Anchoring: The Number You Saw Five Minutes Ago Is Controlling Your Decision Right Now

In 1974, Daniel Kahneman and Amos Tversky spun a rigged roulette wheel in front of a group of participants. The wheel was designed to land on either 10 or 65 — nothing else. After the wheel stopped, the researchers asked: "Is the percentage of African nations in the United Nations higher or lower than that number?" Then: "What do you think the actual percentage is?"

The wheel was obviously random. The participants knew it was random. Nobody believed a roulette wheel had information about African UN membership. And yet — the group that saw 65 estimated an average of 45%. The group that saw 10 estimated 25%. A random number, from a device everyone knew was irrelevant, shifted people's estimates by twenty percentage points.

This is anchoring — the cognitive bias where the first piece of numerical information you encounter disproportionately influences your subsequent judgements. It's not a subtle effect detectable only in labs. It's one of the most powerful and reliable biases in all of psychology, it operates even when you know it's happening, and it's shaping a decision you're going to make today.

Why Your Brain Grabs the First Number

Anchoring isn't laziness. It's architecture.

When your brain is asked to estimate an unknown quantity — "What should this house cost? What's a fair salary? How long will this project take?" — it needs a starting point. Estimation from zero is cognitively expensive. It requires holding multiple variables in working memory, weighing them against each other, and constructing a number from scratch. Your brain doesn't want to do this. It wants a shortcut.

The anchor provides one. Your brain takes the first number available, treats it as a starting point, and adjusts from there. The problem is that the adjustment is almost always insufficient. You move away from the anchor, but not far enough. The anchor's gravitational pull distorts the final estimate, pulling it toward itself regardless of its relevance.

Psychologists call this anchoring and insufficient adjustment. The "insufficient" part is the critical word. Your brain does try to correct for the anchor. It just never corrects enough. The roulette wheel participants didn't estimate that 65% of African nations were in the UN. They adjusted downward — but only to 45%. The anchor didn't replace their thinking. It contaminated it.

The Neural Mechanism

Neuroscientist Todd Mussweiler used fMRI imaging to study what happens in the brain during anchoring. When participants received an anchor before making an estimate, their brains showed selective activation of anchor-consistent information. The anchor didn't just sit passively in working memory. It actively primed a network of related associations.

Give someone a high anchor for the price of a car, and their brain starts accessing memories of expensive cars — features, brands, experiences. Give them a low anchor, and the brain accesses economy cars. The anchor changes what information becomes mentally accessible, which changes the raw material the estimate is built from. You're not just starting from a different number. You're thinking with a different dataset.

The Courtroom, the Hospital, and the Salary Negotiation

If anchoring only affected roulette wheel experiments, it would be a curiosity. It affects every domain where numbers are discussed — and the stakes are often life-altering.

Judges Who Can't Resist Dice

In 2006, Birte Englich, Thomas Mussweiler, and Fritz Strack conducted a study that should have restructured every sentencing system on Earth. They gave experienced German judges a criminal case file and asked them to determine an appropriate sentence. Before sentencing, each judge rolled a pair of dice — which were loaded to produce either a 3 or a 9.

Judges who rolled a 9 gave sentences averaging 8 months. Judges who rolled a 3 gave sentences averaging 5 months. These weren't law students. They were experienced judges with years on the bench. They knew the dice were irrelevant. They rolled them themselves. And the random number shifted their professional judgement by 60%.

Englich's subsequent research found that anchoring effects in sentencing persist even when the anchor comes from the prosecutor's demand (which judges know is strategically inflated), from a journalist's suggestion (which has no legal standing), or from a computer-generated random number that judges were explicitly told to ignore.

The anchor doesn't need to be credible. It doesn't need to be relevant. It doesn't even need to be believed. It just needs to be the first number in the room.

The Doctor's Diagnosis

Dr. Lisa Chen, an emergency physician in Toronto, describes a phenomenon she sees weekly. A patient arrives by ambulance. The paramedic reports: "Sixty-seven-year-old male, possible MI, troponin pending." The word "MI" — myocardial infarction, a heart attack — is the anchor. From that moment, every doctor who sees the patient is anchored on cardiac diagnosis. Tests are ordered for cardiac causes. The history is taken through a cardiac lens. Questions about non-cardiac explanations are asked with less rigour.

When the troponin comes back normal, there's a beat of cognitive friction. The anchor resists updating. Some doctors order a second troponin, not because the clinical picture warrants it, but because the anchor ("possible MI") is so strong that a normal result feels like an error rather than an answer.

Diagnostic anchoring kills patients. A 2005 study in the Journal of General Internal Medicine found that anchoring was a contributing factor in approximately 26% of diagnostic errors in internal medicine. The first hypothesis — the anchor — wasn't just the starting point. It became the gravitational centre that subsequent evidence orbited around instead of displacing.

The Salary You Never Recovered From

Nadia Okafor graduated from law school in 2015 and accepted her first associate position at a mid-sized firm in Atlanta. The offer was $72,000. She was thrilled. It was more money than anyone in her family had ever earned.

What Nadia didn't know: the market rate for her position, at her firm's tier, in that city, was $85,000-$95,000. She didn't research it. She didn't negotiate. The number felt generous relative to her previous income (zero, as a student), and that previous income was her anchor.

Nine years later, Nadia calculated that the $72,000 anchor — compounded through percentage-based raises, bonuses pegged to base salary, and a subsequent job offer benchmarked against her "current compensation" — had cost her approximately $340,000 in cumulative lost earnings. Not because she was underpaid every year. Because the anchor from year one propagated through every subsequent number in her career.

Your first salary isn't just your first salary. It's the anchor for every salary that follows. Every percentage raise is calculated from it. Every new employer asks for your "current compensation" and adjusts from there. The anchor from your first negotiation — or your failure to negotiate — compounds for decades.

Anchoring is the only cognitive bias that gets more expensive over time. Every other bias produces a one-time error. Anchoring produces an error that becomes the baseline for the next error, which becomes the baseline for the next. The chain never self-corrects because each link looks reasonable relative to the one before it.

Why Knowing About Anchoring Doesn't Fix It

Here's the uncomfortable finding: awareness doesn't help.

In study after study, participants who are told about anchoring, who are warned that an anchor is present, who are explicitly instructed to ignore it — still anchor. The effect is reduced slightly, but never eliminated. Timothy Wilson and colleagues demonstrated this at the University of Virginia: even when people were offered a cash incentive to resist anchoring, the bias persisted.

This separates anchoring from most cognitive biases. Motivated reasoning can be mitigated by reducing the emotional stakes. Status quo bias can be overcome by redesigning defaults. Anchoring resists every intervention except one: providing a competing anchor.

The Counter-Anchor

The most effective defence against an anchor isn't ignoring it — your brain can't do that. It's introducing a second anchor that pulls in the opposite direction.

In negotiation, this is well understood. If a seller opens with $500,000 for a property, your brain is now anchored at $500,000. Adjusting downward from there, you'll land higher than if no anchor existed. But if you immediately counter with $350,000 — an aggressive counter-anchor — you've created a second gravitational centre. The negotiation now orbits between two anchors rather than being pulled by one.

This is why pre-suasion works: the person who sets the first number controls the centre of gravity. And it's why experienced negotiators never let the other party anchor first without immediately establishing a counter-anchor. The worst position in any negotiation isn't having a bad number. It's having only one number.

Anchoring Yourself on Purpose

Anchoring isn't only a threat. Once you understand the mechanism, you can use it — on yourself, ethically, to produce better outcomes.

The Project Estimation Trick

Software projects are notoriously under-estimated. The reason is anchoring: the first estimate — usually the optimistic one generated during the planning phase — becomes the anchor for every subsequent revision. Even when evidence mounts that the project will take longer, adjustments are insufficient because they're measured from the original anchor.

The fix: estimate the worst case first. Before asking "how long will this take?", ask "what's the longest this could possibly take if everything goes wrong?" If the answer is twelve months, that number becomes the anchor. Now when you estimate the realistic case, you'll adjust downward from twelve — landing around seven or eight. If you'd estimated optimistically first ("probably four months"), you'd have anchored there and adjusted upward to five or six. Same project, same evidence, different anchor, different estimate — and the pessimistic anchor produces the more accurate one.

The Purchase Pause

Before any significant purchase, look up the cheapest reasonable version of what you're buying. Not to buy it — to anchor yourself. If you're buying a laptop, look at the $400 options before looking at the $2,000 options. If you're buying furniture, visit IKEA before visiting the design store. The cheap option becomes your anchor, and the expensive option must now justify its distance from that anchor rather than being evaluated in isolation.

Retailers understand this perfectly — in reverse. Luxury stores display the most expensive item first. Wine lists put the $300 bottle at the top. Car dealers show you the fully loaded model before the base model. Each is an anchor designed to make everything after it feel reasonable by comparison. The environment is the persuasion. Your defence is to set your own anchor before someone else sets one for you.

The Negotiation Non-Negotiable

If you remember one thing from this essay, remember this: never enter a negotiation without your own number already decided. Not a range. A number. Written down. Before the conversation begins.

Your pre-committed number is your anchor. Without it, the other party's first offer becomes the anchor by default — and your brain will adjust insufficiently from their number instead of yours. Every dollar, every percentage point, every deadline in the negotiation will orbit around whoever planted the first number. Make sure it's yours.

Kahneman spent fifty years studying cognitive biases. He won the Nobel Prize for it. In interviews near the end of his career, he was asked which bias he personally found hardest to overcome.

His answer: anchoring.

Not because it's the strongest bias in the lab — other biases produce larger effects in controlled settings. Because it's the one that resists every tool he'd developed for combating biases. You can't reason your way out of it. You can't introspect your way out of it. You can't even pay your way out of it. The first number lands, your brain grabs it, and every subsequent number is computed relative to it — silently, automatically, with the same confidence your brain brings to every other judgement it produces.

The roulette wheel in Kahneman and Tversky's lab was rigged and random. Every participant knew it. Every participant anchored anyway. The number they saw for three seconds shaped an estimate they gave thirty seconds later about a topic the number had nothing to do with.

Right now, somewhere in your recent memory, there's a number. A price you saw, a salary someone mentioned, a statistic from a headline, a figure on an invoice. You don't remember registering it. Your brain registered it anyway. And the next estimate you make — the next price you evaluate, the next offer you consider, the next deadline you set — will orbit that number like a planet orbits a star.

You won't feel the pull. That's the whole point.