Deduction vs Induction vs Abduction

Sherlock Holmes is the most famous deductive reasoner in fiction. There is just one problem: he almost never uses deduction. What Holmes actually does, in story after story, is a different type of reasoning entirely. Conan Doyle called it the wrong thing, and the mistake stuck for over a century.

This is not pedantry. Understanding the three types of reasoning (deduction, induction, and abduction) makes you a better observer, because each type answers a different question and carries a different level of certainty.

Deduction: the guarantee

Deductive reasoning moves from a general rule to a specific conclusion. The classic example:

1. All men are mortal. (General rule.)
2. Socrates is a man. (Specific case.)
3. Therefore, Socrates is mortal. (Conclusion.)

The power of deduction is certainty: if both premises are true, the conclusion must be true. There is no probability involved, no "most likely." The catch is that you need premises that are actually true and actually general. In real life, absolute rules are rare. "All doctors carry stethoscopes" is not true. "Most practising physicians own a stethoscope" is closer, but "most" breaks the deductive guarantee.

Holmes occasionally uses pure deduction. In The Norwood Builder, he reasons: the only person who benefits from the apparent death is the suspect; the death is staged; therefore the suspect staged it. The logic holds because the premise (only one beneficiary) is established as fact in the story. But these moments are exceptions.

Induction: the pattern

Inductive reasoning moves in the opposite direction: from specific observations to a general rule.

1. Every raven I have seen is black.
2. Therefore, all ravens are (probably) black.

Induction is how science builds theories. You observe patterns, test them, and gradually gain confidence. The conclusion is never certain (the next raven could be white), but it gets stronger with more observations.

Holmes uses induction when he builds his mental library. He catalogues cigar ashes, regional soils, and typeface styles not because any single catalogue entry is a complete case but because the patterns become reliable after enough samples. When he later recognises a specific ash at a crime scene, he is drawing on inductively built knowledge.

Abduction: the best explanation

Abductive reasoning starts with an observation and works backward to the most likely cause. Philosopher Charles Sanders Peirce formalised the concept in the 1890s, around the same time Doyle was writing Holmes:

1. The streets are wet. (Observation.)
2. If it had rained, the streets would be wet. (Explanatory hypothesis.)
3. Therefore, it probably rained. (Best explanation.)

This is what Holmes does in almost every story. He observes a set of clues (callused hands, ink stains, a tan line ending at the wrist), considers which explanation accounts for all of them, and picks the most probable one. He is not moving from general to specific (deduction) or from specific to general (induction). He is moving from effect to cause.

In A Study in Scarlet, Holmes looks at Watson and declares that he has been in Afghanistan. The reasoning: Watson has the bearing of a military doctor, a tan that suggests the tropics, a wounded arm, and a haggard face. The only theatre of operations in 1881 that fits all four observations is Afghanistan. Holmes does not have a rule that says "all men with tan lines are Afghan veterans." He has a set of observations and one explanation that fits them all.

Why the distinction matters for you

When you study a Daily Holmes photograph and write a deduction, you are almost certainly using abduction. You see a cluster of objects and ask: what kind of person would have all of these in the same room? That is inference to the best explanation.

Knowing this changes how you evaluate your own reasoning:

• Deduction is rigid. If you have an airtight premise, you can be certain. But airtight premises about a stranger's room are rare. "This person has a medical degree on the wall, so they have a medical degree" is deductive but trivial. The interesting work is abductive.
• Induction builds your library. The more rooms you study, the better your sense of what clusters of objects typically mean. That growing library is what observation training builds.
• Abduction solves the case. You weigh multiple observations, consider competing explanations, and pick the one that accounts for the most evidence with the fewest assumptions. The best deductions in Daily Holmes are the ones that chain together three or four independent clues into a single coherent profile.

How to practise abductive reasoning

Holmes's methods are abductive at their core. Here is a simplified practice loop:

1. Gather observations. Scan the scene with the six lenses (occupation, hobbies, lifestyle, relationships, recency, absence).
2. Generate two or three candidate explanations. Do not stop at the first one that feels right. Force yourself to imagine at least one alternative. "This could be a music teacher, or it could be a retired musician who gives lessons on the side."
3. Test against the evidence. Which explanation accounts for the most clues? Which one requires the fewest assumptions? Which one does not contradict anything visible?
4. Commit and check. Write your chosen explanation down. Then see how it holds up against the answer.

The generate-alternatives step is the one most people skip, and it is the one that improves accuracy the most. Forcing yourself to consider a second explanation prevents you from locking onto the first plausible story and ignoring contradictory evidence.

Frequently asked questions