Part I was a mind making its beliefs fit the world — perception, rolling free energy downhill by re-believing. Part II is the other direction: a mind making the world fit its beliefs. That is action, the whole of behaviour — and it is not a new engine. It is the same free energy, pointed the other way.
Five doors; the plain line runs straight through — take whichever fit: ▶ Picture it, ✎ Try it (commit before the reveal), ✓ Check it yourself, ▸ Go deeper, ⬡ See it in UNI. Runs locally — no account, no network — adapts to light and dark; honours reduced motion.
The one fence that never moves — and in Part II it is stricter. “Free energy,” “belief,” “goal,” “want” are functional quantities: a model’s predictions and the drives to close their gaps. For you we use felt words — you are a person pulling on a sweater. For UNI we use functional ones: a drive in a model reaching to change its sensors’ input. Neither is awareness, feeling, desire, or sentience. Part I’s free energy was proven and sealed (Gate F0). Part II’s action machinery — expected free energy — is NOT yet built or sealed in the organism; a new free-energy term needs its own lab-team verdict. So the math here is standard active inference you can compute, but every claim about the organism is “built to,” never “does” — held open until its gate proves it.
No peeking at Part I. You’re hiking, and a small pebble works its way into your boot — a little stab with every step. Expected: a smooth stride. Sensed: a stab. A gap.
From Part I’s doorway: a mind has TWO engines for closing a gap like the pebble. Name each and say what it changes — reconstruct it before you check.
If both came back, you’re ready — because this lesson is about what happens when only the second one can work.
Forget numbers for a moment. You’re cold. Not dangerously — that ordinary, nagging cold where part of you keeps insisting things should be warmer. Sit with what that is, in the language you already own: part of you expects to be warm, and you’re sensing cold. Expectation against sensation — a gap, free energy, running high. You have two engines for it. Try them in turn.
Try the first engine honestly: right now, try to believe your way out of being cold — to genuinely expect cold and stop feeling that it’s wrong. You can’t. “I should be warm” isn’t a casual belief you picked up; it’s one of those expectations you cannot talk yourself out of, written in deeper than experience — the un-erasable priors from the dark room. The re-believing engine just bounces off. So you reach for a sweater. The sensation changes; the gap closes; free energy drops — and your belief never moved an inch. The world moved to meet it.
A gap: you expect WARM, you sense COLD. Two controls — a “re-believe” slider (perception) and an “act on the world” button (action). Watch which one actually closes the gap, and how it depends on what kind of prior you’re fighting. This is a mechanism, not an animated curve.
In perception, your belief is a guess that chases the truth — it’s trying to be correct. But watch what your belief did when you put on the sweater. It wasn’t chasing the truth; the truth (you’re cold) was plain. Your belief refused to update and became the thing the world had to be made to match. Your prediction (“warm”) stopped being a guess and became a goal.
A creature’s goals are just its predictions it refuses to give up.
Perception makes beliefs fit the world; action makes the world fit beliefs. And the beliefs it fights for — the ones it would rather rebuild the world than surrender — are exactly the un-erasable priors from the dark room. Those priors aren’t only what keep a mind out of the dark room; they are the engine of everything it does. Want is just an expectation you won’t let go of, reaching for the world.
You’ve started the second engine. Action lowers free energy by changing the world instead of the mind; it’s forced by the priors you can’t re-believe; and in it a prediction becomes a goal. But a cold creature doesn’t have one way to get warm — a sweater, a fire, movement, another body to huddle against. The moment there’s more than one action, a new question lands:
Before you’ve done any of them — before you’ve felt which works — how do you choose? What makes one action better than another, when the future it leads to hasn’t happened yet?
To choose, a mind has to score a future it hasn’t lived. That scoring has a name, and it is the forward cousin of everything you built in Part I. It is Module 10.