic dehumanization of criminals predictssupport for retributivist, instead of rehabilitative, forms ofpunishment (Bastian, Denson, & Haslam 2013; see also Osofsky etal. 2005; Viki, Fullerton, et al. 2012; Giner-Sorolla, Leidner, &Castano 2012). There is also evidence that people find it easier topunish criminals when they perceptually dehumanize them (Fincher &Tetlock 2016; see also§3.2.4). Finally, there is evidence that when victim impact statements includedehumanizing language, this enhances punitiveness towards defendantsin mock trials (Myers et al. 2004).
Other researchers have focused more on the relation betweendehumanization and imprisonment. Drawing on the mind perception thesis (§3.2.2), Deska et al. (2020) found evidence that people regardsoon-to-be-released prisoners as more humanlike than those with moretime to serve. Some theorists have argued that solitary confinement isdehumanizing in the sense that it inhibits or destroys one’sability to exercise distinctively human capacities, such as theability to have meaningful relationships: this view is very much inline with the harms-based approach to dehumanization (e.g., seeCoppola 2019). In contrast, Guenther (2012) argues that the harms ofsolitary confinement cannot be adequately captured by an appeal to“dehumanization”. Importantly, though, Guenther restrictsher attention to the harms-based approach todehumanization—especially those that invoke what she sees asanthropocentric notions of “human dignity”.
Dehumanization is widely regarded as central to sexual and genderoppression. For example, proponents of the harms-based approach oftenargue that that women are dehumanized whenever they are treated asmere sexual objects, as opposed to human beings (LeMoncheck 1985;MacKinnon 1987, 1989, 2006; Dworkin 1979, 1985, 1987; Nussbaum 1995a;see also de Beauvoir 1949; Gardner & Shute 2000; Cudd 2006). Incontrast, some philosophers reject this emphasis on objectification.For instance, Mikkola (2016, 2021) argues that in central cases ofmisogynistic dehumanization, such as rape, women are in fact treatedas subjects, as opposed to objects: nonetheless, they are dehumanizedbecause their “self-set goals, aims, and life plans” areintentionally hampered (2021: 337).
There are also debates about how dehumanization figures in specificsorts of gender oppression. For instance, some feminist philosophersargue that pornography is wrong, in part, because it treats anddepicts women as less than fully human (Dworkin 1981, 1985; Longino1980: 45; MacKinnon 1987; Nussbaum 1995a; LeMoncheck 1997; Neufeld2020; see also Mikkola 2019). In response, some philosophers arguethat pornography does not necessarily involve treating or depictingpornographic actors as objects or as less than human (Saul 2006);while others have questioned the assumption that it is wrong to see ortreat pornographic actors in this way (Soble 2002: 53–54).
Social psychologists have also studied the relation betweendehumanization and different forms of sexual and gender oppression.Much of this work focuses on the psychology of sexual objectification(e.g., see K. Gray et al. 2011; Rudman & Mescher 2012; Heflick& Goldenberg 2014; Roberts et al. 2018; Bevens & Loughnan2019). However, some of these studies focus on dehumanizing attitudestowards people with certain sexual orientations (MacInnis & Hodson2012; Fasoli et al. 2016); while others focus on dehumanizingattitudes towards people who are not cisgendered (Velez et al. 2016;Anzani et al. 2021; Cascalheira & Choi 2023).
In examining the dehumanization of people with disabilities, someproponents of the psychological approach have focused on eugenics. Indiscussing the Nazis’ T4 euthanasia campaign against disabledchildren, Smith (2021: 250–251) draws on his distinction between“demonizing dehumanization” and “enfeeblingdehumanization” (see§4.2). The former occurs when someone is regarded as a threatening subhuman;while the latter occurs when someone is thought of as a docile andnon-threatening subhuman. Smith suggests that the Nazis tended todehumanize people with disabilities in the enfeebling way (for somebroader discussions of dehumanization in the context of eugenics, seeO’Brien 1999; Garland-Thompson 2012; Keith & Keith 2013;Kittay 2016; Wilson 2021; Carlson 2023).
Carlson (2023) distinguishes between cases in which people withintellectual disabilities are “animalized”, and cases inwhich they are seen as “vegetables”. In cases ofanimalization, the victims are “defined and treated as nonhumananimals” (2023: 7); whereas in cases of the latter, the victimsare regarded as “evenlower than animals” (2023:7; see also Carlson 2010: 31–33).
Some disability scholars have argued that moral philosophers oftenexpress dehumanizing attitudes towards people with intellectualdisabilities (Kittay 2005; Carlson 2010, 2023; Carlson & Kittay2009; Keith & Keith 2013; Crary 2018, 2021). One core issue inmoral philosophy concerns what it takes to possess “moralstatus” (see the entry onthe grounds of moral status). Moral philosophers sometimes argue that certain cognitive capacitiesare necessary for having moral status, and that species membership isirrelevant. For instance, Singer (1975, 2010) argues that whether acreature has moral status is not determined by its species per se, butby the interests it happens to possess: in particular, the interest innot suffering. In defending this approach, Singer has argued thathumans with severe cognitive disabilities can have less moral standingthan cognitively sophisticated nonhuman animals (see also McMahan2005, 2010). Some disability scholars argue that these sorts ofcomparisons are problematic because they “animalize”people with intellectual disabilities (Kittay 2005; Carlson 2010,2023). Relatedly, Crary suggests that because Singer’scapacity-based approach denies that being human is morally significantper se, it must forfeit “the pragmatically valuable thought thatanimalizing ideologies are wrong because they dehumanize” (2021:166; see also Carlson 2023: 17).
Social psychologists have presented evidence that people tend todehumanize those with disabilities. Some of these studies focus ondehumanizing attitudes towards those with mental disabilities(Martinez et al. 2011; Betancor Rodríguez et al. 2016; Cage etal. 2019; Boysen et al. 2020; Parker et al. 2020; Cooper & Harwood2024; Kim, Cheon, & Kim 2025); while others focus more on thedehumanization of those with physical disabilities (Kunst, Kteily& Thomsen 2019; Sim & Hugenberg 2023; Sitruk et al. 2023).
Finally, some proponents of the harms-based approach have argued thatpeople with disabilities are dehumanized in distinctive ways. Forinstance, Gambrill (2014) argues that in using theDiagnostic andStatistical Manual of Mental Disorders (DSM), psychiatristsdehumanize patients by reducing “opportunities for freedom,growth, and dignity” (for a similar view, see Szasz 1970).
Researchers often raise concerns about the dehumanization of patientsin medical settings (for some overviews, see Haque & Waytz 2012;Capozza et al. 2016; Hoogendoorn & Rodríguez 2023). Inexamining the reasonswhy patients in medical settings arevulnerable to dehumanization, Haque and Waytz (2012) distinguishbetween “functional” and “nonfunctional”causes of dehumanization (see also Capozza et al. 2016). Theycharacterize “functional causes” as those that arenecessary for providing “effective and efficient healthcare” (2012: 182). For example, they argue that treating andseeing patients as mechanical systems with interacting parts, ratherthan humans with capacities for pain and negative emotionalexperiences, “frees up cognitive resources for clinical problemsolving” (2012: 179) and reduces rates of burnout amonghealthcare providers (2012: 182; see also Vaes & Muratore 2013;Trifiletti et al. 2014). Haque and Waytz characterize“nonfunctional causes” of dehumanization as those that areunnecessary for providing effective medical care, but that result from“other social practices that just happen to producedehumanization” (2012: 183). For example, they argue thatpatients often suffer from reduced agency because of their illnesses,which in turn, causes caregivers to animalistically dehumanize them(2012: 178).
The distinction between functional and nonfunctional causes ofdehumanization is directly connected to a debate about whetherdehumanization is sometimes beneficial in medical contexts. Sometheorists argue that dehumanizing patients is a dysfunctional way forhealthcare providers to deal with stress and burnout, and that itharms the relationship between patient and caregiver (Christoff 2014;Capozza et al. 2016; see also Hoogendoorn & Rodríguez2023). Other theorists argue that in certain contexts, the benefits ofdehumanizing patients outweigh the costs. For example, Palmer andSchwan (2021) argue that replacing human caregivers with AI caregiverswill likely result in beneficial forms of dehumanization (for anopposing view, see Sharkey & Sharkey 2012). This is connected to amore general debate about whether dehumanization is sometimes morallypermissible, and even beneficial, or whether it is always morallywrong (Nussbaum 1995a; Papadaki 2010, 2015; Smith 2021: 23).
Dehumanization and objectification are usually seen as closelyrelated. This section focuses on whether they are equivalent, overlap,or are completely distinct. The range of possibilities is depicted inFigure 1.
(a)
(b)
(c)
(d)
(e)
Figure 1. Venn diagrams representingdifferent views on the relation between dehumanization andobjectification.
One possibility, depicted inFigure 1(a), is that dehumanization and objectification are equivalent. On thisview, there are no members of the categorydehumanizationthat are not also members of the categoryobjectification,and vice versa.
As was explained in§2.1, some proponents of the harms-based approach seem to equatedehumanization with objectification (for further discussion, seeMikkola 2021). Dworkin possibly does so when she claims thatobjectification occurs when a human being “is made less thanhuman” by being “turned into a thing or commodity”(1985 [2000: 30–31]), however, she does not explicitly rule outalternative, non-objectifying, ways of making someone less than human.Cudd (2006: 165–166) uses the phrases“objectification” and “dehumanizingobjectification” interchangeably to refer those cases in whichsomeone’s “full and equal status” as a person isignored or damaged because their ability to express distinctivelyhuman capacities has been hampered.
The second possibility, depicted inFigure 1(b), is that objectification is a type of dehumanization. Some proponentsof the psychological approach defend a version of this view. Forinstance, as was explained in§3.2.1, proponents of the dual model argue that there are two forms ofdehumanization: animalistic and mechanistic (Haslam et al. 2005). Theformer involves thinking of someone as relatively animal-like, whereasthe latter involves thinking of someone as relatively object-like orrobotic. Thus, on this view, objectification is one type ofdehumanization, alongside animalistic dehumanization (see also Fiske2021).
The third possibility, depicted inFigure 1(c), is that dehumanization is a type of objectification. Nussbaum (1995a)defends a version of this view (see also Vaes, Loughnan, & Puvia2014). Nussbaum draws a distinction between permissible andimpermissible cases of objectification. Permissible cases occur whensomeone is treated as an object, but their humanity is still respected(1995a: 276). For example, she argues that objectification during sexis often permissible and “joyous” precisely because eachperson is recognized and treated as a human being (1995a: 275; seealso 2007: 51). Impermissible cases of objectification occur whensomeone is treated as an object, but their humanity is not eventacitly acknowledged. Thus, for Nussbaum, there are two kinds ofobjectification: the impermissible kind that involves dehumanization,and the permissible kind that does not. This suggests a view accordingto which dehumanization is a type of objectification: namely, theimpermissible type.
The fourth possibility, depicted inFigure 1(d), is that dehumanization and objectification are partially overlappingcategories. On this view, we should recognize that
It is not clear whether anyone explicitly defends this view in theliterature. However, elements of it have been defended by varioustheorists. For instance, a proponent of the psychological approachcould defend (1), (2), and (3) in the following way. They could defend(1) by arguing that when someone is seen in a purely instrumental way,they are both objectifiedand dehumanized. Following Haslamand colleagues (2005), they could defend(2) by arguing that when someone is seen as animal-like, but notobject-like, they are dehumanized, but not objectified. Finally, theycould defend(3) by arguing that when a nonhuman animal is perceived in a purelyinstrumental way, the animal is objectified, but not dehumanized (fordiscussion of the objectification of nonhuman animals, see Adams 1990;Loughnan, Haslam, & Bastian 2010; Bastian et al. 2012).
How might a proponent of the harms-based approach defend(1),(2), and(3)? It is clear how they could defend (1) and (3). In defending(1)—the thesis that some things count as both dehumanization andobjectification—they could argue that when someone is treated asa mere object, they are both objectified and dehumanized. In defending(3)—the thesis that some things only count asobjectification—they could argue that when a nonhuman animal istreated as a mere object, the animal is objectified, but notdehumanized. It is less clear how a proponent of the harms-basedapproach could defend (2)—the thesis that some things only countas dehumanization—because they tend to regard dehumanization asentailing objectification (see§7.1). One exception is Mikkola who remarks that “dehumanization doesnot (and need not) involve objectification in the sense of viewing ortreating women as literally object-like or as if they wereobjects” (2021: 337). Using cases of misogynistic dehumanizationto illustrate this point, Mikkola argues that dehumanized women are“treated as subjects all right, just not self-determiningones” and that “dehumanization turns on precisely aimingto hamper and frustrate women’s self-set goals, aims, and lifeplans” (2021: 337).
Finally, dehumanization and objectification might be completelydistinct categories, as depicted inFigure 1(e). On this view, the two categories do not have any members in common,and to think otherwise would be to commit a category mistake (see theentry oncategory mistakes).
The most obvious way to defend this view is to argue thatdehumanization is a way ofthinking about others, whereasobjectification is a way oftreating others (Smith 2021:19–23). On this view, even though there are contexts in whichpeople think of others as less than humanand treat them asmere objects, it does not follow that dehumanization andobjectification are overlapping categories, because one is a type ofpsychological attitude and the other is a way of treating people.
Importantly, the view that dehumanization and objectification arecompletely distinct categories does not entail that they are notcausally related. To illustrate, something cannot be both anelectron and a proton: clearly, though, electrons and protons arecausally related. Discussions of the causal relation betweendehumanization and objectification are relatively rare, however, sometheorists suggest that dehumanization can cause negative forms ofobjectification (Gervais et al. 2013: 12–13); while others havesuggested that certain forms of objectification can causedehumanization (Mikkola 2016: 176; 2021: 338).
Dehumanization is a growing topic of interest within academia andpopular culture. However, various challenges remain.
One challenge concerns the scope of dehumanization. If, for example,the term “dehumanization” is used too liberally to referto any negative thought or feeling about someone else, then it losesits explanatory power: to borrow Paul Bloom’s phrase, “Ifeverything is dehumanization, then nothing is” (2022). Inaddressing this concern, it will be important for proponents of thepsychological approach to further examine how dehumanization isdistinct from other negative attitudes, such as dislike and contempt.For proponents of the harms-based approach, the analogous challenge isto delineate the border between the sorts of harms that qualify asdehumanizing, and the sorts that do not.
A related challenge concerns the proliferation of models and measuresof dehumanization within social psychology (see Landry & Seli2024). To what extent are these different models and measures gettingat different aspects of the same underlying phenomenon? It is possiblethat people possess multiple concepts of being human, each oneengendering a completely distinct form of dehumanization (see§3.1.2 on ambiguity theories). Alternatively, it is possible that peoplepossess a single, multidimensional, concept of being human, and thatdifferent versions of the psychological approach tap into differentdimensions of this concept (see§4.4). Relatedly, it will also be important for researchers to examinewhether the psychological approach to dehumanization is compatiblewith the harms-based approach, or whether they should be consideredcompeting theories (for further discussion, see Smith 2021: chapter 1;Kronfeldner forthcoming).
Finally, another major challenge is addressing the broad contexts inwhich dehumanization occurs. As some theorists have noted, socialpsychologists often neglect the fact that dehumanizing attitudes,along with their consequences, vary across the ideological, social,and institutional contexts in which they occur (Leader Maynard &Luft 2023; see also Steizinger 2018 and Smith 2023). Thus, to fullyunderstand how and why dehumanization matters, researchers must movebeyond the level of the individual and develop models that addressthese higher levels of analysis.
How to cite this entry. Preview the PDF version of this entry at theFriends of the SEP Society. Look up topics and thinkers related to this entry at the Internet Philosophy Ontology Project (InPhO). Enhanced bibliography for this entryatPhilPapers, with links to its database.
[Please contact the author with suggestions.]
bias, implicit |category mistakes |cognitive disability and moral status |colonialism |dignity |disability: and justice |eugenics |feminist philosophy, topics: perspectives on disability |feminist philosophy, topics: perspectives on objectification |feminist philosophy, topics: perspectives on rape |feminist philosophy, topics: perspectives on sex markets |hate speech |human enhancement |human nature |moral status, grounds of |pornography: and censorship |rights: human |social ontology |war
View this site from another server:
The Stanford Encyclopedia of Philosophy iscopyright © 2025 byThe Metaphysics Research Lab, Department of Philosophy, Stanford University
Library of Congress Catalog Data: ISSN 1095-5054