Pourquoi le « bar à salades » pourrait réellement contribuer à la recherche sur l'extrémisme – Une réponse à Horgan et Shayler

Introduction

In its special issue on the “Psychology of Violent Extremism”, this journal recently published a contribution by John Horgan and Morgan Shayler (2026), titled “Why The ‘Salad Bar of Ideologies’ Does Not Help Us Understand Contemporary Violent Extremism.” The piece is timely and important, because it takes stock of recent academic and law-enforcement efforts to define, understand, and eventually tackle forms of political violence that are not clearly motivated or underpinned by a single, coherent ideological worldview. True to their previous work, the authors offer an insightful and sophisticated take on the phenomenon; their contribution amounts to a call to rigorously assess the situation instead of uncritically following the fast flow of political announcements, law-enforcement demands, funding calls, and scientific competition.

At the heart of the paper lie three interlocking arguments. First, our epistemology is inconsistent: the proliferation of partially overlapping labels in policy and scholarship (such as “hybrid”, “composite”, “mixed, unclear and unstable”, let alone “salad bar” extremism) undermines our ability to comprehend the problem. Second, our ontology is naïve: the problem itself could be a measurement artefact rather than a reality, and claims about its novelty may thus be inflated. Third, our empirics are scarce: the processes and causes driving the phenomenon remain largely uncharted (which is unsurprising given the two previous flaws).

Part of a Special IssuePsychology of Violent ExtremismThis article belongs to a curated special issue. Discover more research on this topic from the full collection.Explore the special issue →

Whilst I am sympathetic to these three arguments and support the necessity to voice them, I nonetheless think that the authors exaggerate or underelaborate each of them. Below I examine them one by one, hoping to initiate a dialectical effort towards a more nuanced view – one where the “salad bad” metaphor might well have its helpful uses.

Proliferating Labels

First, the authors lament on the proliferation of partially overlapping and ill-defined concepts covering slightly different realities. They rightly diagnose “an urgent need for conceptual clarity and terminological consistency”, and their table listing these concepts and tracing their uses in the literature is a useful effort not only to visualise this “proliferation” but also to review the field.

This presentation, however, does not really pay justice to the dynamic nature of concept development, that is, to the fact that (most of) these new terms have emerged from a nascent and lively scholarly discussion, where critical examinations of existing concepts and labels feed proposals of less imperfect ones. For instance, Gartenstein-Ross and colleagues’ (2022, 2023) contributions are, precisely, conceptual attempts to correct the problems with everyday metaphors like the “salad bad”, and to consolidate earlier academic categories. Their “composite extremism” (and its four main types) better captures the different realities bundled together in earlier work, or vaguely appraised by imperfect law-enforcement classifications. Horgan and Shayler (2026) are right in noting that there is no “consensus”, but I contend that we are heading in that direction, bearing in mind that a full consensus is never achievable in our complex field (consider our concepts like “terrorism”, the “far-right”, or “extremism”). The authors’ 2×2 typology is, itself, a valuable contribution to this endeavour – but also, as such, one more instance of the conceptual “proliferation” criticized by the authors (indeed it neither corrects the problems with Gartenstein-Ross et al.’s typology, nor does it bring the conceptual effort to completion).

There are indeed too many labels trying to capture extremist ideological hybridity or confusion, but we should not negatively interpret this as a proliferation (literally, an increasing number and spread of something) because concept selection is a normal stage in early collective efforts to comprehend a new phenomenon. Some of these terms are not used any more, and others become adopted much more than others. What’s more, some serve different functions in different contexts: the ‘salad bar’ metaphor itself, for instance, has never been more than an informal label conveniently used, in scholarly and practitioners’ interactions, to quickly pin down roughly what the discussion is about – before honing it down (it is not, and was not intended to be, a rigorous concept).

Overstated Reality

But Horgan and Shayler’s critique is more profound: indeed they doubt that there is even a distinct phenomenon that needs a name. In other words, the conceptual proliferation and fuzziness may simply reflect an unclear – or perhaps inexistent? – object. As such, their contribution is not merely epistemological (how do we name, know, and measure a reality?) but ontological (what is this reality, what is its nature?). As it has been the case with the “terrorism” concept, this line of argument is very important to voice, because it unlocks a whole new level of analysis and fascinating insights.

Above all, Horgan and Shayler contest that hybrid extremism is new, in other words that it is a distinct phenomenon that ought to be conceptualized as such.^a They repeat that scholars and practitioners “overstate the novelty” of the ‘salad bar’ phenomenon, asking whether we are “only seeing more of this because the number of cases are increasing, or is it because we are getting better at identifying them?” (p.13). In other words, what is new is not the phenomenon but the “analytical lens through which practitioners and analysts are now interpreting (voluntary or otherwise) such behaviors” (p. 12). Whilst the authors’ observation that extremist hybridity has always existed (both at the group/organisational and individual levels) is correct, their argument is incomplete and therefore unconvincing: no-one contests that ideological cross-pollination, collaborations, and ideational patchworks have always been there, what scholars and practitioners rather observe is both a significant intensification of hybridisation (more cases, and more sources of ideological inspiration) and a change in what hybridity looks like. To be sure, even “coherent” extremist/terrorist ideologues of the more or less distant past offer worldviews that bear the mark of composite intellectual genealogies (consider for instance Ted Kaczynski, or Salafi-jihadists like Sayyid Qutb or Abdullah Azzam), but this hybridity comes nowhere close to what law-enforcement and expert witnesses find in the digital footprints of contemporary offenders, or to the “manifestos” penned by the likes of Ethan Miller or Solomon Henderson. The latter, for example, conglomerated Islamophobic, anti-Semitic, racist (both black and whites), anti-West, anti-Israel (and Palestine), anti-capitalist, and anti-tech ideations. So stark is the difference that one could not reasonably amalgamate these two families of cases into a single category.

Practitioners’ testimonies and official statistics point to an evolution too significant to be plausibly understood as a fad or a measurement artefact. A 2023 audit of the US Department of Justice’s “Strategy to Address the Domestic Violent Extremism Threat” (US Department of Justice Audit Department, 2023) clearly indicates that consulted DOJ and FBI officials complained that their work has become increasingly difficult due to “the spread of ideologies over the Internet and the prevalence of ‘salad bar’ ideologies that incorporate multiple violent ideologies, making it difficult to categorize them” (p. 15). The UK decided early on the need for a non-residual official category capturing hybrid/mixed ideology because frontline practitioners faced too many cases that proved impossible to categorize as “extreme right”, “Islamist”, etc. In the Prevent radicalization detection programme, the “mixed, unstable or unclear” (MUU) ideology category initially adopted ended up representing 51% of referrals in 2019/20, up from 11% in 2016/17. Replacing this catch-all category with the more granular triad of “multiple ideologies”, “fascination with extreme violence”, and “no ideology”, led to an important observation: by far the biggest number of referrals last year (ending March 2025) belonged to this latter category (almost 5,000), and the two other categories had about 450 referrals each, respectively. In comparison, Islamist and extreme right-wing had roughly 850 and 1700 referrals each (UK Home Office, 2025).

This means that even though the two “classic” extremist ideologies still dominate (as such, this is not the “end of ideologies”), something else is going on: they are followed by a non-negligible (and rising) number of genuinely hybrid constructs that frontline practitioners cannot categorise, and dwarfed by a huge number of individuals immersed in various types of digital communities glorifying extreme violence. I have seen these ideological patchworks first hand in my own work as expert witness in terrorism trials and in my punctual assistance to police investigations, and observed hybridity even in cases where ideology seemed clear at first sight.

How, then, to comprehend the nature of the “salad bad” phenomenon?^b Horgan and Shayler (2026) themselves identify crucial elements – youth, better visibility of lone actors, nihilistic violence – yet do not pursue their ontological task deep enough to connect the dots. In an upcoming paper, Dante Williams and I (2026) suggest that the violence and extreme worldviews we now face and try to capture with labels like ‘salad bar’ or ‘nihilistic extremism’ are the logical by-product of three broad evolutions of contemporary societies.

First, the rise in extremist idiosyncrasy reflects a similar, well-documented increase in hybrid and atomistic systems of thoughts and spirituality throughout society (concomitant to the well-studied erosion of established ideologies and religions, see e.g., Beck, 2014, or Bognár, 2024, on “patchwork religiosity”), which is an important dimension of what Zygmunt Bauman (2000) called “liquid modernity” (i.e., the erosion, instability, and fluctuations in the identities and social structures constituting contemporary societies). Powerful and addictive social media affordances further reinforce short attention spans, quick ideological bites, and fragmentary engagement with political ideas. Why should we deny the reality of rising extremist mixing, unclarity, and instability where we observe similarly idiosyncratic, individualistic, and half-baked identity constructs all around us?

Second, the UK and US examples evoked above, together with multiple similar observations, encourage us to consider “MUU”-type violence as a subset of a much larger interest in committing mass fatality attacks, rather than a form of “terrorism” or even “extremism”. As Rousseau and her team (e.g., 2023, 2025) observed in their Quebec radicalization referral unit, referred youth tend to have a primary attraction and drive to commit violence, before pasting some (typically thin or incoherent) ideological gloss over their emotions, grievances, and urges. What we face is a pool of young people (overwhelmingly men) captivated and attracted by extreme violence and keen to join the list of mass crime offenders; sources of inspiration for these young men vary and cross-pollinize, ranging from mass shooters fandom to gore digital communities, from nihilistic online communities to political extremist content. As such, the rise in “composite extremism” is not separable from the rise in nihilistic violence (by far the main category of Prevent referrals, as we have noted) and mass shootings (especially in the US),^c as well as increasingly thin ideological intellectual engagement. As Lankford and Silva (2025) recently showed, many mass shooters over the past decades already embraced “inconsistent, mixed, or customized beliefs and attack for a combination of personal and ideological reasons” (p.749) without being labelled terrorists or extremists (see also Malkki, 2014), and as such Horgan and Shayler’s argument about novelty is partially right. They should have seen, however, that we now see more of this violence than before, and that it takes different forms in the context of younger generations growing in a world doomed by a highly anxiogenic “polycrisis” (e.g., Lawrence et al., 2024) against which search for meaning and nihilism are not unexpected reactions (Peden, 2025).

Finally, OECD states’ powers of surveillance have become so extensive^d that sustaining organized extremism and terrorism is more difficult than before (Hegghammer, 2021),^e leaving individuals to lone explorations on the internet where many inevitably encounter multiple ideological influences (more on this below). Meleagrou-Hitchens and Ayad’s (2023) Age of Incoherence report is centred around this observation: using the Profiles of Individual Radicalization in the US (PIRUS) dataset, they note that the number of radicalized young people with no formal allegiances or ties to recognized extremist or terrorist groups has jumped 311% in the past ten years alone, as compared to the past five decades. Between 2010 and 2020, the dataset showed a 413% increase in cases where the internet played a primary role in the radicalization process for those under the age of 30. Horgan and Shayler are therefore right in saying that composite extremism has become an object because we see it more, but they should have seen the (concerning) reason why we both at once see it more and have it more.

In sum, the “salad bar” metaphor is a reality, and one which – as always with terrorism and extremism – is a product and mirror of our societies’ deepest structures, currents, and problems.

Uncharted Causal Processes

Faithful to John Horgan’s seminal work on the pathways and causes of terrorism and radicalisation, the authors lament that the exact processes and dynamics producing MUU/composite extremisms remain largely uncharted. Given their epistemological and ontological criticisms, the opposite claim would have been surprising: indeed how could we possibly identify with precision the underlying mechanisms of a possibly artefactual phenomenon we cannot accurately name? Here they note that we have no evidence as to why some individuals construct composite worldviews online whilst others stay in echo-chambers, and they argue that conspiratorial thinking may be playing a major role that remains to be integrated in existing theories.

I would suggest, to the contrary, that evidence has been piling up on both these aspects, even though it remains incomplete. In our paper Where do Mixed, Unclear, and Unstable Ideologies Come From? (Brace, Baele & Ging, 2024, cited in Horgan and Shayler’s article), we used a large dataset of digital extremist behaviour to locate and measure the mechanisms that both keep pulling participants within single-ideology online spaces (centripetal forces) and push them towards other ideological ecosystems (centrifugal forces); we also showed that centrifugal forces – responsible for increased extremist hybridity – have been increasing since the COVID-19 lockdown. Other empirically strong studies have documented similar hybridity-producing mechanisms, and observed their novelty and proliferation. For example, Petersen and Johansen’s (2025) research on what they call “hybridized prefatory extremism” (HYPE) highlights the existence of online spaces wherein participants actively create hybrid forms of extremism (the “Coalfax” case is a good example). About a year earlier, Astley (2025) explained how extremist digital structures and individual pathways together produce “idiosyncratic modes of extremist engagement”. All these are documented empirical realities.

As per conspiratorial ideation, it has for quite some time now been integrated in both theories of extremism and political violence generally (including in my own work, e.g., Baele, 2019; Baele et al., 2026) and frameworks theorizing “salad bar” phenomena specifically. For instance, most recent and ongoing large-scale projects using network analysis to map online extremist ecosystems explicitly highlight the porous borders between these and conspiratorial digital spaces, and chart the connections through which conspiracy theories percolate into extremism (and vice-versa).^f Multiple studies have documented the diverse ways through which conspiratorial constructs cross-pollinize with extremist worldviews in digital settings (e.g., Petersen & Johansen, 2026).

In sum, I would contend that the most significant mechanisms through which at risk individuals end up building hybrid worldviews are by now well-known, as are the dynamics through which most of these composite ideations contain conspiratorial elements as well as far-right tropes;^g these mechanisms and dynamics are indisputably more present now than before. The “salad bar” metaphor may thus be pointing to a reality with surprising accuracy – that of a digital landscape offering a variety of ideologies and worldviews in very close proximity, where some individuals actively steer visitors to taste around (centrifugal ideological effect), some (in some cases the very same ones) encourage them to keep having more of the same (centripetal ideological effect), and some even prepare and advertise ready-made dishes containing several sorts of salad (HYPE).

Conflicts of Interest

The author declares no conflicts of interest.

Endnotes

a. Perliger’s (2026) engaging reply to their article doubles down on this claim)

b. Note here how we use this concept as a usefully vague label capable of roughly situating the discussion.

c. The US Gun Violence Archive (GVA) shows an increase from less than 300 mass shootings in 2013 to more than 600 in 2020, moving from roughly 1000 injuries in 2013 (committed by about 200 perpetrators, and with roughly 250 deaths) to almost 3000 in 2021 (about 500 perpetrators and 700 deaths) (Donnelly et al., 2023).

d. For example, the 2024 budget for the US National Intelligence Program (NIP) was $76b., plus $30b. for the US Military Intelligence Program (NIP) – up from $43b. and $20b. respectively in 2007. In the EU, Europol saw its budget rise eightfold between 2000 (about €30m) and 2025 (about €240m).

e. The current direction of US politics offers an interesting evolution of this trend, however.

f. See for example Maik Fielitz’ project mapping the German Telegram far-right ecosystem ( https://machine-vs-rage.net/ ).

g. Meier (2023) made this observation, and provocatively argued that the “salad bar” label is a “distraction” obfuscating the expansion of far-right ideologies. Like Horgan and Shayler’s arguments, this claim has a valid basis yet overtly simplifies the situation.

Readers of this article also read

‹

Perspective

Why the ‘salad bar of ideologies’ does not help us understand contemporary violent extremism

by John Horgan & Morgan Shayler Cited by 5

Commentary

The erosion of conceptual clarity in the study of political violence

by Arie Perliger Cited by 3

Commentary

Extremism without ideology

by Quassim Cassam Cited by 2

Commentary

The end of ideological violence

by Clark McCauley

Research Article

Deadly persuasion: A multilevel analysis of leader influence tactics and lethality in violent extremist organizations

by Mackenzie Harms, Alex Bruens, Justin Jones, & Gina Ligon

Review

Adverse childhood experiences and violent extremism: A scoping review of the literature

by Teja Pammi et al.

Commentary

Shifts and continuity in contemporary violent extremism: Moving toward a dynamic and developmental understanding?

by Cécile Rousseau & Samuel Veissière

Research Article

System-disillusioned youth radicalize: The role of political alienation in the formation of political radicalism

by Marta Miklikowska & Tomasz Besta

›

References

Astley, J. (2025). Ideological Idiosyncrasies and Individual-level Radicalisation Pathways: Unpacking the Mutating Dynamics of Contemporary Extremism. Journal of Policing, Intelligence & Counter Terrorism, 20(3), 380-398. https://doi.org/10.1080/18335330.2025.2502580

Baele, S. (2019). Conspiratorial Narratives in Violent Political Actors’ Language. Journal of Language & Social Psychology, 38(5-6), 706-734. https://doi.org/10.1177/0261927X19868494

Baele, S., Boyd, K., Coan, T., & Naserian, E. (2026). Lethal Words: An Integrated Model of Violent Extremists’ Language. Studies in Conflict & Terrorism, 49(2), 119–144. https://doi.org/10.1080/1057610X.2023.2213963

Baele, S., & Williams, D. (2026). Mixed, Unclear, and Unstable: Three Risks of Incoherent Conceptualizations of Incoherent Extremism. Working paper, University of Louvain.

Bauman, Z. (2000). Liquid Modernity. Oxford: Polity.

Beck, U. (2014). A God of One’s Own: Religion’s Capacity for Peace and Potential for Violence. Oxford: Polity.

Bognár, B. (2024). Hybrid Social Spaces and the Individualisation of Religious Experience in the Global North: Spatial Aspects of Religiosity in Postmodern Society. Religions, 15(2), 24. https://doi.org/10.3390/rel15020241

Brace, L., Baele, S., & Ging, D. (2024). Where Do ‘Mixed, Unclear, and Unstable’ Ideologies Come From? A Data-driven Answer Centred on the Incelosphere. Journal of Policing, Intelligence & Counter Terrorism, 19(2), 103-124. https://doi.org/10.1080/18335330.2023.2226667

Donnelly, M., Grigorian, A., Inaba, K., Nguyen, N., de Virgilio, C., Schubl, S., Paladugu, A., Swentek, L., & Nahmias, J. (2023). Trends in Mass Shootings in the United States (2013–2021): A Worsening American Epidemic of Death. American Journal of Surgery, 226(2), 197-201. https://doi.org/10.1016/j.amjsurg.2023.03.028

Gartenstein-Ross, D., Zammit, A., Chace-Donahue, E., & Urban, M. (2022). Composite Violent Extremism: A Radicalization Pattern Changing the Face of Terrorism. Lawfare, 22 November 2022, https://www.lawfaremedia.org/article/composite-violent-extremism-radicalization-pattern-changing-face-terrorism

Gartenstein-Ross, D., Zammit, A., Chace-Donahue, E., & Urban, M. (2023). Composite Violent Extremism: Conceptualizing Attackers Who Increasingly Challenge Traditional Categories of Terrorism. Studies in Conflict & Terrorism, online before print. https://doi.org/10.1080/1057610X.2023.2194133

Hegghammer, T. (2021). Resistance Is Futile. The War on Terror Supercharged State Power. Foreign Affairs, 100(5), 44-53. https://www.jstor.org/stable/27121371

Horgan, J., & Shayler, M. (2026) Why The ‘Salad Bar of Ideologies’ Does Not Help Us Understand Contemporary Violent Extremism. advances.in/psychology, e607682, https://doi.org/10.56296/aip00057

Lankford, A., & Silva, J. (2025) What Effect Does Ideological Extremism Have on Mass Shootings? An Assessment of Motivational Inconsistencies, Risk Profiles, and Attack Behaviors. Terrorism & Political Violence, 37(6), 749-768. https://doi.org/10.1080/09546553.2024.2372427

Lawrence, M., Homer-Dixon, T., Janzwood, S., Rockström, J., Renn, O., & Donges, J.F. (2024) Global Polycrisis: The Causal Mechanisms of Crisis Entanglement. Global Sustainability, 7(e6), 1-16. https://doi.org/10.1017/sus.2024.1

Malkki, L. (2014) Political Elements in Post-Columbine School Shootings in Europe and North America. Terrorism and Political Violence, 26(1), 185–210. https://doi.org/10.1080/09546553.2014.849933

Meier, A. (2023) ‘Salad Bar Extremism’ as White Distraction: The (In)coherence of a Category. GNET Insights, 24 August 2023. https://gnet-research.org/2023/08/24/salad-bar-extremism-as-white-distraction-the-incoherence-of-a-category/

Meleagrou-Hitchens, A., & Ayad, M. (2023) The Age of Incoherence? Understanding Mixed and Unclear Extremism. Washington DC.: George Washington Program on Extremism. https://extremism.gwu.edu/sites/g/files/zaxdzs5746/files/2023-06/the-age-of-incoherence-final.pdf

Peden, M. (2025) Nihilistic and Apocalyptic Violent Extremism: Symbolic Rupture and the Crisis of Meaning in the Digital Age. VOX-Pol Blog, 29 July 2025. https://voxpol.eu/nihilistic-apocalyptic-violent-extremism/

Perliger, A. (2026). The Erosion of Conceptual Clarity in the Study of Political Violence. advances.in/psychology, e824472, https://doi.org/10.56296/aip00059

Petersen, L., & Johansen, M. (2025) Spaces of Hybridized Prefatory Extremism (HYPE) on Social Media. Social Media + Society, 11(2), 1-11. https://doi.org/10.1177/20563051251340145

Petersen, L., & Johansen, M. (2026). The Violence of Online Conspiracy Theories. New Media & Society, 28(4), 1464-1481. https://doi.org/10.1177/14614448251336431

Rousseau, C., Frounfelker, R., Ngov, C., & Crocker, A. (2023). Clinical Services Addressing Violent Extremism: The Quebec Model. International Journal of Forensic Mental Health, 22(3), 222-232. https://doi.org/10.1080/14999013.2022.2147254

Rousseau, C., Miconi, D., Ngov, C., & Hassan, G. (2025). Ideological and Nihilistic Violence in Adolescents Referred to a Specialized Clinic for Violent Extremism. Canadian Journal of Criminology & Criminal Justice, 67(2). https://doi.org/10.3138/cjccj-2024-0057

UK Home Office (2025). Individuals Referred to and Supported through the Prevent Programme, April 2024 to March 2025. https://www.gov.uk/government/statistics/individuals-referred-to-prevent-to-march-2025/individuals-referred-to-and-supported-through-the-prevent-programme-april-2024-to-march-2025

US Department of Justice, Audit Department (2023). Audit of the Department of Justice’s Strategy to Address the Domestic Violent Extremism Threat. Washington DC.: US Department of Justice. https://oig.justice.gov/reports/audit-department-justices-strategy-address-domestic-violent-extremism-threat