Chapter 4 — The Brain That Wants to Believe

Chapter 4 — The Brain That Wants to Believe

We have taken the detective apart and found no engine inside — only a very good author and a very willing audience. So let us finally examine the audience, because the audience is you, and the machinery in your skull that makes Holmes feel like genius is the same machinery that will, if you let it, quietly ruin your judgment for the rest of your life. This is the chapter where the autopsy stops being about a fictional man and starts being about the reader holding the book.

The uncomfortable thesis is this: the mental talents that make Holmes thrilling are, in a real human being, malfunctions. Not quirks. Documented, named, laboratory-measured failure modes. Doyle took a catalogue of cognitive bugs and painted them gold and called the result a superpower. Let us scrape off the gold.

The organ that finds faces in clouds

Your brain is, before it is anything else, a pattern-detector — and a wildly, deliberately oversensitive one. You see faces in electrical sockets, in the fronts of cars, in the burnt patterns on toast; the phenomenon has a name, pareidolia, and a famous market: a Florida woman sold a grilled-cheese sandbearing the alleged face of the Virgin Mary for twenty-eight thousand dollars. The broader tendency — to find meaningful patterns in meaningless noise — the science writer Michael Shermer named patternicity. Its darker sibling, the tendency to sense a deliberate agent behind the pattern, he called agenticity. Together they run the whole human range from seeing a duck in a cloud to seeing a conspiracy in a coincidence.

You might expect evolution to have tuned this instinct toward accuracy. It did not. It tuned it toward survival, which is a different and cheaper thing. Consider the rustle in the grass. If you assume "predator" and you are wrong, you have wasted a little adrenaline and looked foolish to no one. If you assume "wind" and you are wrong, you are eaten. The two errors are not symmetric, so evolution built a detector that vastly overproduces false alarms — a smoke detector that shrieks at burnt toast, because a smoke detector that stays politely silent during real fires does not get to have descendants. Biologists formalise this as error management theory, and the arithmetic is unsentimental: when a false positive is cheaper than a false negative, the optimal creature is a superstitious one. You are descended from ten thousand generations of the jumpy. You see patterns everywhere because your silent, sceptical ancestors were lunch.

Now recall what we admire Holmes for. For seeing significance in the trifle. For the conviction that the cigar ash and the bootscrape and the calloused thumb are all speaking, all connected, all meaningful. In a world an author has pre-loaded with meaning, that conviction is genius. In the real world — the world of noise, the world where your shoes carry mud from nowhere in particular — that same conviction is apophenia, the compulsive perception of connections that are not there, a trait the psychiatrist Klaus Conrad first mapped at the onset of delusion. The line between the great detective and the conspiracy theorist is not the strength of the pattern-finding. It is identical in both. The only difference is whether an author rigged the world so the patterns are real. Yours didn't.

The clues you go looking for

Suppose you have leapt — abductively, in the manner of Chapter 1 — to a theory. The suspect is the butler. What does your mind do next?

It does not, whatever you fondly imagine, go hunting for evidence that the butler is innocent. It goes hunting for evidence that he's guilty, and it finds it, because a mind that has decided the butler did it will read every subsequent fact in the butler's disfavour. This is confirmation bias, and in a 1960 experiment the psychologist Peter Wason showed just how deep it runs. He gave people a number sequence — 2, 4, 6 — and asked them to discover the hidden rule by proposing further sequences to test. The rule was simply "any three ascending numbers," but almost nobody found it, because almost nobody tried to break their own guess. Convinced the rule was "even numbers going up by two," they proposed 8-10-12, 20-22-24 — sequence after sequence that could only confirm their theory and never expose it. The one move that finds truth — trying to falsify your own idea, offering 1-2-3 to see if it's allowed — almost no one made. We seek the "yes." We flee the "no." And Holmes, who never in sixty stories seriously entertains a theory in order to destroy it, is the great flatterer of this instinct.

His creator, again, half-knew. Holmes's most quotable methodological line — "when you have eliminated the impossible, whatever remains, however improbable, must be the truth" — sounds like a commitment to falsification, to ruling things out. But watch him work and you'll see he almost never does the ruling-out on his own leading theory. He rules out other people's theories, the clumsy inspectors', and crowns his own. The eliminating is aimed outward, never in. It is confirmation bias wearing the costume of its opposite.

The trap of the vivid story

Here is the deepest one, and the one most precisely responsible for the pleasure of a Holmes story. I am going to ask you a question first devised by the psychologists Daniel Kahneman and Amos Tversky.

Linda is thirty-one, single, outspoken, and very bright. She majored in philosophy. As a student she was deeply concerned with discrimination and social justice and joined anti-nuclear demonstrations. Which is more probable:

(a) Linda is a bank teller. (b) Linda is a bank teller and is active in the feminist movement.

If you felt the tug toward (b), you are in the overwhelming majority — roughly eighty-five percent of people choose it — and you have just committed the conjunction fallacy. Because (b) cannot possibly be more probable than (a). Every feminist bank teller is a bank teller; the set of (b) is entirely inside the set of (a); adding a detail can only ever make a claim less likely, never more. And yet (b) feels truer, because it tells a better story. It fits the vivid picture of Linda. The specific, colourful, coherent narrative overrides the cold logic of probability — and it does so in the exact direction that makes us feel we understand.

Now look at what a Holmesian "deduction" actually is. It is a long, specific, gorgeously detailed story: the man is not merely a soldier but an army doctor, invalided from the tropics, wounded in the left arm, lately returned from Afghanistan. Each added detail makes the picture more vivid, more satisfying, more Holmes — and each added detail, by the iron law that undid Linda, makes the conjunction strictly less probable than the plain claim it decorates. The pleasure of the deduction runs precisely opposite to its probability. We are most dazzled at the exact moment we should be most sceptical. Holmes is the conjunction fallacy rendered as entertainment: the more improbable the specific story, the more we applaud, because our applause is tracking coherence and calling it truth.

Stack these findings up and a portrait emerges of the mind that reads Sherlock Holmes and yearns to be him. It over-finds patterns, by ancient design. It anchors on its first idea — Kahneman and Tversky spun a rigged wheel of fortune, and a number the subjects knew was random still dragged their later estimates toward it — and then defends that idea by seeking only confirming evidence. It mistakes the vivid story for the likely one. And crowning all of it sits what Kahneman called WYSIATI — what you see is all there is — the mind's refusal to account for the evidence it doesn't have, its serene confidence built entirely from whatever happens to be in view. Holmes's world is arranged so that what he sees really is all there is; the frame contains every relevant fact. Your world is not so arranged. Your frame is a tiny lit circle in an enormous dark room, and WYSIATI is the conviction that the room ends where the light does.

The autopsy's verdict

So we close Part I with the body fully opened, and the finding entered into the record. Sherlock Holmes did not practise a science. He practised abduction — the fertile, unsafe leap — inside a universe engineered by his author to punish that leap never; and the thrill we take from him is manufactured by the same suite of cognitive bugs that, uncorrected, produce quacks, conspiracies, wrongful convictions, and the quiet daily catastrophe of confident people being wrong. To try to be Holmes, sincerely, at home, with no author to catch you — is to hand yourself over to every bias in this chapter and call it brilliance. It is the single most dangerous piece of self-improvement advice ever smuggled inside a great story.

Which would be a bleak place to end, if it were the end. It is not even close. Because here is the turn the whole book has been walking toward: every bug in this chapter has a fix. Not a cure — you cannot uninstall three hundred million years of jumpy pattern-matching — but a discipline, an external scaffold, a way of thinking built specifically to catch the mind in the act. There is a mathematics that forces you to respect the base rate. There is a forensic science that counts the men in the ditch. There are professional reasoners who beat the biases by keeping score. The dream Doyle faked is real. It was simply never going to be a lone genius in an armchair. It was going to be a method — and it is time to go and build it.