Inoculating against extremism

Author: Nabil F. Saleh, Jon Roozenbeek, Fadi A. Makki , William P. Mcclanahan & Sander Van Der Linden
Published: 2021

The Internet is gaining relevance as a platform where extremist organizations seek to recruit new members. For this preregistered study, we developed and tested a novel online game, Radicalise, which aims to combat the effectiveness of online recruitment strategies used by extremist organizations, based on the principles of active psychological inoculation. The game “inoculates” players by exposing them to severely weakened doses of the key techniques and methods used to recruit and radicalize individuals via social media platforms: identifying vulnerable individuals, gaining their trust, isolating them from their community and pressuring them into committing a criminal act in the name of the extremist organization. To test the game’s effectiveness, we conducted a preregistered 2 × 2 mixed (pre–post) randomized controlled experiment (n = 291) with two outcome measures. The first measured participants’ ability and confidence in assessing the manipulativeness of fictitious WhatsApp messages making use of an extremist manipulation technique before and after playing. The second measured participants’ ability to identify what factors make an individual vulnerable to extremist recruitment using 10 profile vignettes, also before and after playing. We find that playing Radicalise significantly improves participants’ ability and confidence in spotting manipulative messages and the characteristics associated with vulnerability.


Online recruitment by violent extremist organizations has proven to be a pernicious problem. In recent years, hundreds of US and European citizens have joined extremist organizations in the Middle East or were apprehended in the attempt (Brumfield, 2014). Such organizations increasingly rely on the Internet, as it is considered to enable a safer approach with lower chances of being tracked by law enforcement agencies (Rashid, 2017). Extremist organizations are also active on social media platforms, such as Twitter, to spread propaganda and to recruit and radicalize new members (International Crisis Group, 2018). For example, after the outbreak of violence in Syria in 2011, radical groups became more active on Twitter, making it a hub for disseminating extremist content and directing users to a range of other digital platforms used by these groups (Stern & Berger, 2015). An analysis of approximately 76,000 tweets, captured over a 50-day period in 2013, revealed that they contained more than 34,000 short links to various kinds of jihadist content and connected a network of more than 20,000 active Twitter accounts (Prucha & Fisher, 2013).

As part of their recruitment efforts, radical groups typically share footage containing violence and music videos, as well as multilingual written content such as that found in glossy magazines (Prucha & Fisher, 2013; Hall, 2015). For example, in May 2013, Twitter was used to publicize the 11th issue of Al-Qaeda’s English-language online magazine, highlighting the use of social media by such organizations (Prucha & Fisher, 2013; Weimann, 2014). Aside from sharing propaganda content, radical groups have also used Twitter to communicate with sympathizers and to cultivate a community of supporters. The Islamic State has been particularly effective at this, reaching upwards of 100,000 people on Twitter between 2014 and 2015 (Berger & Morgan, 2015; Hosken, 2015). Another recruitment avenue that is commonly used by extremist organizations consists of direct messaging apps such as WhatsApp and Telegram, which are end-to-end encrypted and therefore difficult to monitor externally (Rowland, 2017; Radicalisation Awareness Network, 2019).

In response, social media companies have taken measures such as suspending accounts used by ISIS supporters, but to little avail, as duplicate accounts can be generated almost immediately (Berger & Morgan, 2015; Lewis, 2015). In addition, easy-access message encryption and other methods for “covering one’s tracks” make the effective tracking and following of extremist organizations and their members extremely difficult (Graham, 2016). Governments and international organizations have therefore also sought to prevent or counter violent extremism through skills development and education (Sklad & Park, 2017), youth empowerment, strategic communications and promoting gender equality (United Nations, 2015).

However, questions have been raised about the effectiveness of such efforts as well. Specifically, the lack of emphasis on “what works” for the prevention of violent extremism and the sparse usage of methods and insights from behavioral science and psychology to develop and test interventions have been noted (Schmid, 2013; Gielen, 2017; Holdaway & Simpson, 2018). Research has also highlighted some methodological issues with existing tools (Sarma, 2017), including validation and reliability problems, low base rates, the lack of generalizability of interventions and an inability to capture the diversity of the backgrounds of radicalized individuals (Knudsen, 2018). The radicalization process is exceedingly complex, and it is related to motivational dynamics, including a quest for identity, a search for purpose and personal significance (Kruglanski et al., 2014; Dzhekova et al., 2016), the pursuit of adventure (Bartlett et al., 2010; Dzhekova et al., 2016) or circumstantial and environmental factors such as extended unemployment or disconnection from society due to incarceration, studying abroad or isolation and marginalization (Precht, 2007; Bartlett et al., 2010; Doosje et al., 2016; Dzhekova et al., 2016). Doosje et al. (2016) break down these factors into micro-, meso- and macro-levels that cut across different stages of radicalization. The micro-level represents individual factors (e.g., a quest for significance or the death of a relative). The meso-level involves group-related factors (e.g., fraternal relative deprivation, or “the feeling of injustice that people experience when they identify with their group and perceive that their group has been treated worse than another group”; see Crosby, 1976; Doosje et al., 2016). Finally, the macro-level consists of factors at the societal or global level (e.g., accelerating globalization). It is important to note that those who harbor extremist sentiments might not necessarily engage in any active form of terrorism, and those who do so might only have a cursory understanding of the ideology that they claim to represent (Borum, 2011; Doosje et al., 2016; McCauley & Moskalenko, 2017). In fact, there appears to be broad agreement that there is no single cause, theory or pathway that explains all violent radicalizations (Borum, 2011; Schmid, 2013; McGilloway et al., 2015).

In a field of research that deals with difficult-to-access populations, oftentimes in fragile and conflict-affected contexts, mixed evidence exists as to what interventions are effective at reducing the risk of individuals joining extremist organizations (Ris & Ernstorfer, 2017). This highlights a strong need for an empirical and scientific foundation for the study of radicalization and the tools and programs used to prevent and combat violent extremism (Ozer & Bertelsen, 2018).

Inoculation theory

The most well-known framework for conferring resistance against (malicious) persuasion is inoculation theory (McGuire, 1964; Compton, 2013). Much like how a real vaccine is a weakened version of a particular pathogen, a cognitive inoculation is a weakened version of an argument that is subsequently refuted, conferring attitudinal resistance against future persuasion attempts (McGuire & Papageorgis, 1961, 1962; Eagly & Chaiken, 1993). The process of exposing individuals to weakened versions of an argument and also presenting its refutation has been shown to robustly inoculate people’s attitudes against future persuasive attacks. For example, a meta-analysis of inoculation research revealed a mean effect size of d = 0.43 across 41 experiments (Banas & Rains, 2010). In fact, McGuire’s original motivation for developing the theory was to help protect people from becoming “brainwashed” (McGuire, 1970).

In recent years, inoculation theory has been shown to be a versatile framework for tackling resistance to unwanted persuasion in a variety of important contexts. For example, inoculation theory has been applied in order to confer resistance against misinformation about contemporary issues such as immigration (Roozenbeek & van der Linden, 2018), climate change (van der Linden et al., 2017), biotechnology (Wood, 2007), “sticky” 9/11 conspiracy theories (Banas & Miller, 2013; Jolley & Douglas, 2017), public health (Compton et al., 2016) and the spread of fake news and misinformation (Roozenbeek & van der Linden, 2019; Basol et al., 2020; Roozenbeek et al., 2020b).

An important recent advance in inoculation research has been the shift in focus from “passive” to “active” inoculations (Roozenbeek & van der Linden, 2018, 2019). In passive inoculation, a weakened argument is provided and also refuted at the same time, in which case the individual would only passively process the exercise, such as through reading. In comparison, during active inoculation, an individual is more cognitively engaged and actively partakes in the process of refuting the weakened argument (e.g., by way of a game or a pop quiz). Research suggests that active inoculation can affect the structure of associative memory networks, increasing the nodes and linkages between them, which is thought to strengthen resistance against persuasion (Pfau et al., 2005). In addition, by focusing on the underlying manipulation techniques rather than tailoring the content of the inoculation to any specific persuasion attempt (Pfau et al., 1997), this advance offers a broad-spectrum “vaccine” that can be scaled across the population. This is especially important for behavioral research, which has been criticized for not tackling more complex social issues such as preventing violent extremism (van der Linden, 2018).