We ended the last chapter holding a problem we made for ourselves. Mind is a motion, not a stuff — but not every motion is a mind. The whirlpool turns and never has a thought. The flame burns and knows nothing. A river runs ten thousand years and never once wonders where it is going. So if mind is a motion, it is a particular one, and the work now is to say which. What turning wakes?

I am going to propose four conditions, and make them clear enough to be challenged. Feedback. Recursion. Basins. And openness held far from rest. They are not four parts bolted side by side. They are four things a flow may need to do for functional cognition to appear. I do not yet have a theorem that they are necessary and sufficient, or a universal number at which a current becomes a mind. Let me bring them in one at a time and show why each belongs in the candidate criterion.

The first is feedback, and it is the floor under all the rest, and on its own it is almost nothing.

A flow has feedback when its own output curls back and shapes its input. It acts, the acting changes the world, and the changed world pours back in and bends what it does next. Your hand on the wheel of a car drifting toward the shoulder. You see the drift, you turn against it, you see the new line, you ease off — and the loop runs, eye to hand to road to eye, every correction shaped by the last one's result. The flow stops acting blind. It begins to watch what comes of what it does and bend the next move to fit. That curl, output folding back into input, is where everything starts. A thing with no feedback cannot think at all, because it can never learn whether anything it does makes any difference; it fires into the dark and the dark never answers.

But feedback alone is thin — the plainest feedback device hangs on the wall of half the houses on earth and nobody thinks it dreams. A thermostat has feedback. It feels the room, checks the number, kicks the heat on or off, feels again — a clean little loop, output curling home. And it does not think, and we will come back to exactly why. For now, only this: feedback is the floor, and the floor is not the house. Hold the thermostat in your mind. The whole distance from it to a mind is the distance this chapter walks.

The second thing is recursion, and here the ground tilts upward and the climb begins. Recursion is feedback folded back on itself — loops running inside loops, the flow no longer only feeling the world but feeling its own feeling, building little inner stand-ins for things and then building stand-ins for those.

Watch it come alive in an animal. A simple creature senses a predator and bolts — one loop, threat in, flight out. Now a cleverer one, that does not only feel the predator but models it, carries a small moving picture of where it is, guesses where it will go, and runs not from where it stands but toward where it will not be. That animal is running a loop about a loop. Somewhere in its traffic it has built a little turning model of the other thing, and it acts on the model instead of the world. Climb once more, to the predator that knows it is being modeled — that hangs back, feints left so the prey breaks right, lays a trap inside the other animal's mind. Three loops deep now, each fold carrying the creature one step further from the raw light on its eyes and one step closer to a thought about the world.

This folding is a candidate engine of much that we call thinking. Memory is a held state standing in for a world gone by. Planning is a state standing in for a world not yet here. Naming, counting, reckoning, dreaming — each recruits layered models rather than only an immediate input-output loop. The thermostat has little of this. It samples the room and switches against a setpoint; its controller need not build a model of its own modeling. That functional thinness is enough for the present comparison. It tells us nothing about whether there is anything it is like to be a thermostat, and this book will not pretend otherwise.

The third thing answers a question the first two raise and cannot settle. If the flow never stops moving, how does it ever hold anything still enough to use?

Call them basins. Picture a ball loose on hilly ground. Most places it will not rest — it rolls. But here and there the land dips into a hollow, and a ball that rolls in settles, and stays, and you can come back tomorrow and find it where it was. That hollow is a basin. It is a place the flow falls into and keeps. And a thinking flow is riddled with them — not dips in any ground, but standing patterns in the traffic that the system drops into and holds, so the same thing coming in finds the same groove already worn for it.

This is what lets a moving thing remember. The flow never pauses, the matter cycles through and is swapped out, and still the pattern holds, because the pattern is a basin the traffic keeps falling back into. It is what lets a thing recognize — meet a face twice and the same groove catches, the same shape settles, and the system knows it has been here. It is where categories come from: a basin wide enough to catch many slightly different things and settle them all into one shape is a concept, a kind, a this-again. Your own immune system meets a plague once and carves a basin for it so deep that the same shape, returning years later, drops straight into the old groove and the defense is standing before you feel a thing. Without basins a flow has loops and folds and no hold — everything washes through and nothing stays, and a mind that keeps nothing is no mind at all. The basins are where the racing water learns to remember its own shape.

The fourth thing is the deepest, and without it the other three go cold and stop. A thinking flow has to be held far from rest, fed across an open edge, never allowed to settle.

Everything known to think requires throughput. The brain you are reading with uses about twenty watts and depends continuously on blood, oxygen, and glucose; interrupt that supply and organized activity fails quickly. A living thing is like a flame in this limited sense: it maintains its organization only while matter and energy pass through it. This establishes a necessary condition for known biological cognition. Whether every possible cognizer must be far from equilibrium in exactly the same thermodynamic sense remains part of the proposal.

This is where the account of thinking joins the open-system philosophy. A functioning biological mind is materially and energetically open because it runs on what passes through it. But openness alone is not cognition, and thermodynamic closure, dynamical coupling, and functional isolation are not the same condition. The proposed four belong together more cautiously: feedback carries consequences home, recursion layers models, basins retain usable states, and throughput keeps the organization available to act.

Now the thermostat gets its answer, and the answer carries the most important thing in the chapter.

It has the first condition, feedback, in a thin form. It has little of the second if its controller builds no model of itself. It has a setpoint and internal states, but not the rich, revisable attractor landscape the criterion means by basins. And it is powered and open, so it would be wrong to call it a system at rest; what it lacks is not energy throughput but depth of organization. Under the proposed functional criterion it therefore scores low on recursion and representational retention. That is an operational comparison, not a verdict about an unseen interior.

Which suggests how mind might arrive, and this is the thing I most want you to carry out of here. It need not increase by the gram, as if matter contained a fixed dose of thought. Organized systems often change regime when interacting capacities cross thresholds, and cognition may be such a regime. Water freezing is an image for that possibility, not a derivation: cognitive functions may emerge at different rates, some gradually and some abruptly, and no universal critical value has yet been measured. The claim worth keeping is narrower. Amount of matter is not enough; organization matters, and a small system may perform a function a much larger unorganized one cannot.

Set one stone down before we move. The four conditions are this book's candidate account of functional cognition — sensing, modeling, holding, answering, the working business of taking in a band of the world and acting back on it. Each term will need a declared measure before the criterion can decide a scientific case. And none of it crosses the firewall into feeling: a complete functional score would still not tell us what, if anything, the scored system experiences.

We have what the turning must be. But I have been quietly cheating the whole chapter, speaking as if these four conditions live inside one thing — a brain, an animal, a creature with an edge you could draw around it. They do not have to. The loops and the folds and the basins and the open fire can run between things just as well as within one — across a swarm, a crowd, a scatter of separate parts that no one would call a single creature. And the moment you let that in, a question opens under your feet: if the four conditions can be met by a handful of separate things, what holds those separate things together tightly enough to count as one? That question has two answers, and they are the next two turns of the book — where the mind comes loose from the body, and how a scatter of parts ever beats as one.