- Heuristics are mental shortcuts: usually helpful, sometimes systematically wrong.
- Availability: easy to recall feels common. Vivid, recent and much-reported events seem likelier than they are.
- Representativeness: typical-looking feels likely, so we forget how common each kind of case is.
- The conjunction fallacy (Linda) and the gambler's fallacy both come from judging by resemblance.
- The fix is a habit: ask how often it really happens, and how many of each there are.
- heuristic
- a mental shortcut that gives a quick answer without working everything out
- availability
- judging how likely something is by how easily examples come to mind
- representativeness
- judging how likely something is by how much it resembles a typical case
- conjunction fallacy
- judging two things together as more likely than one of them alone
- gambler's fallacy
- believing a random outcome is "due" because it hasn't happened for a while
Watch out: Thinking you are immune once you know the names. Knowing about availability does not stop a vivid story from feeling convincing. The fix is a habit, not knowledge: when a judgement feels obvious, ask "How often does this really happen?" and "How many of each are there to begin with?", and then look it up.
- An anchor is a starting number that pulls later estimates towards it, even when it is plainly random.
- In the 1974 wheel study, 10 led to a median estimate of 25%; 65 led to 45%.
- Anchors work on prices, negotiations, forecasts and experts, mostly without being noticed.
- Defences: make your own estimate first, consider the opposite, and make the first offer when you can.
- Anchoring has replicated well, but its size varies with how much people know.
- anchor
- a starting number that pulls later estimates towards it
- anchoring
- the tendency for estimates to stay too close to a starting number
- adjustment
- moving away from the anchor, which people usually do too little
- consider the opposite
- listing reasons the true value could be far from the anchor, which weakens its pull
Watch out: Thinking an anchor only works if you believe it. The wheel-of-fortune number was obviously random, and it still moved people's answers. "I know that price is silly, so it won't affect me" is exactly when it does.
- Confirmation bias: we search for, interpret and remember evidence that fits what we believe.
- In Wason's 2-4-6 puzzle, people tested only what fitted their guess, so they never found their mistake.
- Mixed evidence can leave both sides feeling more certain (Lord, Ross and Lepper, 1979).
- Defences: decide what would change your mind, look for that evidence on purpose, and search with neutral words.
- Testing cases that fit is fine; never testing cases that could prove you wrong is the problem.
- confirmation bias
- seeking, interpreting and remembering evidence in ways that support what you already believe
- disconfirming evidence
- evidence that could show a belief is wrong
- belief polarisation
- when people on both sides see the same mixed evidence and become more certain
- steelman
- the strongest version of the view you disagree with
Watch out: Searching in a way that can only agree. Typing "proof that coffee is bad for you" finds proof that coffee is bad for you. Search with neutral words ("coffee health effects research review"), or search for both sides, and notice which results you are tempted to skip.
- Relative risk ("doubles") says how much a risk changes; absolute risk ("1 in 7,000 to 2 in 7,000") says how big it is. Always ask: out of how many?
- The base rate, how common something is to begin with, matters as much as the evidence.
- With a rare condition, most positive screening results can be false alarms.
- Turn percentages into counts out of 1,000 (natural frequencies) to work it out.
- "The test finds 90% of cases" is not "90% of positives are cases".
- relative risk
- how much a risk changes, as a ratio ("doubles", "halves")
- absolute risk
- how big a risk is, as a count out of a group ("2 in 10,000")
- base rate
- how common something is in the group to begin with
- base-rate neglect
- judging likelihood from the evidence alone and ignoring how common the thing is
- natural frequencies
- counts of people out of a round number, used instead of percentages
- false positive
- a test result that says yes when the true answer is no
Watch out: Confusing "the test finds 90% of people who have the disease" with "90% of people who test positive have the disease". They sound alike but are different questions. The first is about the test; the second, the one you care about, depends heavily on the base rate.
- Memory is rebuilt each time you recall it, from fragments filled in with expectations.
- Later information, such as leading questions and other people's accounts, can change the memory (misinformation effect).
- One verb ("smashed" against "contacted") changed speed estimates by about 9 mph, and later "memories" of broken glass.
- False memories of whole events can be planted in some people.
- Protect important memories: write them down early, in your own words, before comparing stories.
- reconstructive memory
- memory rebuilt from fragments each time it is recalled, rather than replayed
- misinformation effect
- later information changing your memory of an earlier event
- leading question
- a question worded to suggest an answer
- false memory
- a memory of an event, or detail, that did not happen
Watch out: Treating confidence as proof. A memory can feel sharp and certain and still be wrong, and confidence tends to grow each time a story is retold. "I'm absolutely sure" is not the same as "it happened that way".
- Framing: the same facts, described as gains or as losses, lead to different choices.
- In the disease problem, 72% took the sure option framed as lives saved; only 22% took the same option framed as deaths.
- Loss aversion: losses feel roughly twice as strong as equal gains, though the size is debated.
- The endowment effect: owning something makes giving it up feel like a loss.
- Before deciding, restate the choice in the other frame and compare in the same units.
- framing effect
- the same facts described differently leading to different choices
- gain frame
- describing outcomes as what is saved or won
- loss frame
- describing outcomes as what is lost or given up
- loss aversion
- losses feeling bigger than equal-sized gains
- endowment effect
- valuing something more once you own it
- prospect theory
- Kahneman and Tversky's account of choice, in which outcomes are judged as gains or losses from where you stand
Watch out: Deciding from the first frame you hear. Whoever describes the choice picks the frame, and sellers pick the one that helps them. Before deciding, say the choice back to yourself the other way round: "What do I gain? What do I lose?" If your answer flips, the frame was doing the work.
- Knowing about biases isn't enough; a short routine catches them.
- The outside view (how did similar cases go?) corrects the planning fallacy: students predicted about 34 days for their thesis and took about 56.
- A pre-mortem imagines the decision has failed and lists why, before you commit.
- The checklist: three options; outside view; where the numbers came from; what would change your mind; both frames; pre-mortem.
- Write the decision and your reasons down, and review them later.
- inside view
- planning from your own case and plan, imagining it going well
- outside view
- starting from how similar cases went for others
- reference class
- the group of similar past cases you compare with
- planning fallacy
- underestimating how long and how much a plan will take, even knowing past plans overran
- pre-mortem
- imagining a decision has failed and listing why, before you commit
- decision journal
- a written record of a decision, your reasons and what you expect, to review later
Watch out: Using the checklist only to justify a decision you have already made. If every answer conveniently supports your first choice, go back to question 4 and look harder for evidence against it, or ask someone who disagrees to answer the questions with you.