Estimated reading time at 200 wpm: 13 minutes
It is a reasonable assumption that people who have been through university should not make basic errors of thinking. They have been selected for ability, trained in a subject, and surrounded for years by argument, evidence, and criticism. History does not support the assumption. This article follows from Critical Thinking 2.0, in 2023.
Whether or not you agree our Fat Disclaimer applies
Educated people have repeatedly endorsed weak inferences, attacked the speaker instead of the claim, treated two options as if they exhausted the field, and taken sequence for cause. The failures are not confined to one period, one faculty, or one political camp. They appear wherever intelligence and formal training are mistaken for a guarantee of disciplined reasoning.
Critical thinking can be developed. It is not a mysterious gift that some possess and others do not. Reviews of teaching studies have shown that instruction improves the relevant skills, especially when the methods are taught directly rather than left to appear by accident. That is the point. The skill responds to practice, which means using the brain.
Universities almost always say they value critical thinking. They much less often do they teach it as a subject in its own right. The usual arrangement is to list it among the aims of a degree and then assume it will emerge from studying chemistry, law, history, or medicine. Faculties often believe they are teaching it. Many are wrong. Dedicated courses exist, usually in philosophy, but they are not the normal path through a degree. A typical graduate can leave without ever having been shown, formally, how to examine an argument. Higher education does not produce this capacity by default.
The greatest and most laughable situation is where individuals believe they are ‘critical thinkers’ when they are not. These types often demonstrate zero skill at critical thinking.
What the record actually shows
The assumption that a degree produces skills in critical thinking has been tested. It does not hold as a default result of attending university.
Richard Arum and Josipa Roksa’s Academically Adrift found that a large share of students showed little or no gain in critical thinking after two years of college. Later commentary has treated that finding as a standing problem rather than a one-off embarrassment: institutions say the skill is essential, then fail to show that the ordinary degree produces it. Natalia Pasternak put the same point more bluntly in Times Higher Education: higher education does not teach critical thinking by default. Departments claim they already teach it through their regular courses, they disagree about what the phrase means, and many students leave still unable to separate a sound claim from a weak one.
The difficulty is not that the skill is unteachable. A meta-analysis of instructional studies found that critical thinking can be improved when it is taught, especially when students are given dialogue, authentic problems, and mentoring. Other reviews reach the same conclusion: it can be trained, particularly when it is an explicit aim rather than a hoped-for by-product. An OECD report asked the question directly — does higher education teach students to think critically? — and treated the answer as an empirical matter, not a slogan.
What universities usually do is something else. Faculty often believe they are teaching the skill; students more often experience it as implicit. A Harvard study of professional schools found that more than half of instructors said they taught critical thinking explicitly, while students described it as something that was supposed to emerge from discussion and case work. A later faculty survey found the same gap: skills listed in learning outcomes are not always taught, and much of what is called critical thinking never becomes a named object of instruction. A 2026 commentary stated the distinction without decoration. Embedding the skill in subject teaching is not the same as teaching it. The advent of AI may have serious negative effects on critical thinking skills.
Standalone courses exist, usually in philosophy. They are not the ordinary route through a degree. For most graduates, formal training in how to examine an argument remains incidental. That is why the opening assumption fails. Education produces knowledge and fluency. It does not, by itself, produce the habit of testing an inference.
Consistency is not a sealed box
The earlier piece, Critical Thinking 2.0, treated critical thinking as a set of standards and traits that can run through any domain: curiosity, humility, fairness, consistency and breadth on one side; clarity, accuracy, precision, relevance and depth on the other. Logical consistency was already one node among several. This article takes that node and asks what it is made of.
It is made, in large part, of the ability to see when an argument fails. Formal and informal fallacies are the usual ways consistency breaks. Once those failures can be identified, the same check starts to serve the other domains. It does not serve them all equally. Some mappings are tight. Some are real but partial. Curiosity is served only indirectly: a detected fallacy is a reason to keep looking rather than to close the question.
| Domain | How knowledge of fallacies serves it |
|---|---|
| Logical consistency | The home node. Formal and informal fallacies are the usual ways consistency fails. |
| Relevance | Fallacies of relevance are this standard applied to argument. Ad hominem, red herring, straw man, appeal to force: the reasons do not bear on the claim. |
| Clarity | Equivocation, amphiboly, accent, quoting out of context. The wording shifts, so the argument is no longer one argument. |
| Precision | Quantifier shift, illicit conversion, composition and division. A term is used more loosely or more widely than the premises allow. |
| Accuracy | Suppressed evidence, false cause, hasty generalisation. The picture offered is incomplete or the link is not established. |
| Depth | Slippery slope, accident, false dilemma. A thin chain or a forced pair of options stands in for the real complexity. |
| Breadth | False dilemma again, and special pleading. Relevant alternatives or comparable cases are shut out. |
| Fairness | Straw man, tu quoque, ad hominem, special pleading. The other side is misrepresented, or a rule is applied only to them. |
| Intellectual humility | Begging the question, appeal to authority when it replaces reasons, genetic fallacy. The arguer treats a point as settled when it is not, or treats a source as decisive. |
| Intellectual curiosity | Fallacy knowledge does not create curiosity. It does stop a premature close. |
The maps that follow are the working detail under that join. They do not replace the wider picture. They give it a way of testing the argument in front of you.
The taxonomies of fallacy
The phrase “logical fallacy” has drifted. It is now used for almost any mistake in thinking: a bias, a lapse of confidence, a rhetorical habit, a psychological finding. Popular lists grew by addition. Dunning–Kruger sits beside affirming the consequent. The result is a catalogue, not a classification.
A fallacy, in the stricter sense, is a defect in an argument. It can be seen in the form of the inference, or in the relevance of the reasons, or in the language, or in what has been presumed. It is not every way in which a mind can go wrong.
The maps below return to that older cut. The first division is formal versus informal. Formal fallacies fail by structure: the pattern itself does not support the conclusion. Informal fallacies fail by content — by irrelevance, by ambiguity, or by presumption. Under the formal branch the errors are grouped by the logical system they belong to: syllogistic, propositional, quantificational, probabilistic. That is enough architecture. The rest is in the nodes.
This site cannot cater for all browsers and viewing devices. No map seen, means you have the wrong device. Tough.
Each node carries a short note: what the error is, an example, and why it fails. The maps are meant to be used, not merely viewed. Click through the branches that interest you. The article will not rehearse every item. The classification is there for inspection.
Where the argument is most likely to fail
A fallacy is a defect in reasoning. It is also, in practice, something that appears under pressure. We identified more than a dozen situations in which fallacious arguments reliably emerge: overconfidence in a weak grasp of a subject, dissonance after a public commitment, hierarchy and dependency, urgency, reputation, and the compression of language in headlines and clips. Three will do here.
The first is a threat to the self. A claim is bound to competence, status, or face. When the argument is shown to be weak, the a person may reply as if they had been e.g. “What — do you think I’m stupid?” and “No need for insults”. Those are not answers to the inference. They replace it. The original point is abandoned. In its place come a straw man of the critic, an ad hominem about tone, and a red herring about insults. Nothing in the critique required that translation. The translation is the defence.
The second is the half-question. “Are you saying…?” sounds like a request for precision. In the use that matters, it is a substituted claim offered for the speaker to own. A yes accepts the distortion. A no still leaves the worse sentence in play. The Channel 4 interview of January 2018, in which Cathy Newman repeatedly restated Jordan Peterson with “so you’re saying”, is the public exhibit. He declined the restatement and stayed with his own wording. Most people, once they hear the question, either accept the caricature or start defending themselves.
The third is politics. It is the largest arena because the other vulnerabilities are present at once. The argument is public, urgent, and tied to faction and power. Straw man and red herring are ordinary method. Ad hominem and tu quoque stand in for meeting the point. False dilemma and slippery slope force a decision. Appeals to popularity, emotion, authority, and force travel better than a qualification. Genetic fallacy treats the source of a policy as its merit. The “are you saying” move is at home there because a distorted sentence is easier to circulate than the real one.
These are not types of person. They are conditions under which the maps become useful: not as a scorecard, but as a way of seeing that the argument in the room is no longer the argument that was offered.
Not getting down and dirty
Recognition is not the end of the work. Once a defence has started, the new sentences are often worse than the original claim. The task is to stay on the argument and refuse the invitation to trade blows. Eight moves are enough.
1. Restore the claim: The first loss is the wording. What was said is replaced by something larger, smaller, or uglier. Repeat the actual sentence, in its actual scope. Do not improve it and do not accept the improvement offered from the other side. Until the original claim is back in the room, there is no argument to test. Everything else is commentary on a ghost.
2. Separate the person from the inference, once: A single plain sentence will do: the objection is to the reasoning, not to the speaker’s intelligence or character. Say it once. Saying it again begins to sound like a request for permission. The hearer may still take it as insult. That is their translation. You have done what the situation requires.
3. Politely reject the half-question: “Are you saying…?” is a request only in appearance. If the restatement is false, do not answer yes or no. Correct the wording first, or decline the question. Answering it, even to deny it, leaves the worse sentence in play. The disciplined reply is to the claim that was made, not to the claim that was invented.
4. Hold one point: A defensive exchange multiplies issues. Each new charge is an exit from the last. Choose one defect — one missing link, one substitution, one irrelevance — and stay there. A second fallacy can wait. Breadth here is a trap. Depth is the method: the same point, restored, until it is either met or abandoned.
5. Name the move, then return: A short name can help: that replaces what I said; that is about tone; that changes the subject. Then go back to the claim. The name is a pointer, not a sentence. Lingering on the label turns the discussion into a lesson, and the other party will treat the lesson as another insult.
6. Do not pay in the same coin: The loaded reply is an offer to trade. Ad hominem for ad hominem, caricature for caricature, motive for motive. Decline. Matching the move confirms that the subject is now combat. The discipline is not saintliness. It is keeping the only thing that can still be tested: whether the original inference holds.
7. Slow it: Ask for the original sentence again. Ask what would count as evidence. Ask which step is supposed to follow from which. Speed is the friend of the substitute claim and of the appeal to insult. A pause, a repetition, a request for the missing reason — these are not delay. They are the conditions under which an argument can reappear.
8. Leave when there is no argument left: If the only remaining subject is insult, loyalty, or the speaker’s wounded standing, there is nothing to manage. Ending the exchange is a method. It does not concede the point. It recognises that the inference has been vacated. Staying in order to win the last word is another way of getting down and dirty.
