Say one of yours out loud.
*I can’t do maths. I’m not creative. I’m terrible with names. I’m not a natural leader. I don’t have the discipline for that. I’ve never been good with money.*
Now answer one question about it: **when did you last actually test it?**
Not “when did it last feel true.” Not “when did I last avoid a situation where it might apply.” When did you last make a deliberate, effortful, repeated attempt at the thing, under conditions where you could have succeeded?
For most people, with most of their I-can’ts, the honest answer is somewhere between fifteen and forty years ago.
You are walking around with an inventory of your own limits, most of which has not been audited since the last century, and you’ve been making decisions from it every year since.
Here’s where it came from.
## What an Installed Limit Is
**An installed limit is a belief about personal incapacity that was accepted rather than tested. It typically originates in a single event or a short period, is delivered or inferred with authority, and is subsequently maintained by avoidance rather than by evidence. Because the belief causes the person to stop attempting the activity, no disconfirming evidence is ever generated, and the belief remains unchallenged indefinitely while feeling increasingly well-established. Installed limits are distinguishable from genuine limits by their history: a genuine limit has recent, repeated, effortful attempts behind it, while an installed limit has an old verdict and a long absence of attempts. They present in the grammar of fact — “I can’t,” “I’m not a” — rather than in the grammar of experience, which is one of the more reliable ways to spot them.**
Three ways they get in. Almost every one you have came through one of them.
## Route One: The Verdict
Somebody with standing said something about you, once, and it stuck.
A teacher. A parent. A coach. Someone whose opinion, at that moment, counted as a description of reality rather than as an opinion.
*You’re not a maths person. You’ve got no ear. You’re the practical one — your sister’s the clever one.*
Four things have to be present for a verdict to install, and when all four are there it takes on the first attempt:
**Authority.** The person’s assessment counted. You had no standing to dispute it.
**Emotion.** Something was at stake — embarrassment, disappointment, an audience. Emotional load is what makes single-trial learning stick.
**Specificity.** It named a category, not a moment. “That answer was wrong” doesn’t install. “You’re not a maths person” does, because it converts an event into a property.
**No counter-evidence afterwards.** And this is the one that seals it, because the verdict itself produces the conditions for its own survival — you stopped putting your hand up, so nothing ever contradicted it.
Most people can name the moment if they sit with it for ninety seconds. There’s often a room in it. Sometimes a specific chair.
## Route Two: The Generalisation
No one said anything. You did it to yourself, from a sample of one.
You tried something, it went badly, and you drew a conclusion about your nature rather than about the attempt.
This one is more common than the verdict and considerably harder to see, because there’s no villain in it and nothing to be angry about.
The move is always the same and it happens in about a second: an event gets converted into an identity.
*That presentation went badly* → *I’m not good at public speaking.*
*This business failed* → *I’m not an entrepreneur.*
*She left* → *I’m not someone people stay with.*
Notice the grammar shift. The left-hand side is a report. The right-hand side is a permanent property with no expiry date and no conditions attached.
And once you’re on the right-hand side, you stop generating data. That’s the mechanism. Not the belief — the *avoidance the belief licenses*.
## Route Three: The Inherited Ceiling
The third route is the quietest and the most consequential, and it was never stated out loud.
Nobody in your family did that. So it wasn’t a thing people did.
Not forbidden. Just absent — and absence is a far more efficient teacher than prohibition, because prohibition at least tells you the thing exists.
People from families where nobody went to university don’t typically believe they’re forbidden from university. They just have a sense, with no traceable source, that it’s a thing other people do. People from families where nobody ran a business don’t think business is impossible. It’s simply not in the set of available moves.
You can spot these because they have no origin story. The verdict has a room. The generalisation has an event. This one has nothing — it feels like a fact about the world, because you absorbed it before you had the equipment to notice you were absorbing anything.
## The Grammar Test
Here’s the most useful diagnostic on the page, and you can run it in the next thirty seconds.
Take one of your I-can’ts and look at the grammar.
**Genuine limits are stated in the language of experience.** “I’ve tried that four times and I get to about here.” “I trained for two years and topped out.” There’s a history in the sentence. Attempts, dates, conditions.
**Installed limits are stated in the language of fact.** “I’m not a numbers person.” “I can’t sing.” Flat, categorical, present tense, no history at all — because there isn’t one.
Then apply the three-word replacement. Take the sentence and swap “can’t” for one of these:
**Won’t** — you’ve decided against it. That’s legitimate, and it’s a choice rather than a limit.
**Haven’t** — you’ve never seriously tried. Also legitimate, and completely different from can’t.
**Can’t** — you’ve tried repeatedly, recently, with real effort, and hit a ceiling.
Run your list through that.
Most people find that somewhere north of eighty percent of their can’ts are haven’ts wearing a borrowed uniform. And the difference matters enormously, because *haven’t* has an obvious next move and *can’t* has none.
That’s the whole reason the substitution exists. Can’t closes a question. Haven’t opens one.
## Why This Isn’t Just Positive Thinking
I want to be careful here, because there’s a version of this argument that’s rubbish and it’s everywhere.
The rubbish version says all your limits are illusions and you can do anything.
You can’t. Real limits exist. Some are physical, some are about time, some are about the fact that starting a demanding new career at fifty-eight has genuine constraints attached that no amount of belief will move.
I’m not making that argument. I’m making a much narrower one, and the narrowness is what makes it useful:
**You don’t currently know which of your limits are real, because you stopped collecting evidence a long time ago.**
That’s it. Not “your limits are fake.” *Your data is stale.*
Some of your can’ts will survive the audit. Genuinely. And that’s a good outcome, because a tested limit is a useful piece of information you can plan around, while an untested one is just a fog you steer to avoid.
There’s a related distinction worth having: the difference between something being heavy and something being impossible. Most people file heavy under impossible, permanently, because the two feel similar at the moment of contact.
## Where This Connects
An installed limit is the same class of thing as an installed reaction. Both were written early, both were reasonable at the time, both run without consultation.
The difference is where they show up. The reaction fires in the moment — you go quiet in the meeting. The limit fires *before* the moment, by removing options from consideration so quietly that you never experience a decision at all. There’s a full treatment of the reaction side in the seven-year-old making your decisions.
And both are what make a person forecastable.
A man in 1548 had his entire life predicted accurately for twenty years, and the reason wasn’t mysticism. It was that he had a fixed set of responses and a fixed set of things he’d ruled out, and anyone who could read both could run him forward. That’s the story, and your I-can’t list is a substantial part of what a stranger would be reading.
## What This Is Not
**It is not blame directed at whoever said it.** The teacher who told you that in 1994 was managing thirty children and made an offhand remark they forgot within the hour. The install is on your side of the transaction, and it’s the only side you can do anything about.
**It is not an instruction to attempt everything.** You have a finite amount of time and most of your untested limits genuinely don’t matter. Whether you can sing is not load-bearing. What matters is the two or three that are currently removing options you’d want.
**It is not a claim that testing will always go well.** Some of these you’ll test and confirm. That’s a real result and a useful one. The point isn’t that you’ll succeed — it’s that you’ll know, and knowing beats avoiding by a large margin.
## The Cost
Here’s the price, and it’s specific.
An untested limit is doing a job for you. It’s protecting you from finding out.
As long as it stays untested, the possibility is intact. You could have been good at that, if things had gone differently, if you’d started younger, if someone had encouraged you. That’s a genuinely comfortable thing to own, and it costs nothing.
Test it and you replace a comfortable possibility with a definite answer, and there’s a real chance the answer is one you didn’t want.
That’s the trade, and it’s the actual reason people don’t run this audit. Not laziness. They’ve already worked out — somewhere below the level of words — that the current arrangement is more pleasant than either outcome.
And the second cost: if it turns out you could have, all along, then the intervening years were available and you didn’t use them. That’s a genuinely uncomfortable thing to sit with, and I’d rather flag it now than have you meet it by surprise.
## Common Questions
**Where do limiting beliefs actually come from?**
Three routes, mostly. A verdict delivered by someone with authority under emotional conditions; a generalisation you made yourself from a single bad experience; or a ceiling absorbed from an environment where the thing simply wasn’t done. The third is the hardest to see because it has no origin event attached.
**How do you tell a real limit from an installed one?**
By its history. A real limit has recent, repeated, effortful attempts behind it. An installed one has an old verdict and a long absence of attempts. If you can’t name when you last genuinely tried, you’re not looking at a limit — you’re looking at an assumption.
**Why do these beliefs feel so certain?**
Because they’ve never been contradicted, and the absence of contradiction registers as confirmation. The belief causes the avoidance, and the avoidance guarantees no disconfirming evidence is ever generated. It’s a closed loop that gets more convincing with age while acquiring no new information.
**Can you just decide to stop believing it?**
No, and attempting that is why affirmation-based approaches mostly fail. Beliefs formed from evidence only reliably shift on evidence. What changes an installed limit is a small, real, undeniable counter-instance — something you actually did that a person holding the belief couldn’t have.
**What’s the difference between can’t and haven’t?**
“Can’t” asserts a tested ceiling. “Haven’t” reports an absence of attempts. They’re completely different claims and almost everyone uses the first when the second is accurate, which matters because “haven’t” has an obvious next move and “can’t” has none.
**Does this mean I can do anything?**
No, and that version of the argument is worth rejecting firmly. Real constraints exist. The narrow claim is that your information about your own limits is decades out of date, so you don’t currently know which ones are real.
## Where to Start
Ten minutes, one page.
Write down every “I can’t” and “I’m not a” you can catch yourself saying. Aim for ten. They come faster than you’d expect once you start listening for the grammar.
Next to each one, write a single date: **the last time you seriously attempted it.**
Not thought about it. Not avoided a situation involving it. Attempted it, with effort, more than once.
Most of the dates will be old. A good number will be blank.
Then circle the one blank that would change something if it were wrong. Just one. Not the most interesting — the one that’s currently removing an option you’d actually want.
That’s your test case, and testing it is next week’s problem, not tonight’s.
Tonight’s problem is just seeing the dates.
I’ve spent twenty-five years working on how ordinary people develop capabilities that look impossible from the outside, and the single most common thing standing between someone and a capability isn’t difficulty, aptitude, or time.
It’s a sentence somebody said to them once, that they never checked.