LOGICAL FALLACIES
Ten Ways to Reason Yourself Into a Disaster
Ten logical fallacies, and the question that breaks each one.
A logical fallacy is an argument whose premises fail to support its conclusion, while appearing to do so. This article describes ten fallacies that recur in everyday and professional reasoning. Each is defined and illustrated, and each is analysed for the information it omits. The analysis draws on the Meta-Model of Bandler and Grinder (1975), which treats faulty statements as products of deletion, generalisation, and distortion. The article argues that fallacies are best addressed by recovering the missing information, not by asserting a counter-conclusion.
Most bad reasoning doesn’t feel like bad reasoning. It feels like common sense. A fallacy isn’t a mistake. It’s a strategy. Someone runs it in their head, it works for them, and that’s why you can’t argue them out of it. So skip the debate. Ask questions that make the person go back and find the missing piece.
1. Hasty generalisation
“Three patients died on that ward this year. That ward kills people.”
Challenge: “It’s an oncology ward. What were they supposed to die of, boredom?”
The takeaway: A count means nothing without a base rate. Ask how many died elsewhere, and out of how many admitted. (Meta-Model: universal quantifier.)
2. False cause
“Every time I quit smoking, I fall seriously ill. Cigarettes were holding me together.”
Challenge: “Did you quit and then get ill, or get ill and then quit?”
The takeaway: Check the direction of the arrow. Many people quit after the diagnosis, so the illness caused the quitting. (Meta-Model: cause-effect.)
3. Mind-reading
“My mother hasn’t called in a week. She’s punishing me.”
Challenge: “Or she’s dead. Have you checked?”
The takeaway: Silence has many explanations, and the worst ones are the ones you skipped. List them before you pick the one that flatters your grievance. (Meta-Model: mind-reading.)
4. Appeal to authority
“He’s a Nobel laureate, and he says the disease isn’t caused by the virus.”
Challenge: “Laureate in what? Has he treated a single patient?”
The takeaway: Expertise doesn’t transfer between fields. Ask what the authority would point to as evidence, and what other experts in that field say. (Meta-Model: lost performative.)
5. Ad hominem
“Of course the widow says the company was negligent. She’s grieving, she’s biased.”
Challenge: “Was the railing rotten or not? Grief doesn’t rot railings.”
Get plutonium’s stories in your inbox
Join Medium for free to get updates from this writer.
The takeaway: Motive bears on credibility, not on facts. Check the claim independently, especially when the messenger is inconvenient. (Meta-Model: complex equivalence.)
6. False dilemma
“Either I make partner, or these eleven years were a waste.”
Challenge: “Waste by whose accounting? Nobody’s eulogy opens with ‘he made equity partner.’”
The takeaway: Binary framing hides options, and it hides the person who set the terms. Ask who drew the two-item list. (Meta-Model: modal operator of necessity.)
7. Slippery slope
“If I admit the error, I’m fired, then I miss the EMI, then my wife leaves, then I’m on the street.”
Challenge: “You’ve booked the divorce already. Which of these happened to anyone you know who admitted a mistake?”
The takeaway: Multiply the links. Five steps at 60% each: 0.6 × 0.6 × 0.6 × 0.6 × 0.6 ≈ 0.078, under 8%. Catastrophic chains feel likely because each step is vivid, not because each is probable. (Meta-Model: cause-effect, chained.)
8. Bandwagon
“The whole village agreed she was a witch. Two hundred people can’t be wrong.”
Challenge: “Two hundred people, and how many checked?”
The takeaway: Agreement without independent verification is one opinion repeated two hundred times. Ask how many people reached the belief separately. (Meta-Model: universal quantifier, deleted source.)
9. Circular reasoning
“Our internal investigation found no wrongdoing. We conducted it ourselves.”
Challenge: “Who investigated the investigators?”
The takeaway: Ask whether the finding could have come out differently. If the process can’t return “guilty”, it isn’t evidence. (Meta-Model: nominalisation.)
10. Appeal to emotion
“She loved him so much. She couldn’t have poisoned him.”
Challenge: “Love and arsenic aren’t mutually exclusive. What did the toxicology say?”
The takeaway: A feeling is real data about the person and weak data about what they did. Ask what the facts show, apart from how the story feels. (Meta-Model: complex equivalence.)
Reference: Bandler, R., & Grinder, J. (1975). The Structure of Magic I. Science and Behavior Books.
Logical Fallacies, Critical Thinking, Psychology, NLP, Communication
