Reviewed byIn short
A cognitive bias is a systematic pattern in how people judge, remember and decide that departs from what the evidence or the arithmetic would support. The patterns are not random errors: each one is a shortcut that solves a real problem most of the time and fails in a predictable way the rest of the time.
Common cognitive biases with examples
The fifteen biases below are the ones people search for and meet most often. A heuristic is the shortcut itself, a fast rule that usually works; a bias is the predictable error the shortcut produces when it does not. Every row links to a page with a short demo, the original study and what helps. The 187 further down is the number of entries in this Codex, aliases and related effects included, not a settled count of distinct biases.
| Bias | What it does | Everyday example |
|---|---|---|
| Confirmation bias | Seeks out and remembers evidence that fits what you already believe, and skips the rest. | Searching "does this supplement work" and reading only the pages that say yes. |
| Anchoring | A number seen first pulls later estimates toward it, even when the number is arbitrary. | A "was $120" tag makes $80 feel like a bargain. |
| Availability heuristic | Judges how common or likely something is by how easily examples come to mind. | Fearing a flight more than the drive to the airport after a week of crash headlines. |
| Sunk cost fallacy | Keeps a course of action going because of what has already been spent on it. | Sitting through a bad film because the ticket was paid for. |
| Dunning-Kruger effect | People with the least skill in a domain rate themselves well above their actual level. | A first-month chess player rating their game as above average. |
| Halo effect | One favorable impression, such as warmth or good looks, spreads to unrelated qualities. | A confident presenter is assumed to be right about the numbers. |
| Hindsight bias | Once an outcome is known, it looks far more predictable than it was beforehand. | "I knew that stock would fall" after it fell. |
| Fundamental attribution error | Explains other people's behavior by their character while overlooking the situation. | The driver who cut you off is a jerk, not someone racing to a hospital. |
| Survivorship bias | Draws conclusions from the cases that made it through and forgets the ones that did not. | Copying the habits of famous dropouts without counting the dropouts who failed. |
| Loss aversion | A loss weighs more in a decision than a gain of the same size. | Turning down a coin flip that pays $150 on heads and costs $100 on tails. |
| Bandwagon effect | Adopts a belief, product or candidate partly because many others already have. | Choosing the restaurant with the queue outside the door. |
| Framing effect | Chooses differently when the same outcome is worded as a gain rather than a loss. | "90 percent survive" sounds better than "10 percent die" for the same operation. |
| Negativity bias | Negative information weighs more than positive information of the same size. | One critical comment outlasting ten compliments. |
| Self-serving bias | Credits successes to yourself and puts failures down to circumstances or other people. | A good grade means you are smart; a bad one means the test was unfair. |
| Optimism bias | Rates your own chances of good events above average and of bad ones below. | Expecting your renovation to come in on budget when most do not. |
The four problems the mind is solving
The codex, drawn up by Buster Benson in 2016 and illustrated by John Manoogian III, takes the long and overlapping list of named biases and sorts it by the problem each one addresses. Four problems account for all of them.
- Too much information. There is more arriving than anyone can take in, so the mind filters, and the filters have preferences: for what has changed, for what is repeated, for what confirms what it already believes, for what is bizarre or striking, and for flaws in other people rather than in itself.
- Not enough meaning. The information that gets through is thin, so the mind fills the gaps with stories, stereotypes, generalizations from one case, projections of its current mood onto the past and the future, and a flattering picture of itself and the people it knows. This is the largest quadrant, with 81 entries.
- The need to act fast. Decisions have to be made before the evidence is in, so the mind favors the option that feels confident, simple, complete and reversible, and it favors finishing what it has already paid for.
- What should be remembered. Memory has to be compact, so the mind keeps key elements and discards specifics, edits memories after the fact, and stores events differently depending on how they were experienced.
Every shortcut works more often than it fails, which is why they survive. The pages on this site exist for the times they fail: a single demo per bias, built to let you land on the bias yourself, and a plain account of the study that first documented it, where it shows up, how to counter it, and which neighboring biases it is not.
How to use this codex
The wheel below is the codex drawn as a ring: the four quadrants on the inside, the subcategories in the middle ring, and the 187 named biases around the rim. Hover or tab across a leaf to read its name; the leaves that are already written are links. The list under the wheel has every entry with a one-line definition and a filter box, and it is the faster route on a phone.
Each bias page follows the same shape. It opens with the demo, because a bias that you have just experienced is easier to recognize later than one you have only read about. The demo ends on a result that says in plain words which way you went, shows the figure from the original study, and, once enough people have taken it, the share of visitors who went each way. Nothing you enter leaves your browser except that one outcome label. Below the demo come the definition, how the bias works, the original study, a table of everyday examples, what helps, and the boundary with the nearest biases.
The pages are educational. Noticing a bias in yourself is normal, and the demos are built so that most people land on it; that is what the research predicts. None of this is a diagnosis, and none of it is a substitute for talking to a professional if something here touches a difficulty you are living with.