The weekly Executive Briefing: insights, books, tools, and media opportunities worth your attention — curated and distilled into 5 minutes.
You are about to quote this book in a meeting, and some of what gets repeated from it has aged badly. This thinking fast and slow summary gives you the core first. Daniel Kahneman argues that a fast, automatic System 1 makes most of your judgments, while a slow, effortful System 2, which tends to be lazy, lets most of them through.
Below, the book runs part by part, with the ideas that matter most for decisions at work: anchoring, loss aversion, framing and the planning fallacy. Then we test it. We checked 16 of the book’s best-known studies against later replication work, because Kahneman himself said one chapter leaned on evidence that was too thin.
What We Found
Of 16 headline studies in Thinking, Fast and Slow that we checked against later replication work on 15 November 2026, four held up, five failed direct or multi-lab replications, five are disputed, and two came back weaker or reversed.
The four that held come from the decision research Kahneman is known for, much of it done with Amos Tversky: anchoring, framing, the choice patterns behind prospect theory, and the sunk-cost effect. Anchoring came back at least twice as strong as framing, and more than four times as strong as sunk costs, in the project that retested all three.
The five that failed come from two early chapters, on priming and self-control. They include the pen held between the teeth and the idea that willpower drains like a battery. Five more, among them money priming and the hard-to-read font, are disputed: the evidence against them is pooled or secondhand, not a direct rerun.
Thinking Fast and Slow Summary: The Book in Five Parts
Thinking, Fast and Slow was published in 2011 by Farrar, Straus and Giroux. The hardcover runs to 498 pages, and Wikipedia’s entry says it had sold over one million copies by 2012. Kahneman took a half share of the 2002 economics prize, which the Nobel committee awarded for bringing psychological research into economic science.
Part one sets out the two systems. Part two covers the mental shortcuts Kahneman and Tversky catalogued, among them anchoring, availability and the habit of swapping a hard question for an easier one. Part three is about overconfidence, and it holds the planning fallacy.
Part four is the economics: prospect theory, loss aversion, framing and sunk costs. Part five splits you into an experiencing self that lives each moment and a remembering self that keeps score afterwards. The remembering self rates an episode by its most intense point and by how it ended, and pays little attention to how long it lasted.
What Are System 1 and System 2?
System 1 is the fast, automatic and intuitive mode of thinking that answers 2 + 2 without effort. System 2 is the slow, deliberate mode you need for 17 × 24. Kahneman’s claim is that System 2 takes effort and tends to be lazy, so most of the time you run on System 1’s answers.
One of the book’s best-known demonstrations is the bat-and-ball puzzle. A bat and a ball cost $1.10 together, and the bat costs a dollar more than the ball. The answer that arrives first is ten cents, and more than half of the students tested at Harvard, MIT and Princeton gave it. The right answer is five. The fast answer feels finished, so System 2 rarely checks it.
Kahneman’s name for the underlying habit is “what you see is all there is”. The mind builds a confident story from whatever information is in front of it and does not ask what is missing. That is the same failure our Four Agreements summary for working life meets in the rule about assumptions.
A second counter is older than psychology. Leonardo worked his ideas out in writing, and our count of Leonardo da Vinci’s notebooks pulls out five of his habits a busy person can borrow. Writing a judgment down before you learn the outcome is the cheapest System 2 tool there is.

What Is Anchoring?
Anchoring is the pull that a number you have just seen exerts on your next estimate, even when the number is irrelevant. In the book’s example, people asked whether Gandhi was older than 114 when he died gave higher estimates of his age than people asked whether he was older or younger than 35.
It is the sturdiest finding in our count. In 2014 the first Many Labs project reran 13 classic effects across 36 samples and 6,344 participants. Its four anchoring questions, taken from a 1995 paper by Jacowitz and Kahneman, produced weighted effect sizes between 1.17 and 2.42, according to the project’s results table.
For anyone who sets prices, budgets or pay, the lesson is that the first number on the table does real work. Our Never Split the Difference summary covers which of Chris Voss’s techniques transfer to salary and client talks.
Anchors also travel in public. When we traced Shaq’s net worth across 25 pages, the pages that named a source for the $500 million most of them give all pointed to one site. For Serena Williams’s net worth, Forbes and Celebrity Net Worth sit $100 million apart, and which one you saw first shapes what sounds reasonable. Before you see anyone else’s number, write down your own.
What Is Loss Aversion?
Loss aversion is the finding that a loss weighs more heavily on people than a gain of the same size. It sits at the centre of prospect theory, Kahneman and Tversky’s 1979 account of choice under risk, which says people judge an outcome as a gain or a loss measured from a reference point.
Framing follows from it. The same facts land differently when they are stated as losses. In the book’s medical example, more people accept surgery described as having a 90 percent survival rate than surgery described as having a 10 percent mortality rate.
Both ideas have survived retesting. In 2020 an international team, Ruggeri and colleagues, repeated the original prospect theory questions with 4,098 people in 19 countries. Their paper in Nature Human Behaviour reports that the results replicated for 94 percent of items, with some weakening. Many Labs reran Tversky and Kahneman’s 1981 framing problem and found a weighted effect of 0.60.
A second project, Many Labs 2, found framing again in 2018, though at about half the original strength.
The sunk-cost effect belongs to the same family. Money already spent keeps people in projects they would never start today, because stopping turns a paper loss into a real one. Many Labs found it too, with a weighted effect of 0.27, well under framing’s 0.60, and participants were only marginally more likely to go to the game when they had paid for the ticket.
The working test is to ask whether you would buy this position today at today’s price. If the answer is no, the only thing keeping you in is the wish to avoid booking the loss. Our piece on Charlie Munger’s mental models covers the five an operator uses, incentives and inversion among them.
What Is the Planning Fallacy?
The planning fallacy is the habit of forecasting a project from its best-case scenario, so that time and cost come in far above the plan. Kahneman and Tversky coined the term. The book’s example is American kitchen renovations in 2002, where homeowners expected to pay $18,658 on average and ended up paying $38,769.
The classic test came in 1994, when Roger Buehler and colleagues asked 37 psychology students how long their senior theses would take. The average estimate was 33.9 days. The average result was 55.5 days, and only about 30 percent finished in the time they predicted, according to Wikipedia’s summary of the study.
Kahneman’s fix is the outside view. Set your own plan aside and ask how long projects like this one took other people. The formal version is called reference class forecasting. For a launch or a hire, the informal version is enough: find three comparable cases, take their real durations, and start from the middle one.

Which Thinking, Fast and Slow Studies Failed Replication?
Five of the 16 studies we checked failed in a direct replication or a registered multi-lab test: four from the chapter on priming and one from the chapter on self-control. Five more are disputed, because the evidence against them is a pooled analysis, a secondary report or a reanalysis, not a direct rerun. Three of the four that held are Kahneman’s own work, alone or with Tversky.
| Study in the book | Later check | What it found | Verdict |
|---|---|---|---|
| Age-related words slow walking (Bargh, 1996) | Doyen and colleagues, 2012 | No priming with 120 people and automated timing | Failed |
| Pen in the teeth makes cartoons funnier (Strack, 1988) | Wagenmakers and colleagues, 2016 | 17 labs; difference of 0.03 points against 0.82 originally | Failed |
| Reminders of money make people self-reliant (Vohs, 2006) | Lodder and colleagues, 2019 | Pooled test of money priming in general; 47 preregistered experiments, effect 0.01 | Disputed |
| Guilt and the urge to wash (Zhong and Liljenquist, 2006) | Earp and colleagues, 2014 | No effect in any of a series of direct replications | Failed |
| Reminders of death shift attitudes (terror management research, Greenberg) | Many Labs 4 | 21 labs, 2,220 people; no effect under any conditions | Failed |
| Watching-eye posters raise payments for office milk (Bateson, 2006) | Northover and colleagues, two meta-analyses | Pooled tests of eye images in general, 26 and 27 experiments; no effect on giving | Disputed |
| Self-control drains with use (Baumeister) | Hagger and colleagues, 2016; Vohs and colleagues, 2021 | 23 labs, 2,141 people, interval included zero; then 36 labs, 3,531 people, effect 0.06 | Failed |
| Sugar restores self-control (Baumeister’s group) | Lange and Kurzban, 2014, citing two replications | Secondary report: the replications did not find the original effect | Disputed |
| A hard-to-read font improves puzzle answers (Alter, 2007) | Meyer and colleagues, 2015 | Original pooled with 16 replication attempts; no effect | Disputed |
| Parole judges are harsher before a meal (Danziger, 2011) | Glöckner, 2016 | Simulations show the size is overestimated; case order was not random | Disputed |
| Children who wait for a treat do better later (Mischel) | Watts and colleagues, 2018 | Link half the original size; cut by two thirds with family controls | Weaker |
| The hot hand in basketball is an illusion (Gilovich, Vallone and Tversky, 1985) | Miller and Sanjurjo, 2018 | A bias in the original measure; corrected, the conclusion flips | Reversed |
| Anchoring (Many Labs items from Jacowitz and Kahneman, 1995) | Many Labs, 2014 | Weighted effects of 1.17 to 2.42 across four questions | Held |
| Gain and loss framing (Tversky and Kahneman, 1981) | Many Labs, 2014; Many Labs 2, 2018 | Effect 0.60; found again at about half the original strength | Held |
| Prospect theory choices (Kahneman and Tversky, 1979) | Ruggeri and colleagues, 2020 | 19 countries, 4,098 people; 94 percent of items replicated | Held |
| Sunk costs (the paid-for ticket) | Many Labs, 2014 | Weighted effect 0.27 | Held |
The priming chapter took the heaviest damage. Its most famous story, from a 1996 paper by Bargh, Chen and Burrows, is that students exposed to words linked to old age then walked down the corridor more slowly. In 2012 Doyen and colleagues repeated it with a larger sample and automated timing and found no priming. A slower walk appeared in their second experiment only when the experimenters believed it would.
The self-control chapter fared little better. Two large preregistered projects, five years apart and covering 59 laboratories between them, could not find the drain on willpower that the chapter describes. Our article on decision fatigue covers what that leaves standing, including the dispute over the parole judges.
The hot hand is the odd one out, because Tversky co-wrote the original study. The book treats a basketball player’s streak as a pattern seen in randomness. In 2018 Joshua Miller and Adam Sanjurjo showed that the measure used in 1985 was biased. With the bias corrected, they report significant evidence of streak shooting.
Our Outliers summary checked 14 of Malcolm Gladwell’s claims against the studies and six did not hold up, and our piece on Adam Grant’s Give and Take traces 15 of that book’s studies to their source papers.
What Kahneman Said About the Priming Chapter
Kahneman saw the trouble early. The year after the book came out, he sent an open email to researchers who study social priming. Nature’s report of 3 October 2012, republished by Scientific American, says he warned of a “train wreck looming” for the field and proposed that labs replicate each other’s results in a chain.
In February 2017 the psychologist Ulrich Schimmack and two colleagues published a statistical audit of chapter four. They counted 12 cited articles containing 31 studies and scored the set at 14 on a 100-point index of replicability.
Kahneman replied in the comments on 14 February 2017. He agreed that he had “placed too much faith in underpowered studies”, and said the size of behavioural priming effects could not be what his chapter suggested.
How to Use the Book at Work Now
Keep parts two to four. Anchoring, framing, the choice patterns of prospect theory and sunk costs have been retested on thousands of people, and the outside view costs nothing to try. Together they change real decisions about price, pay, budgets and when to stop a project.
Any thinking fast and slow summary that repeats the priming stories without the retests is out of date. If a slide tells you that a word, a picture of eyes or a sugary drink quietly steers behaviour, ask whether the study has been repeated. For five in our table, a direct or multi-lab rerun found nothing. For five more, the eyes and the sugary drink among them, pooled or secondary evidence points the same way.
One likely reason the failed stories still circulate is that a claim you have heard many times feels checked. We meet the same thing with famous people. Seven of 25 well-known claims in our piece on the Cristiano Ronaldo diet are in his own words, and 12 of 21 in our check of the Rock workout.
Quotations drift the same way. Our tracing of David Goggins quotes found eight of the 15 most repeated in his own text or voice, and 11 of 32 pages printing Theodore Roosevelt’s Man in the Arena passage cut or reword it.
If you want a next book, our ranking of the best books for decision making sorts them by what each changes about your next decision. For something lighter, our count of Reese Witherspoon’s book club found that 10 of its last 24 picks are first novels.
How We Did This
On 15 November 2026 our newsroom listed 16 well-known studies and effects that Thinking, Fast and Slow describes, drawn from its chapters on self-control, priming, cognitive ease, the hot hand, anchoring, prospect theory, sunk costs and framing. For each one we looked for the largest later test we could load: a multi-lab replication, a direct replication, a pooled analysis of preregistered experiments, or a published reanalysis.
We recorded the sample and result as each page states them. A study counts as failed only when a direct replication or a registered multi-lab test found no effect, and as held when a large rerun found the effect in the original direction. Where the evidence against a study is a pooled analysis, a secondary report or a reanalysis, we call it disputed: money priming, watching eyes, the sugar drink, the hard-to-read font and the parole judges.
The marshmallow study came back weaker in a conceptual replication, and the hot hand study was reversed by a reanalysis of its own data.
Studies with no large rerun that we could load, such as the planning fallacy thesis study, were left out, and several journal pages that would not open were replaced by university repository copies. This thinking fast and slow summary draws on Wikipedia, LitCharts and the 2017 audit of chapter four for the book’s contents. The full dataset, with the source and supporting text for every row, is available on request.
Frequently Asked Questions
Is Thinking, Fast and Slow still worth reading?
Yes. The chapters on anchoring, framing and prospect theory rest on work that has been retested at scale and held. The chapters on priming and self-control describe several results that later failed, and Kahneman said in 2017 that he had trusted small studies too much.
What is the main idea of Thinking, Fast and Slow?
That most judgments come from a fast, automatic system that goes predictably wrong in certain situations, such as a number on the table or a choice framed as a loss, and that the slower, effortful system is too lazy to step in as often as it should.
Did Daniel Kahneman retract any of the book?
He did not issue a revised edition that we could find. He did write, in a public comment on 14 February 2017, that he had changed his views about the size of behavioural priming effects and that the chapter’s results could not be as strong as it suggested.
The bat-and-ball sum takes ten seconds to check, and most people do not check it. Kahneman did not check some of his favourite studies either. When others did, he said so in public.

Written by
Editor of The Executive Insight. Writes about leadership, decision-making and the parts of building a business that nobody puts in the plan.
Explore other articles
-
Decision-Making
A career change at 50 usually goes wrong for one reason: people try to change everything at once, industry, function, income level and identity in a single jump. The version that works moves one of those and keeps the rest, so the experience you already paid for carries straight into the new role. That is […]
13 min read -
Decision-Making
It is Sunday night and you are dreading a job you were proud to win three years ago. The Steve Jobs Stanford speech speaks to that feeling. Delivered on 12 June 2005, it is three short stories, about connecting the dots, love and loss, and death, and each one works as career advice. Our newsroom […]
16 min read -
Decision-Making
It is late, the house is quiet, and three tabs are open. One is a $49-a-month certificate from a company everyone has heard of. One is a bootcamp that promises a new job in sixteen weeks. One is a university program whose price makes you close the tab. Choosing between online courses for career change […]
9 min read









