Why do people become violent extremists?
Not because of one thing. The honest answer from two decades of research is that the road into political violence has several distinct on-ramps — and that telling them apart is what makes prevention possible.
Keep scrolling
There is no single extremist profile
The first thing the evidence takes away is the idea of a single 'violent type'. In a preregistered study my colleagues and I ran across 58 countries, support for striking first and support for striking back turned out to rest on different psychological foundations — offensive and defensive violence are not one propensity expressed twice. The same lesson comes out of work measuring extremist archetypes: people drawn toward violence sort into recognisably different types, not points along one continuum. Any policy built on the assumption of a single profile starts on the wrong foot.
Grievance needs a mechanism
Perceived threat to one's culture is among the most reliable predictors of support for violence that we have found — but it does not work directly. Cultural threat predicts violent extremism via the need for cognitive closure: what does much of the work is the discomfort of uncertainty, the craving for a definite answer, rather than the ideology itself. Alongside threat sits relative deprivation — the sense that your group is being left behind, and narratives like the 'great replacement', which convert a perceived loss of group status into extreme hostility toward immigrants.
When the group becomes the self
Some people do not merely belong to a group; they fuse with it, so that an attack on the group feels like an attack on their own body. Our meta-analysis of identity fusion shows how strongly that visceral oneness predicts extreme action on the group's behalf. It can also attach to a person: fusion with a political leader predicts willingness to persecute immigrants and political opponents — a finding my colleagues and I published before it became topical, and one that has not become less so.
What artificial intelligence changes
The pathway I have described is increasingly walked online, and recent work sets out a framework for radicalization in the age of AI: recommender systems and conversational agents do not just increase the volume of extremist content, they reshape the route into it — faster, more personalised, and harder for outsiders to observe. How that same machinery works from the supply side is the subject of the misinformation strand of my research.
What actually helps
If the entry into extremism is social, the exit is too. Our work on resilience against radicalization in schools treats resistance as something that can be measured and built, not merely the absence of risk. And because counter-terrorism policy needs public consent, it matters that people's lay beliefs about what causes terrorism shape which policies they will accept. Prevention that misreads the psychology loses twice: it targets the wrong mechanism, and it forfeits the public's trust while doing so.