They are not described as wrong. They are described as something less than people.
Literal meaning: Dehumanization describes the psychological and rhetorical process of denying the full humanity of a person or group — portraying them as animals, vermin, machines, or morally contaminated — in ways that reduce the psychological barriers to harming them. It is both a cognitive process and a language practice.
Origin: The concept was theorised by Albert Bandura as part of his broader framework of moral disengagement — the psychological mechanisms by which ordinarily moral people come to commit harmful acts. Bandura identified dehumanisation as a key disengagement mechanism: if the victim is not fully human, the inhibitions against harming them do not apply. Genocide scholars including Gregory Stanton, James Waller, and Ervin Staub subsequently documented dehumanising language as a consistent precursor stage in historical atrocities.
The systematic denial of full humanity to a person or group — a documented precursor to persecution and violence.
The Appeal: Dehumanisation serves a psychological function for those who engage in it: it resolves the cognitive dissonance between moral self-image and harmful intent. If the target is not fully human, harming them is not a moral violation. The rhetorical forms vary — animal metaphors, disease metaphors, criminality metaphors — but the function is consistent.
The Friction: The research is unusually robust. Bandura’s moral disengagement framework, developed across decades of empirical work, identifies dehumanisation as a reliable mechanism. Waller’s interviews with genocide perpetrators document how dehumanising language preceded and enabled violence. This makes digital platforms’ treatment of dehumanising language as protected expression analytically significant: the platforms are not merely hosting offensive speech, they are providing infrastructure for a documented harm precursor. The Incel forums show the mechanism in miniature: “non-person” is the standard word for women there, used 114 times in a single thread of 453 posts (De Vettor et al. (2025)). That infrastructure can also be profitable. Ragebaiting networks that manufacture outrage for advertising revenue routinely surface comments deploying the classic vermin frame. In one network documented by Justice for Prosperity (2026), the comment “ongedierte mag je bestrijden” (“vermin may be exterminated”) drew 133 likes and no moderation, even though it breaches Meta’s own rule against comparing groups to animals. Here dehumanisation is not the actors’ conviction but a byproduct of engagement sold for money. Subhuman is the endpoint of the language. Groomer, Social Contagion, and Woke Mind Virus are less extreme instances of the same mechanism — applying pathology or predator frames that reduce target group’s perceived humanity. Deplatforming is the primary response; its effectiveness at preventing downstream harm is contested.
Why This Matters: Dehumanization names the mechanism, not just the outcome. Once you know the sequence — dehumanising language, reduced inhibitions, escalated harm — the presence of dehumanising rhetoric in mainstream discourse is not merely offensive. It is a stage in a process.
Related terms: Subhuman · Groomer · Social Contagion · Great Replacement · Ragebaiting · Deplatforming · Woke Mind Virus · Incel
Read more:
- Selective Moral Disengagement in the Exercise of Moral Agency — Bandura, A. (2010). Journal of Moral Education, 31_(2), 101–119.
- Becoming Evil — Waller, J. (2002). Oxford University Press
- Polarisatie als verdienmodel — Justice for Prosperity (2026). JfP — on outrage manufactured for ad revenue, including unmoderated vermin-frame comments