Logic

Deduction vs Induction

Sherlock Holmes said 'deduction', but in reality he used induction and abduction. Understanding the difference between types of reasoning is the key to knowing when conclusions can be trusted and when to be skeptical.

  • **Science:** All scientific laws are inductive generalizations, always open to revision
  • **Law:** Legal precedents are applied deductively, but are themselves created inductively from practice
  • **Medicine:** Diagnosing from symptoms is induction; treating by protocol is deduction

Deduction

**Deduction** is reasoning from the general to the specific. If the premises are true, the conclusion is GUARANTEED to be true. A deductive argument is either valid (the conclusion follows necessarily) or not.

**Key property:** In a valid deductive argument it is impossible for the premises to be true while the conclusion is false. The information in the conclusion is already contained in the premises - we merely 'extract' it.

**Deduction preserves truth but doesn't create new knowledge.** The conclusion is logically contained in the premises. We don't learn anything new - we make explicit what was implicit in the premises.

'All mammals are warm-blooded. Whales are mammals. Therefore, whales are warm-blooded.' This reasoning is:

Induction

**Induction** is reasoning from the specific to the general. We observe cases and draw a conclusion about a pattern. An inductive conclusion is PROBABLE, but not guaranteed - even with true premises.

**The problem of induction (Hume):** Why do we believe a pattern will continue? 'The sun rose 10,000 days in a row' does not prove it will rise tomorrow. Induction works in practice but has no logical justification.

0

1

Sign In

**Induction expands knowledge, but with the risk of error.** Unlike deduction, induction 'goes beyond' the data. This is its strength (we learn something new) and its weakness (we might be wrong).

A scientist studied 500 samples of metal X and found they all conduct electricity. They conclude: 'Metal X conducts electricity.' What is the problem with this conclusion?

Certainty vs Probability

**Deduction yields certainty, induction yields probability.** This is a fundamental difference. A deductive conclusion either follows or it doesn't. An inductive conclusion can be more or less probable.

**Practical implication:** Mathematics and logic use deduction - their conclusions are absolute. Science uses induction - its conclusions are always provisional and may be revised.

**The strength of an inductive argument** is measured by probability. The more observations, the more varied they are, and the fewer counterexamples - the stronger the induction. But it never becomes deduction.

The law of universal gravitation has been confirmed by millions of experiments. Can it be said to be proven with absolute certainty?

Generalization

**Generalization** is the inductive transition from observations to a general rule. The quality of a generalization depends on the sample size, its representativeness, and the absence of systematic errors.

**The key question:** How well do the observations represent the entire population? If the sample is biased, the generalization will be flawed even with a large number of observations.

**Hasty generalization** - a logical fallacy where a conclusion is drawn from an insufficient or non-representative sample. 'I've encountered two rude taxi drivers - all taxi drivers are rude.'

The more examples, the more reliable the generalization

Representativeness matters more than size - a biased sample is flawed at any size

A million biased observations are worse than a hundred representative ones. If the sample systematically excludes certain cases, increasing its size will not fix the error.

A researcher surveyed 1,000 users of their app and found 80% are satisfied. They conclude: '80% of people are satisfied with our app.' What is wrong?

Key Takeaways

  • **Deduction:** from general to specific, the conclusion is guaranteed, but adds no new knowledge
  • **Induction:** from specific to general, the conclusion is probable, expands knowledge, but carries risk
  • **The problem of induction:** even a million confirmations don't guarantee the rule will keep working
  • **Generalization:** requires a representative sample - size without diversity is not enough

Related Topics

Deduction and induction are the foundation for other types of reasoning:

  • Valid Argument Forms — Deductive schemas that guarantee the truth of the conclusion
  • Inductive Strength — How to measure the reliability of an inductive conclusion

Вопросы для размышления

  • Which of your beliefs are based on deduction, and which on induction?
  • Why do we keep using induction if it is not logically justified?
  • How can you distinguish a strong inductive generalization from a hasty one?

Связанные уроки

  • ml-03
Deduction vs Induction