The Mathematics of the Frightened Primate

You are standing in the cereal aisle, or its equivalent—a doctor’s letter in your hand, two job offers open in a browser, a person across a table who has just asked you something enormous—and the machinery in your chest has seized. Some animal older than language has grabbed the wheel, and it does not want a good decision so much as it wants the ache of deciding to stop. This is the human predicament in miniature: a creature shaped to outrun leopards, asked instead to weigh a pension it cannot picture and a risk it cannot see. We are frightened primates doing arithmetic with our feelings. And most of what we call the stress of a modern life is just the friction of that mismatch—an old mind grinding against the machinery of modern choice.

Here is the good news, and it is better than it sounds. Across the last century, in rooms full of people who were mostly not thinking about your happiness, a handful of mathematical ideas were worked out that do precisely this part of the job for you. Decision theory, game theory, network theory, the mathematics of when to stop—these are not spreadsheets for the soul, and the aim is not to shrink a life to numbers. They are lenses. Each one takes a particular kind of hard choice and dissolves the part of it a formula handles better than a gut ever could: not the whole decision, only the mechanical core, the part that keeps you awake doing bad sums in the dark. You can offload it—the way you offload long division to a calculator—and keep your scarce attention for the part that genuinely needs a human. What follows is a small kit. Four lenses, and where to aim them.

One Number for the Knot

Start with the oldest lens and the plainest. Most dread is a knot of two questions tangled into one: how likely is this, and how much would it matter? The gut feels them fused, a single hot lump of maybe. Pull them apart and multiply, and the lump becomes a figure you can look at. Expected value is nothing grander than that—the probability of an outcome times what the outcome is worth, added up across the ways things could fall. A one-in-ten shot at a thousand pounds is worth a hundred; a near-certain loss of a small, recoverable sum is worth almost none of the fear you are paying it. Write the two numbers down and the knot loosens, because you were never really afraid of the outcome. You were afraid of holding both halves of it at once. The paper holds them, so you don’t have to.

Expected value tells you what a bet is worth; it says nothing about how to hold a belief that keeps shifting under new evidence. For that there is a second lens, named for a Presbyterian minister whose one great idea was read to the Royal Society in 1763, two years after he had died. Bayes’ theorem is a rule for updating. You begin with a prior—your honest estimate before the news arrives—and each new fact nudges it toward a posterior, a revised estimate, by an amount that depends on how surprising the fact would be if you were wrong. The discipline it teaches is almost a spiritual one: believe things by degrees. Most of us lurch between certainties—sure she’ll call, then sure she won’t—when the sane motion is a dial, not a switch. A careful mind is rarely at zero or at one. It sits at seventy percent, and stays willing to move.

One error the updating rule exposes so reliably that it has earned its own name. In 1978 a group of researchers put a question to sixty doctors and students at Harvard’s teaching hospitals: if a disease afflicts one person in a thousand, and the test for it carries a five percent false-positive rate, how worried should someone be who tests positive? Nearly half answered ninety-five percent. The true answer is about two percent—because among a thousand people you turn up one real case and some fifty false alarms, and one in fifty-one is nothing like nineteen in twenty. The doctors had ignored the base rate, the sheer rarity of the disease, letting the vivid positive drown out the quiet background number. Base-rate neglect may be the commonest mistake in all of human judgment, and its antidote is a single reflex: before you react to the striking new signal, ask how common the thing was to begin with.

The Games You Are Already In

The first lens is for choices you make alone, against an indifferent world. The second is for the harder kind—choices knotted together with other people who are choosing too, each of them reacting to you as you react to them. This is game theory, and its central idea arrives deceptively calm. In any such tangle there tends to be a Nash equilibrium—named for John Nash, who defined it in 1950, still in his early twenties—a set of choices from which no one can do better by moving alone. It is where the game comes to rest. And the field’s most disquieting discovery is that the resting place can be worse for everyone than somewhere they might have reached together. The equilibrium is stable. It is not the same thing as good.

The purest illustration was devised at the RAND Corporation in 1950 by Merrill Flood and Melvin Dresher, and dressed in its unforgettable costume by the mathematician Albert Tucker: two prisoners, questioned in separate rooms, each offered a deal to inform on the other. Stay mutually silent and both get light sentences; betray mutually and both get heavy ones. But whatever your partner does, betraying shaves a year off your own term—so both of you, reasoning flawlessly, betray, and both end up in the cell marked worst-for-the-pair. That cell is the equilibrium. The prisoner’s dilemma is the little engine that explains why rational people build irrational worlds. Enlarge it to the size of a village and you have Garrett Hardin’s tragedy of the commons, from a much-cited 1968 essay: a shared pasture where each herdsman gains by adding one more cow and all of them lose when the grass is gone. Overfishing, gridlock, a warming sky—the same small betrayal, scaled up until it has a body count.

So is cooperation merely a story that rational animals tell to comfort themselves? Around 1980, the political scientist Robert Axelrod set out to test it. He invited game theorists to submit strategies for playing the prisoner’s dilemma not once but hundreds of times over, then pitted them against one another in a computer tournament. The winner, sent in by the mathematician Anatol Rapoport, was almost insultingly plain: tit for tat. Cooperate on the first move; after that, simply do whatever your opponent did last. It was nice—never the first to betray—and retaliatory, and forgiving, and clear, so that any rival could read its intentions at a glance. When Axelrod ran the whole tournament again, with everyone now gunning for the champion, tit for tat won a second time. The finding landed like a revelation: cooperation is not a moral luxury but a winning strategy, and it prevails whenever the future is long enough to matter.

Name the game you are in, and you are already half out of its trap.

There is a gentler province of the same theory, and it belongs to Thomas Schelling, who took a Nobel for it in 2005. Try his puzzle. You have to meet someone in New York tomorrow, but you cannot reach them to arrange it—no place fixed, no hour agreed. Where do you go, and when? Remarkably, a great many people give the same reply: the clock at Grand Central, at noon—not because a rule says so, but because each is trying to land where the other will think to look. Schelling called these focal points: answers we converge on without exchanging a word, because they simply shine a little brighter than the rest. Almost every unspoken agreement runs on them—which side of the pavement to take, where a border falls, what counts as a fair share. Coordination is less about talking than about guessing where the other mind will go.

Where to Stand

Pull back from the pair to the whole web of people, and a third lens sharpens. In 1973 a young sociologist named Mark Granovetter published a paper—already rejected once—with a title that reads like a paradox: the strength of weak ties. He had looked at how people truly find jobs, and the result cut against intuition. Most found work not through close friends but through acquaintances, the people they saw only occasionally. The reason is structural, almost geometric. Your close friends all know one another and know what you know; their news is already your news. A weak tie, by contrast, reaches into a different cluster altogether—a separate pool of rumor, openings and ideas your inner circle cannot see. The loose acquaintance is a bridge to elsewhere. The people likeliest to change your life are rarely the ones closest to it.

Networks have a shape, and the shape obeys rules worth knowing. A few nodes—the hubs—sit astride far more paths than the rest, and influence, information and contagion pour through them; this is roughly what centrality measures, and why a rumor started beside a hub outruns one started at the rim. The webs we live inside are also strangely shallow: Stanley Milgram’s 1967 experiment, passing packets across America from hand to hand, found that the chains which arrived had crossed only a handful of people—the origin of that worn phrase, six degrees of separation. The practical lesson is not an equation but a stance. To make an idea travel, hand it to a hub; to hear news the herd has not heard, tend your weak ties. Where you stand in the web decides what can reach you—and, unlike your genes or your childhood, you are usually free to move.

Look, Then Leap

The final lens is for a torment the others miss: not which option to choose, but when to stop looking. Picture flat-hunting in a market so hot each place is taken by evening, with no going back. Look too little and you pounce on the first thing; look too long and the good ones vanish while you dither. Mathematics offers a precise, faintly eerie answer, known as the secretary problem. Spend the first thirty-seven percent of your search committing to nothing—only watching, learning what good even looks like—and then take the next option that beats everything seen so far. The fraction is no accident: it is one divided by the number e, 0.368 and onward, dropping out of the equations like a fixed star. Obey the rule and you secure the best candidate about thirty-seven percent of the time—far better than any quantity of agonized hesitation, and it scarcely matters whether you are choosing among ten or ten thousand.

The deeper permission comes from Herbert Simon, who won his own Nobel in 1978 for working out how real minds decide under real limits. Simon drew a line between two kinds of chooser. The maximizer wants the best and will ransack the world to be certain of it; the satisficer sets a bar—good enough, along the dimensions that matter—and takes the first option to clear it. His coinage, satisficing, sounds like a synonym for settling. It is nearer to sanity. A world of too many options is not a gift but a tax; each unexamined alternative levies a small toll on your attention, and the hunt for the perfect coat or contractor or sentence can cost more than the difference it buys. A good stopping rule is not laziness—it is the mathematically correct way to spend a finite mind. It is the maximizer, forever chasing the last one percent, who is quietly being irrational.

The Part That Needs a Human

Notice what the four lenses share. Each marks a spot where the gut and the arithmetic disagree—and where, on the narrow question actually asked, the arithmetic is simply right. The gut anchors: Kahneman and Tversky spun a rigged wheel of fortune, let it halt on a meaningless number, and watched that number tug people’s later estimates toward it, as though a random ten or sixty-five knew anything at all. The gut honors sunk costs, pouring good money and good years after bad because it cannot stand to waste what is already spent. The gut, as we saw, forgets the base rate. These are not stupidities; they are the reflexes of a mind that had to be fast to survive, and they misfire in precisely the slow, abstract, invisible choices that a modern life is built from. The math is a corrective lens for a specific astigmatism of the mind.

“All models are wrong, but some are useful.”—George E. P. Box

But a lens is not an eye, and here is the border you must not cross. None of this can tell you what to want. Expected value needs you to say what a thing is worth to you; Bayes needs a prior you had to feel your way toward; the stopping rule cannot tell you what you are even looking for. The mathematics does the mechanical middle—the probability, the equilibrium, the fraction—and hands the two ends back, because the ends are made of values, and a value is not a computation. That is the entire art of it. Not to live by spreadsheet, but to let the formula carry what was only ever arithmetic, so that the frightened primate, its hands suddenly free, can spend itself on what was never arithmetic at all—the wanting, the meaning, the face across the table. The number was never there to make the choice for you. It was there to quiet the noise, so that you could.