Extremism, Radicalisation and Lone-Actor Violence

Clark McCauley and Sophia Moskalenko's two-pyramid model, developed in Friction: How Radicalization Happens to Them and Us (2011), is the dominant conceptual framework in the current radicalisation literature.

The opinion pyramid traces the distribution of extreme views across a population. At the base, a large proportion holds some sympathy for extreme grievances. At the middle, a smaller proportion endorses political violence in principle. At the apex, a very small fraction believes violence is the correct action now. Movement up this pyramid is shaped by social identity processes, group polarisation, and narrative exposure.

The action pyramid runs parallel: at the base, legal activism and political expression; in the middle, confrontational and illegal support acts; at the apex, mass-casualty violence. The two pyramids are partially independent. Someone can reach the top of the opinion pyramid without ever approaching violence. Someone can commit violence without being near the top of the opinion pyramid.

This independence has a direct policy implication: counter-terrorism strategies focused exclusively on monitoring or changing extreme views may intrude on large numbers of people who will never act while missing individuals in less-monitored communities who are moving up the action pyramid. For forensic risk assessment, the action pyramid, operationalised in TRAP-18 warning behaviours, is the more relevant domain than ideological content.

Arie Kruglanski's complementary 3N model (Needs, Narrative, Network) proposes that radicalisation requires a personal significance quest (need), an ideology framing violence as significance-restoration (narrative), and a social network validating that framing (network).

TRAP-18 was developed by Reid Meloy, Jill Morin Roshdi, Steffen Guldimann, and Bram Hoffmann (published 2014 in Journal of Threat Assessment and Management). It distinguishes eight proximal warning behaviours, dynamic, observable, changing in the period immediately preceding an attack, from ten distal characteristics, background factors that amplify risk when proximal behaviours are escalating.

The eight proximal warning behaviours:

1. Pathway, research, planning, preparation, or probing of a target; moving from ideology to execution.
2. Fixation, increasing exclusive focus on a cause or person, crowding out other life domains.
3. Identification, desire to be a warrior, soldier, or martyr; adoption of ideological identity as primary self-concept.
4. Novel aggression, unrelated violence serving as a test or expressing emerging propensity.
5. Energy burst, dramatic activity increase in attack-related domains (materials, communication, final arrangements).
6. Leakage, intentional or unintentional communication of violent intent to a third party.
7. Last resort, evidence of a final-push mentality; the actor has internally crossed a threshold.
8. Direct threat, explicit communication of intent to use violence against a specific target.

Selected distal characteristics include: personal grievance and moral outrage, framing ideology, failure to affiliate with an extremist group (lone-actor signal), virtual-community dependence, thwarted occupational goals, relevant mental disorder, and access to weapons.

Meloy and colleagues validated TRAP-18 retrospectively against 73 lone actors across ideological categories, jihadist, far-right, far-left, incel, showing that leakage, pathway, and fixation were present in the majority of cases. Independent validation in UK MAPPA caseloads (James and colleagues) and EU samples (Pressman and colleagues in the Netherlands) supports face validity and case-structuring utility, while noting that retrospective case-file validation does not directly establish prospective sensitivity and specificity. Reference instruments for general violence risk (HCR-20 V3) and sex-offender risk (Static-99R) live in Module 4; TRAP-18 is the specific lone-actor protocol.

The VERA (Violent Extremism Risk Assessment) was developed by Elaine Pressman and John Flockton (2010, revised to VERA-2R). It is a structured professional judgement (SPJ) instrument for known extremists already in contact with the justice or security system, not for initial threat-identification screening. Five assessment domains:

1. Beliefs and attitudes, violent ideological commitment, dehumanising outgroup narratives, perceived obligation to act.
2. Context and intent, proximity to operational milieu, expressed intent, preparation behaviours.
3. History and action, prior violent extremist action, criminal history, network affiliations.
4. Commitment and motivation, dedication to cause, personal significance invested in ideology, in-group status from commitment.
5. Protective factors, family deterrents, employment stability, cognitive flexibility, institutional attachments.

VERA-2R is used by security services and courts in the Netherlands, Belgium, Germany, and the UK, and by Australia's ASIO Behaviour Analytics Unit. The EU Internal Security Fund has funded research on its use in deradicalisation programmes.

A critical methodological constraint applies to both TRAP-18 and VERA-2R: the base rate of lone-actor attacks is very low even in the most affected countries. Even a highly sensitive instrument will produce many false positives for every true positive in a low-prevalence environment. Pressman and colleagues acknowledge this in their validation literature: the instruments' primary value is structuring case management and identifying modifiable risk factors, not predicting rare events from a population screen.

Pittsburgh Tree of Life Synagogue, October 2018 (US). Robert Bowers killed 11 people during Saturday services. His Gab platform history documented an escalating antisemitic narrative over the weeks before the attack, with explicit linking of Jewish-led immigration organisations to what he framed as white genocide. Retrospective TRAP-18 analysis identifies multiple proximal warning behaviours: leakage (explicit online statements of intent), pathway (weapons acquisition and location reconnaissance), and fixation (increasingly exclusive ideological focus). A not-guilty-by-reason-of-insanity plea was entered and rejected; defence neuropsychological evidence did not satisfy the MPC substantial-capacity standard.

Christchurch, March 2019 (New Zealand). Brenton Tarrant killed 51 people in two mosque attacks, published a 74-page manifesto beforehand, and livestreamed the attack. He had not come to the attention of New Zealand or Australian security services despite international travel and legal weapons acquisition. The Royal Commission (He Whenua Tōtū Nei, 2020) identified systemic failures to monitor online extremist activity on Facebook and 8chan. He pleaded guilty, receiving life without parole, New Zealand's first such sentence. Retrospective TRAP-18: pathway, fixation, identification, energy burst, and leakage all documented.

Pulwama, February 2019 (India). A car bomb killed 40 Central Reserve Police Force personnel. The perpetrator, Adil Ahmad Dar, was identified as a Jaish-e-Mohammed member, making this a cell-facilitated rather than purely lone-actor attack. This illustrates a significant limitation of lone-actor frameworks in the South Asian context: Indian-subcontinent extremist violence frequently involves cross-border facilitation from Pakistan-based organisations. The NIA undertook the primary investigation under the Unlawful Activities (Prevention) Act 1967 (as amended). BNS § 113 creates the general offence of terrorist act; the UAPA remains the primary prosecution statute for terrorism in India.

Utøya, July 2011 (Norway). Anders Breivik killed 77 people, 8 in an Oslo bomb attack, 69 on Utøya island, in the deadliest peacetime attack in Norwegian history. The initial psychiatric evaluation found paranoid schizophrenia; a second court-ordered evaluation found narcissistic personality disorder without psychosis and legal responsibility. The court accepted the second evaluation. Retrospective TRAP-18: all eight proximal warning behaviours are documented, making Utøya the most complete illustration of the instrument used in training. The case also shows how extensive the behavioural record can be, years of planning, an 1,500-page published manifesto, without triggering pre-attack intervention.

United Kingdom: Channel and Prevent. Prevent is the UK's multi-agency counter-terrorism prevention strategy; Channel is its referral pathway. Any public-sector professional (teacher, social worker, healthcare professional, police officer) can refer an individual to a Channel panel. The panel assesses vulnerability and, where identified, creates a voluntary support plan, mentoring, educational support, mental health services, or ideological counter-narratives.

Channel has attracted substantial academic and civil-society criticism: disproportionate referral of Muslim young people relative to the actual distribution of terrorism risk, conflation of holding extreme religious or political views with violence vulnerability, and the absence of rigorous outcome evaluation for support plans. The government's independent Shawcross review (2023) concluded the programme had under-focused on far-right extremism in some regions while over-focusing on Islamist extremism. The 2023 Prevent statutory guidance issued under the Counter-Terrorism and Security Act 2015 updated referral thresholds in response.

Germany: EXIT. Founded by Bernd Wagner in 2000, EXIT focuses on former far-right extremists using motivational interviewing, social support, and practical assistance (employment, housing, identity document changes). Published outcome data show substantial sustained disengagement rates among participants, though the absence of a randomised control group limits causal inference. EXIT is the most extensively studied deradicalisation programme in the world.

India: NIA and UAPA framework. The National Investigation Agency Act 2008 established the NIA after the 2008 Mumbai attacks. The UAPA 1967 (as amended 2019, with the individual-designation power) is the primary prosecution statute. The 2019 amendment, allowing the Ministry of Home Affairs to designate specific individuals as terrorists rather than only organisations, has been challenged in the courts on due-process grounds. India does not yet have a national institutionalised deradicalisation programme comparable to Channel or EXIT; individual state police forces run community engagement initiatives in high-tension contexts, but these lack national standardisation or outcome evaluation.

Disengagement vs deradicalisation is the conceptual distinction that determines programme design. Disengagement, ceasing to participate in extremist activity, is substantially more achievable than deradicalisation (changing underlying beliefs). Research by John Horgan (Georgia State) and Anne Speckhard (interviews with former IS fighters) consistently shows that programmes providing alternative social identity pathways and practical support achieve disengagement; programmes focused on direct ideological challenge rarely achieve deradicalisation and sometimes harden views.

Paul Gill, John Horgan, and Paige Deckert's systematic case study analysis of 119 lone-actor terrorists (Crime and Delinquency, 2014) found approximately 32% had a documented mental disorder, compared with approximately 15% of the general population. Elevated rates are real but tell a nuanced story: the overwhelming majority of individuals with severe mental illness never commit violence, and the majority of lone-actor attacks are not attributable to psychotic motivation.

Specific clinical configurations appearing at elevated rates: narcissistic personality features with a specific grievance narrative (the "aggrieved entitlement" pattern prevalent in incel-motivated attacks); paranoid ideation fused with ideological content without reaching full psychotic disorder; depression combined with ideological motivation for suicide-by-attack (the "lethal suicide" pattern, Joiner, Van Orden and Witte); and mixed affective-ideological presentations.

TRAP-18 Distal Characteristic 7 (relevant mental disorder) explicitly notes that mental disorder increases risk when it is intertwined with other distal characteristics and proximal warning behaviours, not when present in isolation. This is the framing forensic assessments must reproduce.

For court reports in terrorism-related cases, three questions structure the psychological contribution: (a) is there a diagnosable mental disorder, and does it engage BNS § 22 in India, the MPC substantial-capacity standard in the US, or the M'Naghten and partial-responsibility provisions in the UK; (b) does the mental state interact with ideological motivation in ways that increase or decrease future violence risk; and (c) are mental health treatment needs present that form a component of an effective management plan?