07:30
Rob, before bed after his shift
He forwards an article to the family chat: a study finds students who use AI score lower on tests. He adds three worried faces. Nadia replies with a different article saying the opposite.
The way of thinking that asks what is being claimed, on what grounds, and what would change your mind. It works by breaking reasoning into its parts and holding each one to a standard.
Take the claim apart and ask what would have to be true
Critical thinking, in Ennis’s definition, is reasonable reflective thinking focused on deciding what to believe or do. Its tools take an argument apart: the claim, the evidence, the assumptions, the inferences that connect them. Paul and Elder list eight elements of reasoning and a set of standards to hold each against. It is strongest when a decision rests on a claim that could be wrong. It is weakest when there is no claim to test yet, and it depends heavily on knowing the subject.
In ordinary life
A headline says that people who drink coffee live longer. A critical thinker does not ask whether they like the headline. They ask what was actually measured, in whom, compared with what, and what else coffee drinkers might have in common. Most of the time the study is more careful than the headline, and says less.
Kahneman describes a mind that jumps to a coherent story from whatever evidence is in front of it. He calls it "what you see is all there is". Critical thinking is the slower habit of asking what you are not seeing, and whether the story would survive it.
The skill is not doubt. Doubting everything equally is as lazy as believing everything. The skill is weighing: some claims deserve more scrutiny than others, and the useful question is always what evidence would change your mind. Willingham’s caution applies here. You can only weigh evidence in a field you know. Critical thinking about coffee needs some idea of how nutrition studies work.
The educational experience
Every page follows the same school on the same kind of day. The school has decided to increase students’ use of generative AI, and the plan is due for review. The five people and the situation stay fixed. Only the way of thinking changes.
All five, interconnected
The five threads as they were written, interleaved, so the hand-offs between them are visible. Pick a person to follow their day on its own. The people are illustrative, and the same five appear on every way’s page.
07:30
Rob, before bed after his shift
He forwards an article to the family chat: a study finds students who use AI score lower on tests. He adds three worried faces. Nadia replies with a different article saying the opposite.
09:00
Mr Osei, preparing the review
The vendor report says writing quality rose by a third. It is the main evidence in his plan. He asks Ms Lim to check it before the governors see it, mainly as a formality.
10:45
Ms Lim, free period
She does not understand the statistics. She uses Paul and Elder’s elements instead. What is the question? What was measured? What is assumed? The report measured essay grades, marked by the tool’s own rubric, on essays written with the tool. The claim is that students write better. What was measured is that the essays scored better.
11:30
Nadia’s history class
Ms Lim gives the class the two articles from Rob’s family chat, without saying where they came from. In pairs they list what each one measured. Nadia notices that the "AI lowers scores" study tested students without the tool, after they had used it all term. The "AI raises scores" one tested them with it.
11:50
Nadia, same lesson
She says both studies might be right. One measured what students can do with the tool. The other measured what they can do without it. The class argues about which one the school should care about. Nobody wins, which Ms Lim counts as a success.
13:00
Ms Lim and Mr Osei
She shows him the gap between the claim and the measure. He is irritated, then interested. He asks what evidence would convince her the plan works. She says: better writing done without the tool, a term later. He writes it down.
15:00
Priya, drafting her advice
She has eleven reports on her desk, six for AI-supported writing and five against. She stops counting them and sorts them by what they measured. They sort cleanly into with-the-tool and without-the-tool, and they barely disagree once they are sorted.
16:30
A call between Priya and Mr Osei
They agree the plan can go ahead with one change. The school will measure unassisted writing at the end of each term, and the plan will be reviewed against that, not against the vendor’s rubric.
19:45
Nadia and Rob, at dinner
Nadia explains the two studies. Rob asks which one is right. She says that is the wrong question, and then has to explain why, which takes the rest of dinner.
The evidence did not change during the day. What changed was the question asked of it. Sorting the claims by what they measured dissolved most of the disagreement and left one real question: which capability does the school want. That question is a value judgement, and critical thinking was only able to find it, not answer it.
Implications
The same five dimensions for all four ways. This is the part that can be tabulated. The two narratives above are the part that can be argued with.
The gap between what is claimed and what was shown. Critical thinking notices when a measure has been swapped for the thing it stands in for, when an assumption is carrying the argument, and when two sources that seem to disagree are answering different questions.
Anything that has not yet been claimed. Critical thinking needs something to evaluate, so it is weak at the start of a problem and at generating options. It can also make a group very good at finding flaws in every proposal and very slow at making one.
It sees arguments more clearly than it sees structures. A set of individually sound decisions can still add up to a bad pattern, and no amount of scrutinising each decision will reveal it.
Knowledge first. Willingham’s argument is that critical thinking cannot be taught as a general skill cut loose from content, because what counts as good evidence differs between history and chemistry. Teachers need to teach it inside their subject. Students need to be allowed to reach conclusions, not only to find faults.
Good work names what evidence would change the writer’s mind. Weak work lists objections to every side and settles on none.
Behind the question of what to measure. In the day above, critical thinking cleared away the false disagreement and then stopped at the real one: whether the school cares more about what students can do with the tool or without it. That is a question about purpose. The Delphi Report’s experts included fair-mindedness among the dispositions of a critical thinker, which is an admission that the method needs values it cannot supply.
A poster of question stems, and a generalised scepticism that treats every claim as equally doubtful. The hollow version teaches students to say "but what about bias?" about everything. That is a verbal habit, and it can make a class more confident and less accurate at the same time.
Reading
Published works only. Start with the first on the list.
Ennis, R. H. (1996). Critical Thinking. Prentice Hall.
Critical thinking as reasonable reflective thinking focused on deciding what to believe or do.
Paul, R., & Elder, L. (2006). The Miniature Guide to Critical Thinking: Concepts and Tools. Foundation for Critical Thinking.
The eight elements of reasoning and the intellectual standards to hold them to.
Facione, P. A. (1990). Critical Thinking: A Statement of Expert Consensus for Purposes of Educational Assessment and Instruction (The Delphi Report). American Philosophical Association.
An expert consensus on the skills and dispositions involved.
Willingham, D. T. (2007). Critical thinking: Why is it so hard to teach? American Educator, 31(2), 8–19.
Why critical thinking does not transfer well without knowledge of the domain.
Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
Why the mind builds a coherent story from whatever is in front of it.