No convincing evidence outgroups are denied uniquely human characteristics: Distinguishing intergroup preference from trait-based dehumanization

No convincing evidence outgroups are denied uniquely human characteristics: Distinguishing intergroup preference from trait-based dehumanization. Florence E. Enock et al. Cognition, Volume 212, July 2021, 104682, https://doi.org/10.1016/j.cognition.2021.104682

Highlights

• The dual model predicts outgroups are attributed human traits to a lesser extent.

• To date, predominantly desirable traits have been investigated, creating a confound.

• We test attributions of desirable and undesirable human traits to social groups.

• Attributions of undesirable human traits were stronger for outgroups than ingroups.

• We find no support for the predictions of the dual model of dehumanization.

Abstract: According to the dual model, outgroup members can be dehumanized by being thought to possess uniquely and characteristically human traits to a lesser extent than ingroup members. However, previous research on this topic has tended to investigate the attribution of human traits that are socially desirable in nature such as warmth, civility and rationality. As a result, it has not yet been possible to determine whether this form of dehumanization is distinct from intergroup preference and stereotyping. We first establish that participants associate undesirable (e.g., corrupt, jealous) as well as desirable (e.g., open-minded, generous) traits with humans. We then go on to show that participants tend to attribute desirable human traits more strongly to ingroup members but undesirable human traits more strongly to outgroup members. This pattern holds across three different intergroup contexts for which dehumanization effects have previously been reported: political opponents, immigrants and criminals. Taken together, these studies cast doubt on the claim that a trait-based account of representing others as ‘less human’ holds value in the study of intergroup bias.

Keywords: DehumanizationIntergroup biasPrejudiceSocial cognition

7. General discussion

In this paper, we question the central claims of one of the most prominent psychological accounts of dehumanization - the dual model - which holds that outgroup members are perceived as lesser humans than ingroup members by being denied human specific traits (Haslam, 2006). We first revisited work relating to how the lay concept of ‘human’ is best characterised. We then tested its predictions about outgroup dehumanization in a series of seven experiments. Our results present a serious empirical challenge to the dual model.

The dual model argues that there are two sense of humanness: human uniqueness and human nature. Uniquely human traits can be summarised as civility, refinement, moral sensibility, rationality, and maturity. Human nature traits can be summarised as emotional responsiveness, interpersonal warmth, cognitive openness, agency, and depth (Haslam, 2006). However, the traits that supposedly characterise ‘humanness’ within this model are broadly socially desirable (Over, 2020a; Over, 2020b). We showed that people also associate some undesirable traits with the concept ‘human’. As well as considering humans to be refined and cultured, people also consider humans to be corrupt, selfish and cruel.

Results from our pretest provided us with grounds for re-examining predictions made by the dual model of dehumanization about the nature of intergroup bias in trait attributions. The dual model account holds that lesser attribution of human specific traits to outgroup members represents a psychological process of dehumanization that is separable from ingroup preference. However, as the human specific attributes summarised by the model are positive and socially desirable, it is possible that previous findings are better explained in terms of ingroup preference, the process of attributing positive qualities to ingroup members to a greater extent than to outgroup members.

In seven highly-powered experiments, we tested the predictions of the dual model against this alternative. We pitted the two hypotheses against each other by comparing attributions of uniquely human traits that varied in whether they were socially desirable or undesirable to ingroup and outgroup members. The dual model holds that subtle dehumanization is evidenced by denying outgroup members uniquely human traits relative to ingroup members. We reasoned that whereas outgroup members may be denied desirable human traits, they are likely to be attributed undesirable human traits to a greater extent than ingroup members.

Across three distinct intergroup contexts, we found no evidence for either animalistic or mechanistic dehumanization of outgroup members. Instead, we found strong and reliable intergroup preference effects. Desirable traits were ascribed more strongly to ingroup members than outgroup members and undesirable traits more strongly to outgroup members than ingroup members, irrespective of perceived humanness.

A possible defence of the dual model account could be to argue that we chose three intergroup contexts in which animalistic and mechanistic dehumanization does not occur. However, we chose to investigate judgements of political opponents, immigrants and criminals specifically because previous research has suggested that they are dehumanized on a range of measures (Banton et al., 2020; Bastian et al., 2013; Markowitz & Slovic, 2020; Pacilli et al., 2016; Viki et al., 2013). In addition, we also showed in every experiment that outgroup members were explicitly rated as less human than were the ingroup on the blatant dehumanization scale (Kteily et al., 2015). Prior work shows that measures of animalistic and mechanistic dehumanization correlate positively with blatant dehumanization scores (Kteily et al., 2015). Though they are not claimed to measure the same construct, they have been shown to reliably co-occur. These findings confirm that these are the sorts of intergroup contexts in which we would expect to see trait-based dehumanization should the process occur.

We acknowledge that without testing all possible intergroup contexts, it remains a possibility that some outgroups could be denied human specific attributes relative to ingroups even when valence is appropriately controlled for. In other words, it could be the case that trait-based dehumanization occurs independently of ingroup preference in some social settings. It may be particularly interesting for future research to investigate intergroup contexts that are not so strongly associated with competition, threat and animosity.

However, the possibility that some, as yet untested, groups may be denied human unique attributes does not detract from the importance of our critique. To accurately measure trait-based dehumanization in future research, studies must consider the central role of valence. Prior work utilising the dual model framework has reported dehumanization to be extremely widespread in society, affecting not just marginalised groups but doctors, patients and even cyclists (Delbosc, Naznin, Haslam, & Haworth, 2019; Haslam & Stratemeyer, 2016). Rigorous measurement and tighter experimental control may change some or all of the conclusions from previous research.

Across our experiments, we observed strong intergroup preference effects, with desirable traits more strongly ascribed to the ingroup and undesirable traits more strongly to the outgroup. Our results demonstrate both ingroup favouritism (assigning greater positivity to the ingroup) and outgroup derogation (assigning greater negativity to the outgroup) (Brewer, 1999; Hewstone et al., 2002). However, we also suggest that group specific stereotypes are likely to play an important role in these processes. In many social contexts, trait attributions may reflect social stereotyping as well as intergroup preferences (Fiske et al., 2002). For example, previous work suggested that Anglo-Australians were ‘animalistically’ dehumanized both by themselves and by Ethnic-Chinese participants, whilst Ethnic-Chinese people were ‘mechanistically’ dehumanized both by themselves and by Anglo-Australians (Bain et al., 2009). These effects may be more compatible with stereotype content than with trait-based dehumanization. Future work would benefit from addressing the distinction between stereotyping and trait-based dehumanization.

An outstanding question relates to whether other psychological models of dehumanization more accurately capture the ways in which different social groups are perceived. For example, infrahumanisation theory predicts that people tend to believe ingroup members experience uniquely human emotions more strongly than do outgroup members (Leyens et al., 2000, Leyens et al., 2001). It would be valuable for future research to examine the utility of this theory by testing whether participants perceive ingroup members to experience human emotions more strongly overall or whether they perceive ingroup members to experience prosocial emotions more strongly but outgroup members to experience antisocial emotions more strongly. Further work could helpfully investigate how these findings bare on the claim that outgroups are sometimes dehumanized by being denied mental states (Harris & Fiske, 2006).

Taken together, our studies suggest that the dual model does not accurately characterise the ways in which outgroups are perceived in at least the social contexts examined here – political groups, immigrants and criminals. Prejudice and discrimination are pressing social problems. If psychological research is to contribute to the interdisciplinary mission to reduce prejudice and encourage more egalitarian behaviour, then it must start by accurately characterising the psychological biases underlying discriminatory behaviour. We suggest that the dual model of dehumanization conflates apparent evidence for dehumanization with ingroup preference. As a result, it may obscure more than it reveals about the psychology of intergroup bias.