Perceptual dehumanization theory: A critique.

Abstract

Central to perceptual dehumanization theory (PDT) is the claim that full engagement of a putative module for the visual analysis of faces is necessary in order to recognize the humanity or personhood of observed individuals. According to this view, the faces of outgroup members do not engage domain-specific face processing fully or typically and are instead processed in a manner akin to how the brain processes objects. Consequently, outgroup members are attributed less humanity than ingroup members. To the extent that groups are perceptually dehumanized, they are hypothesized to be vulnerable to harm. In our article, we challenge several of the fundamental assumptions underlying this theory and question the empirical evidence in its favor. We begin by illustrating the extent to which the existence of domain-specific face processing is contested within the vision science literature. Next, we interrogate empirical evidence that appears to support PDT and suggest that alternative explanations for prominent findings in the field are more likely. In the closing sections of the article, we reflect on the broader logic of the theory and highlight some underlying inconsistencies.

Are upside-down faces perceived as “less human”?

“…A significant Orientation × Face Type × Attribute interaction was found, F (4, 516) = 25.01, p < .001, η 2 p = .16. For the traits creative, thoughtful, considerate, and empathetic, the pattern was similar and followed the predictions of Over and Cook (2022). Trustworthy faces were seen to possess these attributes to a lesser extent when inverted (all p s < .001), while untrustworthy faces were seen to possess these attributes to a greater extent when inverted (all p s < .003).…”

“…According to the alternative view, inversion does not lead to impaired attribution of uniquely human traits. Rather, inverted faces appear less distinctive than upright faces because observers are less able to detect and encode distinguishing features (Over & Cook, 2022). Consequently, all trait ratings will tend toward the average when face stimuli are shown inverted.…”

“…Over and Cook (2022) predict that faces that appear trustworthy when viewed upright will be rated as less empathetic, thoughtful, creative, and considerate when viewed inverted. Conversely, faces that appear untrustworthy when viewed upright will be rated more empathetic, thoughtful, creative, and considerate when viewed inverted (Over & Cook, 2022).…”

“…Recently, however, an alternative interpretation of these results has been presented as part of a broader critique of PDT (Over & Cook, 2022). Over and Cook (2022) argue that what appears to be evidence for perceptual dehumanization in this paradigm may be better explained by the observation that orientation inversion impedes observers’ ability to derive accurate perceptual descriptions of faces (McKone & Yovel, 2009; Rossion, 2008; Yin, 1969). When observing inverted faces, individuals are less able to detect and encode stimulus variation: expressions seem more neutral, and face shapes appear less distinct (i.e., closer to the population average).…”

“…Consistent with this view, inversion has been shown to adversely affect a wide range of perceptual decisions about faces that are unrelated to theoretical models of dehumanization including judgments about identity (Yin, 1969), age (Murphy & Cook, 2017), gender (Murphy et al, 2020), similarity (Biotti et al, 2019), expression (McKelvie, 1995; Prkachin, 2003), attractiveness (Bäuml, 1994; Cook & Duchaine, 2011), and adiposity (Thompson & Wilson, 2012). The fact that inverted facial percepts are impoverished might plausibly explain why individuals who appear thoughtful, considerate, creative, and empathetic when their faces are viewed upright appear to exhibit these attributes to a lesser extent when their faces are viewed upside-down (Over & Cook, 2022).…”