The cognitive bias field guide
Pitfalls in individual thinking: what one mind does to itself, from attention and memory to how it judges other people.
Five catalogues of the ways reasoning fails — inside one mind, inside a group, inside a discussion, inside the rules chosen to prevent it, and inside one field’s beliefs about learning. 924 named entries, each with what stands behind it and what it looks like in a school or a product team.
Why the guides exist
The literature on how thinking fails is large and it is scattered across fields that do not read each other. Cognitive psychology names what goes wrong inside one head. Social and organisational psychology name what happens to a judgement once a group forms around it. Informal logic and rhetoric name the moves that derail an exchange. Philosophy, statistics and engineering each keep their own lists of the rules of thumb meant to prevent all of it, and education keeps a literature on its own myths that its practitioners rarely read. Anyone who wants the whole picture has to hold five literatures at once, and most of us meet the material the other way — a name dropped in a meeting, a post online, a study half-remembered and probably misreported.
These five guides consolidate it into one place and one form. Each takes one of those bodies of work and sets it out as a catalogue: the name as the field gives it, what it is, what stands behind it, and what it looks like where I work. The point is not to memorise nine hundred names. It is to have something to check against — the failures worth knowing about are exactly the ones that feel like clear thinking while they are happening, and a blindspot is not visible from inside the eye that has it. Read with one of these open, a decision gets a second pass it would not otherwise have had.
Every entry uses the name the literature uses, with a definition short enough to hold in mind. Where two names describe the same thing, or one name has drifted to mean two, the entry says so rather than picking a winner quietly.
Each entry carries its provenance: the founding study where there is one, and a plain statement where there is not. An entry resting on nothing more than repetition is marked that way and kept, because knowing a popular name is thin is worth as much as knowing a solid one is solid.
Every entry lands in education. The first volume gives each bias a section on what it means for learning and one on what it means for a learning product; the second says where the fallacy lands in a school; the third supplies a reply that can actually be said in the room; the fourth sets each rule in a classroom, on a platform or before a procurement panel; and the fifth is about education itself, so every claim in it ends with what to say in the meeting where it is made.
Who it is for
There is no reader model in these guides, and that is deliberate. The two series on the work side sort their advice by who is holding it, because a ministry officer, a vendor and a school lead each control a different part of a product. Nothing here works that way. A bias operates the same on a teacher, a procurement officer and a person choosing lunch, and a discussion goes wrong by the same moves whoever is in the room.
What is particular is the examples. I work on educational technology inside a school system, so that is where the illustrations come from: the pilot that looked successful because the keen schools volunteered, the committee that agreed before anyone had disagreed, the argument about screen time that never stated the claim it was defending. The reasoning carries anywhere. The cases are the ones I can vouch for, and I would rather a reader outside education did a small translation than have me invent situations I have not been in. The fifth volume is the exception: education is its subject rather than its setting, and its readers are inside it.
Inside a guide
Each volume sets every one of its entries in the same sections, so a reader who learns the shape once can use all five. The sections differ between volumes because the five kinds of fault need different things said about them.
| Volume | What an entry carries |
|---|---|
| 01 · Cognitive bias | What is it · Why it matters · In practice · How to counter · Relevance to learning · Application to edtech · Where it came from |
| 02 · Collective fallacies | What it is · Why it matters · The documented case · How to recognise it early · What works against it · Where this lands in education |
| 03 · Faulty arguments | The move · Where it breaks · The legitimate version · In the wild · What to say back · Where it came from |
| 04 · Working defaults | The rule · Where it came from · Why it works · Where it misleads · Its counterweight · In education and edtech · How to use it |
| 05 · False certainties | The claim as it is said · What the source actually said · What the evidence shows · What it costs · The kernel · What to say in the room |
What each volume contains
One mind, one group, one exchange, one set of defaults, one field. The five are built on one engine and meant to be read together: a bias is something that happens to you, a collective fallacy is something a structure does to everyone in it, a faulty argument is a move somebody makes, a working default is a rule someone chose on purpose, and a false certainty is a true finding held with more confidence than it earned. Nothing is repeated between them. Each opens as a deck; press R for the reading version, where Appendix A is an A to Z index of every entry, so a name you already have takes one click to find. The fifth, and its index, sit behind a password.
A set of five
5 volumes
Pitfalls in individual thinking: what one mind does to itself, from attention and memory to how it judges other people.
Pitfalls in collective interaction: what happens to reasoning once more than one mind is involved.
Pitfalls in collective decision making: the moves that derail a discussion, and what to say back.
Rules chosen on purpose: the razors, laws and principles people use to think well, and where each one misleads.
Pitfalls in a field’s own beliefs: what education holds true about learning that the evidence does not carry, and why it cannot tell.
How the guides were designed
Each volume tries to hold everything its field has named, rather than the dozen entries that would make a good talk. Over nine hundred entries across five volumes, with the long tail kept.
A shortlist is someone’s taste wearing the clothes of a reference. The value of a catalogue is that the entry you need is in it — and the one you need is rarely the famous one, because the famous ones are the ones you already watch for.
Nobody reads these front to back, and they are too long to. They are built to be looked up rather than read, which is why the index, the search and the glossary matter more here than in anything else on this site.
Each entry is tagged with its provenance — the founding study, a body of observation, or an honest note that there is no single study and the name rests on usage.
I kept collecting these because I wanted to know how many would survive being looked at properly. A good number do not. Replication has been unkind to parts of this literature, and a catalogue that hid that would be doing the same thing it is warning about.
Some well-loved names come out looking thin, which reads as debunking when it is only reporting. The entries stay in either way: a name in wide use is worth knowing even when the evidence under it is not much.
The first three split the material by where the fault sits: in one mind, in a group, or in an exchange between people. The fourth collects the rules people adopt on purpose to prevent those faults, and the fifth turns all four on one field’s beliefs about learning. Nothing is repeated between the volumes.
They are different kinds of thing and they are fixed by different means. A bias is something that happens to you and is countered by design. A collective fallacy is something a structure does to everyone inside it and is countered by changing the structure. A faulty argument is a move somebody makes and is countered by answering it. A working default is a rule someone chose, and is checked against the rule that pulls the other way. A false certainty is a true finding held too firmly, and is countered by naming the distance between the finding and the claim.
A few names could sit in two volumes, and one or two arguably should. Where that happens the entry goes where the remedy is, and the other volume points at it.
Every entry is worked through a case from teaching, schooling or educational technology rather than the finance and politics examples the sources tend to use.
An abstraction does not transfer until it has been seen doing something. Education is the domain I can write about without guessing, and it is the domain the rest of this site is about.
A reader from another field has to translate. That is a real cost, and the alternative — generic examples about coins and urns — costs more, because nobody recognises themselves in them.
The vocabulary
759 entries from four of the volumes as one list, each with the line its entry opens with and a link that opens the guide at it. A name that two guides share appears once for each, because each treats it differently. The fifth guide’s index sits behind the guide’s own password.
Field guides to how thinking goes wrong · Five volumes · 924 entries · September 2026