Untilt
A map of cognitive biases born from a course I couldn’t shake.
Starting over
I should begin with a confession, and the subject lets me: to write this piece, I had to start over from scratch.
We’d spent some time in Madrid — the occasion for three great museums, the Prado, the Thyssen and the Reina Sofia. Like a dutiful art columnist, I meant to get one piece out of each. I spent a while at it; inspiration didn’t come. And yet I kept at it: I’d started, invested a few hours and some thought, it would have been a shame not to see it through. So there I was, mid-way through preparing an article on cognitive biases, falling for one of them — the sunk cost fallacy.
The classic example is the bus you wait for. Five minutes. Then ten. Then fifteen. Then thirty. And the longer you wait, the harder you cling to waiting, to justify the time already spent — time that is precisely lost, unrecoverable, hence the name — when cutting your losses and calling a taxi would get you there sooner. It’s probably the one bias known to people who know nothing about cognitive biases. I told the whole story in my newsletter — honesty compels me to add that we still walked the whole Reina Sofia conscientiously, rather than admiring Guernica and leaving.
I eventually closed the Madrid document and wrote this one instead. It’s a fairly good way in, I think, to what Untilt is — and to what led me there.
From agile to neuroleadership
For years I taught agile-method workshops inside my company. The short frame wasn’t enough, so I widened it — a bit of constructivism to explain how people learn, a bit of Palo Alto to explain how people communicate. After enough questions from trainees, I ran into mine: how does a brain that decides actually work?
That’s what led me to Foundations in NeuroLeadership at the NeuroLeadership Institute. A course built for managers, but one that takes the science seriously — some neurobiology, a lot of bias literature, and a single thread: when you understand the brain’s shortcuts, you decide a little better. I didn’t go to fix a specific problem. I went out of curiosity, with the vague sense that an ordinary working day holds a quantity of decisions that deserves better tooling than intuition alone.
The flip side of a very bright coin
The main thing the course left me with isn’t a method. It’s a shift of angle. Biases aren’t bugs; they’re the trade-offs a brain makes to function with too much information and too little time. They’re adaptive — except when the context that built them no longer matches the one you’re using them in.
Of everything I’ve learned in neuroscience, the idea I find most helpful is the brain as a prediction machine: an organ that constantly projects patterns — meaning — onto the reality it perceives. Our biases are just the flip side of a very bright coin, that of pattern recognition. The same mechanism that makes us see a face in a power socket makes us overstate a regularity, mistake a coincidence for a cause, take the improbable for the impossible. (Ask yourself how many people you need in a room for a fifty-fifty chance that two of them share a birthday: the answer feels absurdly low.)
What I like about this idea is the quiet ethics that follows from it. If my brain automatically pins a cause to an effect, then I can learn to be wary of it — and a little more forgiving of other people, whose brains do the same. If what I credit to talent is maybe, statistically, luck, then some modesty is in order. The same coin again: the faculty that fools us is the one that lets us have intuition, and intuition is just a pattern the brain caught almost in spite of us. You wouldn’t want to be without it. You only stand to gain by knowing when it’s playing tricks on you. (I’ve devoted a few newsletter posts to specific biases — the halo effect in the middle of a hiring round, or the peak-end rule that decides what you’ll remember of your holiday.)
What to do with this information?
That’s the question that comes back at the end of each of those posts, and I hold to it: I refuse any definitive answer. Knowing a bias exists doesn’t tell you what to do about it — that depends on each person’s goals, and a human being isn’t, at every instant, a perfectly rational agent anyway. That’s exactly what makes our choices perfectly human in their imperfection.
But there’s a gap that reading doesn’t close. Reading the literature on biases is easy; remembering it at the precise moment it would help — a hire, a budget call, a career move — is something else. Decisions don’t wait for you to crack open a textbook. They arrive on a Tuesday at 2pm, in a meeting room or in front of a screen. Just as I have books piling up in my Tsundoku, I had a pile of biases carefully listed and carefully forgotten by the time they mattered.
If bias science is going to be useful outside the classroom, it has to show up then. That’s what Untilt tries to be. Not a coach, not a personality quiz, not an oracle that tells you what to do — true, in other words, to my refusal to prescribe. A map you consult just before you decide, and that lets you decide.
What’s inside
50 cognitive biases, 37 decision types, in English and French. Two ways in:
- Manual. You describe the situation, pick the decision type, and the app surfaces the biases most likely to apply — each with a description, scientific sources, and a few strategies for stepping back from it.
- AI-guided. You describe the situation in free language; a model call classifies it and proposes the likely biases — which then drop into the same mechanic as the manual path.
No account. No tracking cookies. The full knowledge base is bundled inside the app, with bilingual full-text search and references to the source papers for anyone who wants to check the work. The underlying move is the same as for everything I build: since I can create just about any application with AI, I make one whenever no service fits my need exactly.
P.S. — The map keeps growing. The bias × decision matrix was drafted mostly by Claude, from the scientific literature and the materials of my course; it expands in batches, whenever my Claude Max plan has room — a build detail that, along the way, changed how I work. It’s why the numbers feel both precise and provisional. And if you’ve read this far telling yourself you’d already put in enough time not to stop now — well, you now know what that’s called.