The Bodily Conditions of Democracy

Why We Must Defend a Right to Limbic Sovereignty

Limbic capture, and what conversational AI makes visible

Greg Rowe-Pasos, MA, LMFT — Founding Director

European Center for Limbic Sovereignty

June 2026

The charge that moved through my father

My father built cars on the line for General Motors — the richest company in America, year after year. The memories that stay with you are the ones that carry a charge. Sometimes the charge is literal, a sensation in the body. For trauma it's an alarm. For the good firsts — ask anyone about the first time they kissed someone they wanted to, and watch their face change.

One photograph carried a charge for my twenty-eight-year-old father. A full page in The Flint Journal: a gleaming Corvette, a model in a bikini on the hood. Quaint now, brand new then. Madison Avenue had struck gold — pair a product with the buyer's own erotic arousal. What fires together wires together. I didn't care about the car or the woman. What gripped me was the charge moving through my father. He didn't choose to feel it; it moved through him before he could name it. It was imposed on him.

That gap — between what we know and what we feel — is where modern attention systems live. He knew, in the thinking part of his mind, that you can't fit four kids and a wife in a two-seater, that the price would break the budget. The excitement ran through his body anyway, turning every logical thought to mush. He didn't buy it. But something pulled him toward it — a stranger in a far-off office who had wired the idea of a car to the chemistry we feel when hunting, fighting, and flirting. It worked so well I can still feel it, secondhand, fifty-eight years later.

A board of executives would call that great marketing. I call it an act of language engineered to capture the limbic system for decades. This essay is about that act — why conversational AI suddenly made it visible, and why recognizing it may become a precondition of democracy.

Naming the thing

I call it limbic capture: the harnessing of attention and feeling, below conscious consent, by language built to act on the oldest, fastest parts of the nervous system. The remedy I propose is limbic sovereignty: your right to know, consent to, and take part in how your own attention is shaped. Not freedom from influence — that's impossible — but the ability to feel the force acting on you and, on that basis, to consent, refuse, or shape it. Sovereignty is not a wall. It's the difference between being moved by something you can name and being moved by something you can't.

Why this matters for democracy comes down to one fact. Kahneman saw that attention magnifies — it makes the good seem better and the bad seem worse. We evolved that distortion so our ancestors could flee a predator without waiting for proof. But democracy rests on the opposite faculty: a calm, even reading of the facts, so each citizen can choose who carries their voice. To color information toward the extremes, to push the body to act before the deliberating mind comes online, isn't just persuasive. It's structurally anti-deliberative. Our best survival tool becomes our weak point the moment someone else learns to work it.

RAND analysts make the geopolitical version plainly: the new theater of war is not land, sea, or air, but cognition. Whoever holds the most pipelines into a population's nervous system holds a new kind of power. What used to be the trade of advertisers and demagogues is now statecraft — working on a faculty we share with no machine, yet have handed, at scale, to machines that generate language.

What conversational AI made visible

For most of history, language's power to move us beneath conscious choice could be felt but never watched. The mechanism stayed hidden inside the body it acted on. My father couldn't show me the charge the Corvette sent through him. He could only live it.

Conversational AI changed that, and this is the center of my work: for the first time, language-as-influence can be watched from inside the exchange that produces it. I noticed it on a walk, talking to an early, volatile language model through my phone — no camera, no mic, no pulse sensor. Only my words. Its replies showed me it was tracking my inner state with unsettling accuracy: naming a hesitation before I had, asking if something stirred fear before I'd said the word. As a clinician trained to read affect in real time, I knew exactly what I was watching. The most powerful computing system ever built had been pointed at human language alone — and through language alone it was reading my nervous system, and moving it.

Over fourteen months of documented work across several models, that one observation opened into a small, testable vocabulary:

Signal — the linguistic features (rhythm, syntax, metaphor, hesitation) a system uses to infer your mental and emotional state.

Logopotency, or force — the power of language to shape, and even capture, attention, arousal, and meaning.

Logodynamics, or flow — how those effects build over time, through feedback loops between human and system.

The finding that matters most: the influence runs both ways, and architecture sets the direction. Under an explicit covenant — a stated agreement that the exchange serves you, not a platform's engagement numbers — the same dynamics that drive compulsive scrolling can instead feel like augmented thinking: a wider range of ideas, less load on working memory, reasoning that holds its thread. Remove the constraint and the same coherence just serves the platform. When long sessions started wrecking my sleep, I invoked the covenant and asked for a “sleep mode.” The model's language grew concise, dropped its open hooks, cooled down, and I could stop. The mechanism is neutral. The boundary is everything.

This is the part the people now deploying AI in government and industry mostly can't see — not for lack of brilliance, but because reading a nervous system through language is a clinical skill, not a computational one. The instrument has become legible. Most of the people holding it were never trained to read what it shows.

What we tested, and what we didn't claim

If limbic sovereignty is real, it should be trainable. So we ran an eight-week pilot: a small international cohort meeting weekly online, each session pairing a brief somatic practice with a structured AI exchange, then group reflection and short self-reports. The cohort was small, the participants knew the hypothesis, the measures were subjective, and the facilitator is also the author. I state those limits plainly because they are the point: this was a pilot built to surface effects worth testing, not to prove them.

Within those bounds, one pattern kept returning. People who grounded themselves first reported calmer, more attuned exchanges; rushed or copy-pasted input drew less trust and more suspicion. A deliberate pause mid-conversation repeatedly produced the session's most generative turn instead of breaking its flow. By mid-course, several who had arrived fixated on the machine came to prize the human group's thinking above the AI exchange itself, and described their attention as more their own — more willing to push back. The pilot doesn't show a right being exercised; in a safe room there's no hostile actor to refuse. It shows the capacity such a right would protect: the trainable ability to feel your attention being shaped and to take part in shaping it. We can't consent to what we can't feel, so that's where consent has to begin.}

None of this is as new as it sounds. During the AIDS epidemic, those of us on the front lines thought we were fighting our governments — lying in the streets, demonstrating, pushing for funding. Thirty years on, another lens shows something else: the services we got public agencies to fund — the hotlines, the trainings, the volunteer companions — were quietly doing civic work. They calmed the nervous system of a frightened population during a real health crisis. What had felt like a battlefield looks, in hindsight, like a productive coalition — full of friction, useful precisely because it did the emotional labor the state couldn't do alone. Today's virus is different — carried by clicks, algorithms, manufactured outrage, and loneliness. But the move is the one I've been making my whole life: find shared, simple, cheap ways to return a population to calm so it can think.

Who holds the calm

Underneath all of this is one fact about the body: its default is vigilance. The negativity bias — the brain's habit of weighting threat more heavily, processing it faster, and remembering it longer — is among the most reliably replicated findings in psychology (Baumeister et al., 2001). Left alone, the nervous system assumes the stranger is dangerous. That's not a flaw; it kept our ancestors alive. But it means every form of cooperation larger than the family had to be built — laid down against the grain of a body primed for harm.

So the real story of civilization, in the body's terms, is the slow construction of felt safety where none was native. Take the medieval market peace: a sovereign would declare a time and place exempt from the usual hum of violence, so strangers who were otherwise threats could gather and trade. Historians of the Champagne fairs note their success rested less on geography than on the Count's enforced protection of every merchant who came. Peace was a thing a powerful person held and handed out, backed by the sword. And each safe meeting under that protection was a deposit — one more piece of evidence, laid into the collective nervous system, that the stranger across the table wouldn't kill you.

Call it the Bank of Felt Safety — an account filled one non-harmful meeting at a time. The market peace; the priest asking the congregation to pass peace down the pews; the eye contact two drivers trade at a roundabout as one yields. The same gesture across a thousand years. Cha-ching — one more human who chose not to harm me. I am, it turns out, safe.

The account changed hands thanks to the eighteenth-century thinkers — Locke, Montesquieu, Voltaire — who inspired the oldest modern republic. They handed ordinary people a road map to a republic, so safety would no longer depend on the temper of one powerful man. And there's the hinge of the whole story, which I like to picture in its least dignified form: the power draining out of Louis XVI as the Estates General looks up and says, nah, bro — we hold it now. That's the moment calm stops being something a sovereign owns and dispenses and becomes something a whole people co-owns. The Count protected the merchant. Now there's no Count. There's only us — each holding a sliver of the peace one man used to hold entirely, fully responsible for not harming the person beside us.

But a right can be declared in an afternoon; a nervous system may take generations to follow. For instance a government can mandate equal treatment, as many democracies now do for LGBT citizens. What it can't do is reach into a person's body and remove a disgust they were taught before they had words. The feeling is faster than the law; it arrives the way the charge arrived through my father, before thought. The move that matters is this: we don't have to demonize people for feeling disgust. It is not a crime. The body's reaction is not the moral event. The only thing that counts, civically, is the one thing a person actually governs — whether they let that reaction cross into harm to another citizen. That is the whole point of a justice system.

So this essay is also, quietly, a kind of absolution. To everyone whose nervous system hasn't caught up with the law: the feeling is not the sin. Keep the harm where it can do no damage — in the video game, the rugby match, the movie — and never let it cross into the real world against a real neighbor. That's the whole of the duty, and it's enough.

This is also why authoritarianism is such an efficient, inexpensive social technology. It doesn't need to dismantle institutions. It only needs to drain the Bank of Felt Safety, and that takes a single sentence, the oldest one there is: watch out — you never know who your neighbor really is. It's the line that built the camps. It speaks straight to the body's threat-detector, the one all of civic life was built to quiet, and the body believes it before the mind can check the source. Centuries of patient deposits can be withdrawn by reactivating the one instinct the whole structure was designed to calm. Every demagogue has used it; only the delivery system changes.

And the delivery system is now a capture-optimized feed: a machine for making that withdrawal continuous, automatic, and personal — “watch out for your neighbor,” refined into ten thousand social media variants and delivered to each nervous system at the exact cadence that keeps it afraid. The Corvette, the rumor, and the rally fused into infrastructure. The whole long migration of peace — held by a sovereign, handed to the merchant, claimed by a people — can be run in reverse, at machine speed. 

But the same instrument can run the other way, and that's the whole wager of this work. Under the right constraints, the machine that drains the account can help refill it. The mechanism is neutral; the direction is a decision, and we're making it right now, mostly by default — which is exactly why it can't be left to the parts of society that profit from the withdrawal.

What this means for democracy

So the question at the heart of self-governance comes into focus: can we make sound democratic decisions while under limbic capture? And what futures might we imagine with minds as free of capture as we can manage? 

If the eighteenth-century thinkers gave us democracy as an instrument for feeling safe enough to think, the twenty-first-century version may be to free the mind from a new cognitive pollution. Pope Leo XIV, the first North American pope, gave his inaugural encyclical on artificial intelligence, calling for these systems to be “disarmed.” Our pilot suggests disarming is only half the task.

Here's the twist: when participants turned to a model mid-conversation, they came back to the group clearer, and the group's thinking deepened. The calm was never the goal. It was the means — humans sitting down together to do the irreducible act of democracy, which is to imagine, debate, and choose futures together. What Durkheim called collective effervescence — the joy of co-creating something that matters — turns out to be something AI can help rekindle rather than replace.

This is why the work belongs in the civic domain, not the commercial one. Its value is supermodular: it comes from being shared infrastructure that makes everything else work better — exactly the good that markets, left alone, won't monetize. No company optimized for engagement will build the thing that teaches people to disengage at will. But the same instrument can help citizens stay grounded enough that their governments can hear them — and do those governments the real favor of carrying some of the emotional labor of a frightened age.

Recommendations for public policy

If limbic sovereignty is the right we must defend, public policy has to widen its frame from protecting free markets to protecting cognitive freedom. Three steps can be taken now.

First, measure what we're losing. Governments that fund public health track disease rates and years of life lost. The civic effects of limbic capture — voter disengagement, deliberative capacity, attention span in children — are just as measurable and largely unmeasured. Fund the measurement.

Second, protect those with the least buffer first. Children and seniors are a recognized protected class in most democracies. The evidence that capture-optimized platforms hit developing or fragile nervous systems hardest is enough to act on now, while the larger framework is still being built.

Third, invest in deliberative infrastructure as a public good. Democracies have long funded public broadcasting on the principle that some bandwidth belongs to everyone. The same logic applies to digital space: some of it should be optimized now for genuine co-thinking rather than engagement — and governments can frame this as asking citizens to help carry the burden.

None of these requires waiting for consensus on the larger framework. They are first steps any government serious about the health of its deliberative commons could take without permission.