Cognitive training for war : how to prepare for uncertainty

Over the course of the ten-year per­iod bet­ween 2015 and 2025, cog­ni­tive pre­pa­ra­tion for war­fare has been the sub­ject of intense research and expe­ri­men­ta­tion, both within defence ins­ti­tu­tions and in the aca­de­mic and indus­trial worlds. French and allied doc­trines have conver­ged towards the same orien­ta­tion : trai­ning must become expe­ri­men­tal, ins­tru­men­ted and refutable.

In 2015, France’s Direc­to­rate Gene­ral of Arma­ments sta­ted that “the chal­lenge is no lon­ger to repro­duce, but to expe­riment”. In 2021, the Concept d’emploi des forces ter­restres (Land Forces Employ­ment Concept) pla­ced “confron­ting change” at the heart of cog­ni­tive super­io­ri­ty. Final­ly, Le Com­bat aéro­ter­restre en 2040 (Air-Land Com­bat in 2040), a text from the Com­man­de­ment pour le Com­bat Futur (Future Com­bat Com­mand), empha­sises the need for pre­pa­ra­tion based on constraint, speed, adap­ta­tion and beha­viou­ral robust­ness. These three texts, sup­por­ted by ana­lyses from the RAND Cor­po­ra­tion (2020–2022) and NATO MSG-190 (2021), esta­blish a consen­sus : trai­ning must increase adap­ta­bi­li­ty rather than confor­mi­ty to a model.

At the same time, the civil and para-aca­de­mic fields have deve­lo­ped a set of prac­tices grou­ped under the term spe­cu­la­tive sce­na­rio plan­ning. Ins­pi­red by cri­ti­cal desi­gn and pros­pec­tive inno­va­tion methods, these approaches use fic­tion as a tool for reflec­tion. Based on nar­ra­tives set in plau­sible futures, they make it pos­sible to ques­tion the social, poli­ti­cal or tech­no­lo­gi­cal effects of war by remo­ving over­ly res­tric­tive bar­riers of plau­si­bi­li­ty. Their sta­ted objec­tive is to sti­mu­late the ima­gi­na­tion, broa­den per­cep­tions of what is pos­sible and engage the public in thin­king about the future.

In spe­cu­la­tive sce­na­rio buil­ding, sur­prise is a nar­ra­tive device, not an expe­ri­men­tal variable

Cultu­ral­ly, the result is indis­pu­table : spe­cu­la­tive sto­ry­tel­ling has brought war back into the demo­cra­tic debate. Cog­ni­ti­ve­ly, howe­ver, stu­dies conduc­ted in both the mili­ta­ry and civi­lian spheres show that it has not pro­du­ced any mea­su­rable pro­gress in terms of pre­pa­red­ness. Nar­ra­tive devices create repre­sen­ta­tions ; they do not gene­rate mecha­nisms. Their effec­ti­ve­ness lies in unders­tan­ding, not in reac­tion. In some cases, we can even iden­ti­fy a per­for­ma­tive intent. For example, the CIA’s “The World in 20.. as seen by the CIA” reports, publi­shed at each US pre­si­den­tial elec­tion, are as much an exer­cise in sce­na­rio-based fore­sight as they are a poli­ti­cal sta­te­ment, pre­sen­ting the world with a par­ti­cu­lar stra­te­gic vision that shapes the minds of both the Uni­ted States’ allies and competitors.

Reports by the RAND Cor­po­ra­tion (Buil­ding Bet­ter Games, 2020) and NATO MSG-190 (2021) iden­ti­fy three recur­ring limi­ta­tions in these formats :

• linear sequen­cing, which is incom­pa­tible with the conti­nuous dyna­mics of infor­ma­tion warfare,
• delayed feed­back, which sepa­rates lear­ning from action rather than inte­gra­ting it,
• the absence of beha­viou­ral metrics, which pre­vents the nar­ra­tive from being lin­ked to obser­vable performance.

In spe­cu­la­tive sce­na­rio buil­ding, sur­prise is a nar­ra­tive device, not an expe­ri­men­tal variable. The cohe­rence of the nar­ra­tive takes pre­ce­dence over the mea­su­re­ment of the deci­sion. This is par­ti­cu­lar­ly evident in the IPCC’s cli­mate sce­na­rios, which present nar­ra­tives of pos­sible futures. They are des­cri­bed as nar­ra­tives trans­la­ted into sets of socio-eco­no­mic assump­tions. They make it pos­sible to iden­ti­fy the most signi­fi­cant risks for an orga­ni­sa­tion or ter­ri­to­ry by choo­sing a pro­bable future. They do not allow the impact of each variable to be measured.

Ins­ti­tu­tions that use these approaches pur­sue a legi­ti­mate goal of rai­sing awa­re­ness. The aim is to fos­ter a com­mon culture of com­plexi­ty and encou­rage stra­te­gic crea­ti­vi­ty. Howe­ver, none of the foun­ders¹ ever clai­med, or even sug­ges­ted, that these tools could pre­pare an indi­vi­dual for deci­sion-making in a dyna­mic envi­ron­ment or for the cog­ni­tive constraints spe­ci­fic to ope­ra­tio­nal situa­tions. Their objec­tive was to raise awa­re­ness, broa­den the ima­gi­na­tion, pro­duce repre­sen­ta­tions, open up dis­cus­sion and pro­voke intel­lec­tual debate. The cur­rent confu­sion the­re­fore stems less from a lack of tools than from an inap­pro­priate real­lo­ca­tion of their purposes.

But peda­go­gy, howe­ver par­ti­ci­pa­to­ry it may be, can­not replace expe­ri­men­ta­tion. The dif­fe­rence lies in the nature of the evi­dence : sce­na­rio plan­ning demons­trates through mea­ning ; expe­ri­men­ta­tion demons­trates through data. Sce­na­rio plan­ning raises awa­re­ness of the need ; expe­ri­men­ta­tion puts pro­ce­dures in place.

Mili­ta­ry-civi­lian stu­dies conduc­ted over the past ten years – in cog­ni­tive engi­nee­ring, deci­sion-making psy­cho­lo­gy and sys­te­mic model­ling – have shown that effec­tive trai­ning requires three pro­per­ties : time constraints (pro­lon­ged expo­sure to time pres­sure), endo­ge­nous feed­back (imme­diate effect of action on the envi­ron­ment) and mea­su­ra­bi­li­ty (pro­duc­tion of objec­tive data). This approach is alrea­dy being consi­de­red in the world of cyber­de­fence : for example, major USCYBERCOM exer­cises pit teams against each other, with each action modi­fying the ove­rall sys­tem, gene­ra­ting imme­diate feed­back and, above all, enabling per­for­mance to be mea­su­red. The results are convin­cing : after seve­ral years of trai­ning, some teams have achie­ved res­ponse time gains of over 30%. None of these cri­te­ria are ful­ly met by spe­cu­la­tive sce­na­rio plan­ning. Its contri­bu­tions remain qua­li­ta­tive. Spe­cu­la­tive sce­na­rio plan­ning is above all an exer­cise in com­mu­ni­ca­tion and mediation.

The pro­blem is also one of pace. Infor­ma­tion war­fare moves at the speed of the net­work ; nar­ra­tive devices ope­rate at the pace of dis­cus­sion. The gap bet­ween the speed of the attack and the slow­ness of the debate creates a metho­do­lo­gi­cal vul­ne­ra­bi­li­ty. As long as trai­ning does not ali­gn with the real time of the conflict, it will lag behind by one cycle of adap­ta­tion, or worse : it will create the false impres­sion that the threat is being taken into account (this is the cri­ti­cism often level­led at armies for pre­pa­ring for yesterday’s war ins­tead of anti­ci­pa­ting tomorrow’s). The urgen­cy is the­re­fore cog­ni­tive before it is poli­ti­cal : lear­ning fas­ter than the threat is changing.

The ten years of mili­ta­ry-civi­lian research stu­died here lead to a clear diag­no­sis. Spe­cu­la­tive sce­na­rio plan­ning is a method of accul­tu­ra­tion, not a trai­ning tool. It pro­motes awa­re­ness of the issues among the inex­pe­rien­ced public but can­not intrin­si­cal­ly deve­lop either resi­lience or res­pon­si­ve­ness. It remains use­ful for buil­ding sha­red ima­gi­na­ries, but it is expe­ri­men­tal simu­la­tion – open, ins­tru­men­ted, fal­si­fiable – that pro­duces las­ting learning.

It is not a ques­tion of oppo­sing two worlds, but of cla­ri­fying their res­pec­tive regimes of truth. Sce­na­rio plan­ning explains ; simu­la­tion veri­fies. The for­mer relies on the cohe­rence of dis­course, the lat­ter on the resi­lience of rea­li­ty. It is bet­ween these two approaches that the future of cog­ni­tive pre­pa­ra­tion is being played out today. At a time when civil and mili­ta­ry autho­ri­ties are reco­gni­sing the need for a cer­tain fru­ga­li­ty, we must nar­row our focus : long-term fore­sight will always remain essen­tial to the defence of the coun­try, but it can­not be achie­ved at the expense of imme­diate les­sons and adap­ta­tions to emer­ging threats.

For scrip­ting to be direct­ly pro­duc­tive, it must be accom­pa­nied by the abi­li­ty to change the para­me­ters of the story’s envi­ron­ment. Not to tell one par­ti­cu­lar sto­ry, but to allow hun­dreds to be ima­gi­ned. It is not so much the consis­ten­cy of the sto­ry that mat­ters as the weight that the para­me­ters car­ry in the eva­lua­tion of the tools or concepts invol­ved. This is the approach that led the aero­nau­tics indus­try to deve­lop the concept of digi­tal twins, in which pre­cise model­ling of air­craft, tur­bines and pro­duc­tion lines allows variables to be chan­ged and the imme­diate conse­quences to be obser­ved. It is thus pos­sible to replay the same ove­rall action an infi­nite num­ber of times, varying the para­me­ters and iden­ti­fying the consequences.

From then on, the sce­na­rio is, so to speak, inci­den­tal. The key thing is that the sys­tem model­ling is rele­vant. With a robust model, we can ima­gine any sce­na­rio we want. All we need to do is modi­fy the sets of rules of action for the various actors to assess the conse­quences of each of the variables.

Sce­na­rio plan­ning allows us to take an intel­lec­tual step back, which pre­vents us from imme­dia­te­ly dis­mis­sing the very pos­si­bi­li­ty of the object or cir­cum­stances under stu­dy. Expe­ri­men­ta­tion should allow us to ite­rate the para­me­ters until we pro­duce imme­dia­te­ly use­ful les­sons, rather than just food for thought.

Ten years of mili­ta­ry-civi­lian research have led to a clear diag­no­sis : sce­na­rio plan­ning explains, expe­ri­men­ta­tion veri­fies. The for­mer relies on the cohe­rence of dis­course, the lat­ter on the resis­tance of rea­li­ty. The future of cog­ni­tive pre­pa­ra­tion lies bet­ween these two approaches.