Cognitive sovereignty:
the blind spot of constitutional law.

The law of acts

French constitutional law was designed to protect liberties against acts. Administrative acts, legislative acts, police acts. Its entire architecture rests on this grammar: an act infringes a guaranteed liberty, the judge intervenes.

This grammar presupposes two implicit conditions. That the infringement be identifiable as an act. That the liberty be identifiable as a legal object. As long as these two conditions are met, the edifice works, and it works remarkably well: freedom of expression, the protection of personal data, the secrecy of correspondence enjoy one of the most sophisticated litigation arsenals in Europe.

What escapes the act

These two conditions cease to be met when what is at stake is not an isolated act but an infrastructure, and when what is affected is not an already-named liberty but the very condition in which a judgement can form.

A democracy is not reducible to the right to vote or to freedom of expression. It presupposes, in order to function, that citizens have the cognitive conditions from which to form a judgement: an informational space in which to deliberate, representations of the world that are not entirely preformed, a capacity to judge that is still exercised. These conditions do not appear among fundamental rights. They are not qualified as civil liberties. They do not constitute an identifiable object of litigation. Yet they are, increasingly, produced by private architectures whose logic is neither democratic, nor public, nor graspable through existing categories.

The finding is not technical. It is constitutional.

The neighbouring categories

One will object that law and legal thought have taken hold of the matter. The objection deserves to be taken seriously, because it is almost accurate. Five bodies of categories today approach the object. Each names a fragment of it. None covers it.

Cognitive security, first, as developed by cybersecurity and informational defence: it defends cognition against manipulation, disinformation, influence operations. Its register is defensive and operational; its object is the attack, not the ordinary condition of judgement in the absence of any attack.

Neurorights, next: cognitive liberty, mental privacy, mental integrity. They have entered positive law, notably through the French regulation of brain-imaging techniques. But they remain moored to neurotechnology, to interfaces, to cerebral data. Their object is the affected brain, not ordinary judgement under an algorithmic environment.

Freedom of thought, third, the forum internum of international conventions: an absolute right to keep one's thoughts, not to be manipulated, not to be punished for them. It is the legally closest category, and that is what makes it instructive: it protects the interior against interference. It forbids intrusion; it says nothing of what must exist for judgement to form.

The positive law of manipulation, fourth: the European regulation on artificial intelligence prohibits certain systems deploying subliminal, manipulative or deceptive techniques, where they impair the capacity to make an informed decision and are likely to cause significant harm. A real anchor in hard law. But its trigger is characterised manipulation and individual harm, not the collective and enduring condition of deliberation.

Cognitive sovereignty, finally, a term now in wide circulation, carried of late by a wave of work, mostly anglophone, dispersed across ethics, philosophy, engineering and human rights. Several reach positions close to the finding above. None stands within the register of continental public constitutional law; none reasons from the French jurisdiction.

Locating the void

The picture above makes it possible to situate the lacuna exactly, and to distinguish it from what it is not.

The void is not the absence of protection: manipulation is prohibited, brain imaging is regulated, the forum internum is guaranteed. Nor is the void the absence of a term: the term is now claimed on all sides. The void is the absence of a legal object of its own. Each of the five categories protects something against something: a cognition against an attack, a brain against an intrusion, a forum internum against an interference, a decision against a manipulation. All presuppose the object already there, and defend it. None asks what the law would need to be able to see for the condition of judgement to exist before it as an object.

It is a lacuna of a particular kind: it cannot be filled by extending a neighbouring category, for each extension imports the register of its category of origin, the attack, the intrusion, the interference, the harm, whereas what is at stake presents itself under none of these forms. It presents itself as an ordinary infrastructure, adopted without constraint, functioning without incident.

The existing categories treat the symptoms: data protection, platform regulation, the fight against manipulation. They do not name the object.

Closing

A democracy does not presuppose only guaranteed liberties. It presupposes the conditions in which a judgement can form, and these conditions have today, in French constitutional law, no name, no status, no judge.

As long as this condition remains unnamed, it will continue to be what it is today: a blind spot.