Cognitive warfare: what seven years of military-civilian research reveals

Attack­ing the enemy’s thought pro­cesses is not a par­tic­u­larly ori­gin­al concept: the prac­tices developed by the mas­ters of Soviet dis­in­form­a­tion provide very con­crete and rel­at­ively well-doc­u­mented examples of this. How­ever, due to the com­plex­ity of the oper­a­tions involved and the resources required, these prac­tices are more akin to small-scale crafts­man­ship than mass pro­duc­tion and are still thought of as a sec­ond­ary form of dis­in­form­a­tion. Nev­er­the­less, at the begin­ning of the 21^st Cen­tury, advances in neur­os­cience and a bet­ter under­stand­ing of how the brain works sug­gest that cog­nit­ive pro­cesses can now be tar­geted in a more sci­entif­ic manner.

In 2017, the term “cog­nit­ive war­fare” was used for the first time by Vin­cent Stew­art, dir­ect­or of the US Defence Intel­li­gence Agency (DIA). How­ever, it was more of a buzzword than a sci­en­tific­ally defined concept. Some­time later, at the end of 2018, “cog­nit­ive war­fare” was still just a con­veni­ent expres­sion used to describe all forms of inform­a­tion­al and psy­cho­lo­gic­al manip­u­la­tion. The term was cir­cu­lat­ing in aca­dem­ic circles and was often accom­pan­ied by ref­er­ences to sci­ence fic­tion or cyber­net­ics. The first attempts at open­ing it up centred around a mix of ima­gin­ary pro­spects, war games and stra­tegic com­mu­nic­a­tion. These approaches were inter­est­ing in that they raised aware­ness among insti­tu­tions, stim­u­lated stra­tegic ima­gin­a­tion and allowed pos­sib­il­it­ies to be explored. But they belonged to anoth­er register: that of pro­jec­tion, not measurement.

From 2022 onwards, anoth­er pro­ject was launched in the major centres of French mil­it­ary doc­trine. The armed forces ceased to treat cog­nit­ive war­fare as a future-ori­ented top­ic and instead approached it as an observ­able sys­tem. The Centre de doc­trine et d’enseignement du com­mandement (CDEC) (Com­mand Doc­trine and Train­ing Centre) then con­duc­ted a series of in-depth ana­lyses cov­er­ing the peri­od 2022–2023. In 2023, the Centre d’enseignement milit­aire supérieur-Terre (CEMST) (Centre for High­er Mil­it­ary Edu­ca­tion – Land) took over, incor­por­at­ing mod­el­ling and tools from the decision sciences.

This res­ul­ted in the concept of Cog­nit­ive Net Assess­ment (CNA), based on the work of Andrew Mar­shall and intro­duced in reports coordin­ated by Lan­glois-Ber­th­el­ot (2023–2024). These res­ults mark the decis­ive junc­tion between this insti­tu­tion­al work and sci­entif­ic research. Where “war­games” explore future scen­ari­os, CNA seeks to con­struct a rig­or­ous meth­od. Rather than ima­gin­ing the future, it seeks to under­stand the mech­an­isms of sta­bil­ity and imbal­ance in con­tem­por­ary cog­nit­ive environments.

These mech­an­isms have become increas­ingly import­ant since 2022, when gen­er­at­ive arti­fi­cial intel­li­gence (AI) became access­ible to the gen­er­al pub­lic. These AI mod­els are a game-changer, as they allow cog­nit­ive war­fare to move bey­ond small-scale oper­a­tions and enter the era of “mass pro­duc­tion” and, there­fore, exist­en­tial threat.

Tra­di­tion­al Net Assess­ment, developed by the Pentagon in the 1970s, com­pared dynam­ics rather than stat­ic resources. It assessed real asym­met­ries, speeds of adapt­a­tion and slow dis­rup­tions. CNA applies the same logic to the sphere of per­cep­tion and col­lect­ive decision-mak­ing. It does not seek to map dif­fuse nar­rat­ives or “influ­ences”, but rather to under­stand how a col­lect­ive main­tains or loses its inter­pret­at­ive coher­ence in the face of inform­a­tion flows.

Three con­cepts struc­ture this approach: (1) Decision-mak­ing super­pos­i­tion: the moment when sev­er­al con­tra­dict­ory rep­res­ent­a­tions of real­ity coex­ist without any one pre­vail­ing; (2) Cog­nit­ive col­lapse: a sud­den shift towards a single nar­rat­ive, often under the effect of an emo­tion­al or inform­a­tion­al shock; (3) Cog­nit­ive entropy: a meas­ure of men­tal and inform­a­tion­al dis­order with­in a social sys­tem. These con­cepts reflect the con­vic­tion that cog­nit­ive war­fare should be treated as a ques­tion of dynam­ics rather than dis­course. CNA makes it an engin­eer­ing field whose object­ive is to under­stand the fra­gil­ity of a cog­nit­ive sys­tem to pro­tect and strengthen it.

As Lan­glois-Ber­th­el­ot demon­strated, CNA is based on two com­ple­ment­ary indic­at­ors: a cog­nit­ive entropy index, which meas­ures the dis­per­sion and redund­ancy of cir­cu­lat­ing nar­rat­ives, and a super­pos­i­tion ten­sion index, which estim­ates the prox­im­ity of a col­lapse threshold. Togeth­er, they make it pos­sible to identi­fy areas of cog­nit­ive instabil­ity and take action before a break­down occurs. What fun­da­ment­ally dis­tin­guishes this approach from the foresight exper­i­ments of the 2020s is this subtle bal­ance between oper­a­tion­al­ity and sci­entif­ic rigour. The res­ults can be repro­duced, com­pared and dis­cussed with­in a shared meth­od­o­lo­gic­al frame­work. The dis­cip­line is defin­it­ively mov­ing away from nar­rat­ive towards meas­ure­ment and con­trolled experimentation.

AI occu­pies a cent­ral place in this con­cep­tu­al archi­tec­ture. It accel­er­ates atten­tion cycles, pro­motes micro-tar­get­ing and cog­nit­ive isol­a­tion, but it also provides the tech­nic­al means to mod­el them through detec­tion of arti­fi­cial flows, sim­u­la­tion of inform­a­tion propaga­tion, and machine learn­ing from weak sig­nals. AI becomes a double mir­ror: a factor of instabil­ity on the one hand, and an instru­ment of obser­va­tion on the oth­er. In CNA, it enables the con­struc­tion of dynam­ic rep­res­ent­a­tions of men­tal envir­on­ments, not to mech­an­ic­ally pre­dict indi­vidu­al beha­viour, but to meas­ure the col­lect­ive cog­nit­ive load and resi­li­ence of a social sys­tem in the face of inform­a­tion­al disturbances.

This approach did not ori­gin­ate from a single school of thought, but from a need for inter­dis­cip­lin­ary con­ver­gence. Cog­nit­ive research­ers, data engin­eers and doc­trine officers find com­mon ground here. This fruit­ful hybrid­isa­tion has made it pos­sible to pro­duce a stable lan­guage – attract­ors, entropy, col­lapse – and to estab­lish last­ing bridges between mil­it­ary and sci­entif­ic cul­tures. The 2023–2024 reports laid the found­a­tions for this com­mon gram­mar. They enabled France to align its work with emer­ging inter­na­tion­al stand­ards, while estab­lish­ing its own path: a sci­ence of cog­nit­ive resi­li­ence rooted in meas­ure­ment, not speculation.

This sci­entif­ic momentum con­tin­ued in 2024 with a new mile­stone: the sys­tem­ic dia­gnos­is of cog­nit­ive vul­ner­ab­il­it­ies, coordin­ated by Lan­glois-Ber­th­el­ot in asso­ci­ation with Chris­tophe Gaie (“Ser­vices du Premi­er Min­istre”). Where­as Net Assess­ment sought to char­ac­ter­ise the over­all sta­bil­ity of an envir­on­ment, this approach aims to under­stand the pre­cise moment when a social sys­tem loses its abil­ity to self-reg­u­late. Draw­ing on the work of Bateson, Mor­in and Fris­ton, this approach con­siders social cohe­sion to be an emer­gent prop­erty of a sat­ur­ated inform­a­tion sys­tem. Crises do not res­ult solely from extern­al attacks, but from the intern­al amp­li­fic­a­tion of unreg­u­lated feed­back loops.

Soci­ety is rep­res­en­ted as a net­work of multi-scale inter­ac­tions between indi­vidu­als, insti­tu­tions and sym­bols. Cog­nit­ive vul­ner­ab­il­it­ies then appear as struc­tur­al effects observ­able over time: inform­a­tion over­load leads to emo­tion­al polar­isa­tion, which weak­ens medi­ation and accel­er­ates desyn­chron­isa­tion between social groups. The Lan­glois-Ber­th­el­ot and Gaie mod­el artic­u­lates three dimen­sions – col­lect­ive nar­rat­ives, insti­tu­tion­al medi­ations, polit­ic­al reg­u­la­tions – and eval­u­ates not what indi­vidu­als think, but the speed at which their rep­res­ent­a­tions are recon­figured. Cog­nit­ive sta­bil­ity then becomes the abil­ity to main­tain mul­tiple inter­pret­a­tions of real­ity without nar­rat­ive collapse.

Sev­en cog­nit­ive fields serve as res­on­at­ors of cohe­sion: nation­al belong­ing, mor­al eco­logy, social norms, his­tor­ic­al memory, insti­tu­tion­al legit­im­acy, stra­tegic autonomy, and inter-eth­nic cohe­sion. Ana­lys­ing their inter­ac­tions allows us to map cog­nit­ive entropy, com­par­able to an energy map of the social body. AI plays an obser­va­tion­al role, with semant­ic graphs detect­ing nar­rat­ive dens­i­fic­a­tions, identi­fy­ing cor­rel­a­tions between fields and meas­ur­ing cog­nit­ive trans­itions. It does not draw con­clu­sions: it facil­it­ates read­ing without repla­cing human interpretation.

An ini­tial intern­al exper­i­ment, con­duc­ted in a lim­ited, non-pub­lic set­ting, tested this meth­od­o­lo­gic­al archi­tec­ture. Without going into detail about the meth­ods used, this imple­ment­a­tion con­firmed the pos­sib­il­ity of dynam­ic mon­it­or­ing of cog­nit­ive cohe­sion and early detec­tion of areas of sym­bol­ic ten­sion. These par­tial res­ults, obtained with­in a lim­ited scope, are now guid­ing work towards the integ­ra­tion of these meas­ures into stra­tegic obser­va­tion sys­tems. The strength of the sys­tem lies in the con­ver­gence between mil­it­ary and civil­ian cul­tures. One provides long-term man­age­ment and the form­al­isa­tion of crit­ic­al thresholds; the oth­er provides a detailed under­stand­ing of sym­bol­ic dynam­ics. Togeth­er, they lay the found­a­tions for an applied sci­ence of cog­nit­ive sta­bil­ity, cap­able of meas­ur­ing social cohe­sion with meth­od­o­lo­gic­al rigour.

Today, cog­nit­ive war­fare has become a field of engin­eer­ing in its own right. CNA is its cent­ral archi­tec­ture: a mon­it­or­ing and sim­u­la­tion tool that replaces alert nar­rat­ives with objec­ti­fi­able indic­at­ors. The pri­or­it­ies for 2025–2028 are clear: con­sol­id­ate met­rics, integ­rate sim­u­la­tion into oper­a­tion­al plan­ning, and teach tem­por­al decision man­age­ment. After six years of gradu­al mat­ur­a­tion, cog­nit­ive war­fare is enter­ing a phase of equi­lib­ri­um: the meth­od is sta­bil­ising, the tools are becom­ing more pre­cise, and the approach is gain­ing in coher­ence without los­ing con­cep­tu­al caution. 

Thus, the work car­ried out by Lan­glois-Ber­th­el­ot and Gaie extends the move­ment begun with CNA: it shifts cog­nit­ive war­fare from the realm of spec­u­la­tion to that of sys­tem­ic engin­eer­ing, where cohe­sion becomes a meas­ur­able, and now tested, vari­able of nation­al resilience.