Skip to content

Mindset

8 min left
BOOK SUMMARY · THINKING

Think Again Summary: The Key Ideas, What Holds Up, and How to Use Them

Adam Grant's Think Again in plain words, and what the research says about each big idea.

Hacks Vitae artwork (AI-assisted)
The answer

Adam Grant argues that the ability to rethink may matter more than raw intelligence, and that we would do better to treat our views as hypotheses, as scientists do.

Its framework, from a 2002 paper by Philip Tetlock, and its Mann Gulch story hold up. Several findings it leans on are disputed or smaller than they sound.

Why it matterspending

By Hacks Vitae Published Updated Last checked October 2026 8 min read 9 sources ↓

Think Again is a book about changing your mind, by an organisational psychologist who teaches at Wharton. Here are its big ideas in our own words, what research says about the studies it leans on, and a few small practices drawn from the parts that hold up.

About the book

Think Again: The Power of Knowing What You Don’t Know is by Adam Grant. Viking published it in hardcover on 2 February 2021; Google Books lists it at 320 pages, and a paperback followed from Penguin Books on 26 December 2023. The publisher sums up its argument in one sentence: “We think too much like preachers defending our sacred beliefs, prosecutors proving the other side wrong, and politicians campaigning for approval—and too little like scientists searching for truth.”

It is a general-audience book that, in a cover blurb’s words, “weaves together research and storytelling”. The author’s own discussion guide groups its themes in three levels: individual rethinking, interpersonal rethinking (opening other people’s minds) and collective rethinking (schools, workplaces and communities).

We read the publisher’s pages, the author’s website and discussion guide, a 2021 interview with him, and short passages of the 2021 Viking edition through the Google Books search-inside tool. Page numbers below are from that edition.

The big ideas

1. Drop your tools

The book opens with the Mann Gulch fire of 1949, in which a crew of smokejumpers was overtaken by a wildfire in Montana. The foreman, Wagner Dodge, survived by lighting an escape fire and sheltering in the burned-over patch. Grant uses the story as a metaphor: “You may not carry an ax or a shovel, but you do have some cognitive tools that you use regularly.” As we read it, the point is that the habits and beliefs we rely on can become a weight we fail to put down.

2. Preacher, prosecutor, politician, or scientist

Grant describes three modes we fall into. In preacher mode we defend beliefs we hold sacred; in prosecutor mode, in the book’s words, “we marshal arguments to prove them wrong”; in politician mode we campaign for an audience’s approval. He recommends a fourth, the scientist. In a 2021 interview he put it this way: “when I start to form an opinion, I should treat it as a hypothesis”.

3. Confidence and competence are not the same thing

A chapter on overconfidence draws on the Dunning-Kruger effect, which the book describes this way: “it’s when we lack competence that we’re most likely to be brimming with overconfidence.” Grant uses the label “Mount Stupid” for the point where a little knowledge makes us feel expert, and his discussion guide speaks of “confident humility”. The book also argues that feeling like an impostor can be useful: in its words, it “can make us better learners”.

4. Clever is not the same as open-minded

The publisher’s description puts it bluntly: “Intelligence is no cure, and it can even be a curse: being good at thinking can make us worse at rethinking.” It goes on: “The brighter we are, the blinder to our own limitations we can become.”

5. Think like a scientist, even in business

Grant describes an experiment with Italian startups, most of which “had yet to bring in any revenue”. Founders who were taught to treat their plans as hypotheses and test them did better than founders who received the same training without that method.

6. Change minds by listening, not lecturing

The book’s middle section is about rethinking with other people. Grant describes motivational interviewing, a counselling method built on open questions, affirming, reflective listening and summarising, and tells the story of someone the publisher calls a “vaccine whisperer” who used it with hesitant parents. One of his guiding principles, in the publisher’s words, is to “argue like he’s right but listen like he’s wrong.”

7. Build places where rethinking is normal

The book also covers habits and groups: building a “challenge network” of people who will tell you when you are wrong, encouraging disagreement about tasks without letting it become personal, and, in its last part, schools and workplaces that build “a learning culture”, along with regular “career checkups” that ask whether a plan still fits.

PreacherDefends beliefs held sacred.
ProsecutorMarshals arguments to prove the other side wrong.
PoliticianCampaigns for an audience's approval.
ScientistTreats a new opinion as a hypothesis.

The first three come from Philip Tetlock; the scientist is the mode Grant recommends.

Fig. 1Four modes of thinking, as the book describes themGraphic: Hacks Vitae, from this article.

What holds up

What the sources show for each idea

01 The preacher, prosecutor and politician modes come from Philip Tetlock Holds up Grant says so, and the paper exists.
02 The less competent we are, the more overconfident we are (the Dunning-Kruger effect) Still argued The overestimation is real; its size and cause are argued over.
03 A little knowledge makes beginners overconfident ("Mount Stupid") Partly There is lab evidence for a beginner's bubble; the famous curve is not from the original study.
04 Smarter people are worse at changing their minds Still argued One well-known study found it; two large replications did not.
05 Startups taught to think like scientists earn far more Partly The trial was real; the revenue gain is smaller and less certain than the story suggests.
06 Motivational interviewing persuades vaccine-hesitant parents Partly Promising, with weaker evidence than the story suggests.
07 On a multiple-choice test, sticking with your first answer is the safer bet Doesn't hold up Changing answers more often helps than hurts.
08 At Mann Gulch, the foreman survived by lighting an escape fire his crew did not follow Holds up The record matches the book's outline.

The book’s core framework is sourced and its opening story is accurate. Several of the studies it leans on, though, are disputed or depend on small samples, and two of its practical claims hold up only with limits. The rows above give each claim, our verdict and the source.

A note on the Italian startup numbers

This is one of the book’s most concrete examples, so it is worth looking at closely. The book reports that the scientific-thinking group “averaged over $12,000 in revenue”. In the published paper, revenue is counted in euros, and only 17 of the 116 startups earned any revenue during the study: 9 in the trained group and 8 in the comparison group. An average across so few earning firms can move a lot with one or two of them. The later, larger study by the same team, which pooled the original trial with three new ones, supports the decision-making part of the finding: trained founders were more willing to drop ideas. It also found that they earned more on average, by about 7,000 euros, though the difference was statistically clear in only one of the three new trials, and the authors note that many firms earned nothing at all.

17 of the 116Startups that earned any revenue during the original study.
9 and 8Earning firms in the trained and comparison groups.
About 7,000 eurosExtra average revenue for trained founders in the follow-up, statistically clear in only one of the three new trials.

The later study was run by the same team and pooled the original trial with three new ones.

Table 1The Italian startup trial, closer upGraphic: Hacks Vitae, from the figures in this article.

How to use it

These practices come only from the parts that hold up. They describe what people do; what fits your situation is your call.

  1. Naming the mode. Noticing, in the middle of an argument, whether you are defending, prosecuting or campaigning is a quick check drawn from Tetlock’s framework. It is a way of looking rather than a tested technique, but it costs nothing to try.
  2. Revisiting test answers you feel unsure about. The research on answer-changing suggests that a second look, guided by how confident you feel about each answer, more often helps than hurts.
  3. Asking before telling. Motivational interviewing’s core moves, open questions and reflecting back what you hear, have some evidence behind them, mostly for short-term change, and they are easy to try in an ordinary conversation.
  4. Treating a plan as a hypothesis. The startup studies suggest that one clear benefit of testing a plan is being readier to drop a weak one early, an idea that carries over to many projects beyond business.

Who it’s for, and who can skip it

It suits readers who like ideas carried by stories, and anyone who wants to get better at disagreeing or at changing their own mind. Readers who want to judge the evidence for each claim will find the stories more vivid than the studies are strong; the research above is the place to look for that. Readers who already know Tetlock’s work on forecasting, or the cognitive-bias literature, will recognise much of the first part.

If you liked this

How we checked this · Last checked October 2026
  • We read the publisher's page, the author's book page and discussion guide, a 2021 interview, short search-inside passages of the 2021 Viking edition and the publisher's sample of the prologue and chapter 1.
  • We read the two Mann Gulch history pages, the 2018 Quebec study, the 2018 review, and the 2020 startup paper and its 2024 follow-up in full.
  • For the other studies listed we read the abstracts.

Checked October 2026. What we read: the publisher’s page, the author’s book page and discussion guide, the 2021 interview, and short search-inside passages of the 2021 Viking edition, the publisher’s sample of the prologue and chapter 1; the two Mann Gulch history pages in full; the 2018 Quebec study and the 2018 review in full; the 2020 startup paper and its 2024 follow-up in full; and the other studies above (abstracts).

How we check every article →

Sources

Primary written at the time, by a witness, or the original text Scholarly edited or peer-reviewed work Popular press, blogs and general websites