Think Again

Grant, an organisational psychologist at Wharton, proposes four mental modes: preacher, defending sacred beliefs; prosecutor, attacking others'; politician, seeking approval; and scientist, forming hypotheses and testing them.

Most people default to the first three and treat changing their mind as defeat. His case is that the scientist mode is learnable and that the organisations which run it outperform.

Preacher, prosecutor and politician are all defending something. Only the scientist is trying to find out.

Devmark’s reading — not a quotation from the book

The central construct is confident humility: certainty in your capacity to learn combined with doubt about what you currently know. Grant distinguishes it from both impostor syndrome and the armoured confidence usually rewarded in hiring.

He reports the finding that the best forecasters update frequently and in small increments, and that the worst are distinguished by conviction rather than by information.

Task conflict versus relationship conflict — the finding that high-performing teams argue about the work more and about each other less. Motivational interviewing as a technique for changing someone else's mind by asking rather than telling. The practice of scheduling a rethinking cycle rather than waiting for evidence to force one.

The chapter on how not to argue is the most immediately usable: fewer reasons, better ones, because weak arguments dilute strong ones.

Adding weak arguments to a strong one makes the case worse, which is counterintuitive and well supported.

Devmark’s reading — not a quotation from the book

Grant writes in the standard popular-psychology mode — a construct, studies, anecdotes — and some of the studies cited have had replication difficulties, as is common across the genre.

The core claims about updating and about task conflict are well supported. The book is strongest read as a set of practices rather than as a settled account of cognition.

Grant's most memorable case concerns firefighters who died carrying equipment they had been ordered to drop, because the tools were part of who they understood themselves to be. He uses it as the book's central metaphor for identity-protective thinking: the difficulty is rarely the new information but the cost of having been the person who believed the old thing.

They did not die because the fire was faster. They died carrying what they could not put down.

Devmark’s reading — not a quotation from the book

His practical recommendation is to cultivate a small group whose job is to disagree with you — chosen for their willingness to say so rather than their supportiveness — and to make the role explicit so that criticism arrives as service rather than attack. It is a deliberate structural fix for the fact that most feedback environments quietly select for agreement.

A long section covers motivational interviewing: asking questions, reflecting the answer back, and leaving the person to notice their own inconsistency rather than being shown it. The reported results with vaccine-hesitant parents and with entrenched positions generally are better than direct argument, which reliably produces counter-argument.

Grant's account of why public disagreement degrades is that people compress continuous questions into two camps, and that adding complexity — more positions, explicit uncertainty, acknowledged trade-offs — measurably reduces polarisation in how the same evidence is received. The practical version is to present findings with their caveats intact rather than stripping them for impact.

The distinction he leans on is between argument about the work and argument about each other. High-performing teams in the research he cites have more of the first and less of the second, and the failure mode for most groups is either suppressing both or letting the first turn into the second. The managerial recommendation follows directly.

It is written in the standard popular-psychology shape: a construct, a sequence of studies, a stack of anecdotes, a prescription. Some of the supporting research is robust and some is single-study or has weathered the replication crisis poorly, and the book does not distinguish between them for the reader. The argument is better than several of its citations.

The reusable habits are three: notice when you are defending rather than learning, name the smallest thing that would change your mind before you argue, and build a relationship with at least one person who will tell you when you are wrong. None of that requires the four-mode framework.