Welcome to Unit 2. Panic feels like pure sensation — a wave, not a sentence. But there’s always a forecast in it, and it’s specific.

Not I feel terrible. Something more like: this is my heart, this is the one that doesn’t stop, I’m going to collapse here and there’s nobody who’d know what to do.

That’s a prediction. It has a claim, a timeframe and an outcome, and every one of those can be checked against what happens. Which is the point — you cannot test a forecast you have never stated.

Why is it hard to catch? The reading happens faster than language. By the time you’d have words, you’re already in the middle of it, and afterwards it’s compressed to I panicked. So it goes uninspected for years, and stays exactly as convincing as it was the first time.

There’s a second reason it stays hidden, and it’s less comfortable: saying it plainly is embarrassing. Written down in daylight, I will die on the floor of a supermarket looks absurd, and most people soften it before it reaches the page — into I get anxious in shops, which is true, useless, and safely untestable.

The softened version is the problem. It can never be disproved, because it doesn’t predict anything.

Getting it into words does something on its own, before any evidence arrives. A vague dread has no edges. I will faint within two minutes and be found on the floor of the supermarket is a claim about the world — and it can turn out to be wrong in a way that a general sense of doom never can.

So how do you get at it? Take the last attack on your record and work from the sensation: what were you afraid that meant? If the answer is I don’t know, just that it was bad, push once more — bad how? What would have happened if it had kept going?

That last question is usually the one that produces it. The prediction is nearly always about something continuing past the point where it stops.

Here’s a worked example. Sensation: heart thumping in a meeting. First answer: I felt awful. That’s a feeling, not a forecast. Bad how? Like I had to get out. Closer, still a feeling. And if you hadn’t got out? I’d have gone bright red, lost the thread, and everyone would have seen something was wrong with me.

There it is. Three checkable claims — the colour, the losing the thread, and what other people noticed — where a minute earlier there was only awful. The last one is testable by simply asking someone afterwards, which is a thing almost nobody does.

One thing not to do: don’t argue with it as you write it. The temptation is to add but I know that’s irrational — and that sentence, which people have often said to themselves for years, has never once helped, because the fear was never being maintained by a shortage of arguments.

Write the prediction as it actually arrives, in its own words, at full strength. It’s about to be tested. It doesn’t need softening first.