The Yes You Never Asked For

A table. A sheet of paper. Three numbers on it: 2, 4, 6.

The person across the table knows these three numbers follow a rule. The rule is simple. Her task is to guess it. To do this, she can name any set of three numbers, and the experimenter will say “yes” if the set follows the rule, and “no” if it doesn’t. Once her confidence is solid, she declares the rule and finds out whether she is right.

The rule that suggests itself almost immediately: the numbers increase by two. Indeed, 2, 4, 6 run exactly that way. The person then checks what already seems like truth to her, and for 8, 10, 12 she says “yes.” For 14, 16, 18 she also says “yes.” For 20, 22, 24 she says “yes” again. A few more sets, and she announces: “Each subsequent number is two greater than the previous one.”

The answer is wrong.

The real rule turns out to be simply “three increasing numbers.” 2, 4, 6 fit it. As do 1, 5, 199. As do 3, 17, 84. No “by two” step exists anywhere, because the person invented a rule for herself and then searched only for what confirmed it. She had every right to ask about 5, 4, 3, and she would have heard “no.” About 1, 2, 3, she would have heard “yes.” And then the “by two” hypothesis would have collapsed instantly. But she didn’t ask. She named only the sets that were supposed to get a “yes.”

The experiment was devised by British psychologist Peter Wason in 1960. Out of 29 participants, only six arrived at the correct rule without first declaring a false one. Most named sets that fit their hypothesis, and not a single one that could refute it. People searched for confirmation of their own rightness.

And found it.

The normal reaction to this result is to mentally distance oneself from those 23 participants. They chased confirmation when refutation lay a single step away, so we wouldn’t fall for it. We would check. We would ask.

But they didn’t notice the apparatus.

In 2011, Hugo Mercier and Dan Sperber proposed a theory that overturns the conventional view of the function of reason. Evolutionarily, thinking was shaped for argumentation, to produce and evaluate reasons in a dispute. To persuade others. To defend one’s own position. To search for grounds for what we already believe.

That’s its working mode.

When a person sits at the table with the sets of numbers, she is not solving an abstract logic puzzle. She already has a hypothesis, and from then on her apparatus operates as it has for hundreds of thousands of years: it searches for arguments in favor of that hypothesis. Confirmation. Proof of her own rightness before an imaginary opponent. And there is no opponent. There is only a sheet of paper and her own head, and the head does what it was built to do.

Here it’s worth pausing and feeling the consequence: when we think in solitude, we are not thinkers. We are lawyers without a court.

The same theory predicts the reverse. If thinking is calibrated for group argumentation, then in a group it should work better. And indeed: when people discuss the task with others, they both produce stronger arguments and are capable of selecting the best among others’ arguments. In discussion, the very same person who was just chasing confirmations alone suddenly turns out to be a capable thinker. She hears a counter-argument and hears it well, because that is precisely what her apparatus is designed for. For a dispute with a real opponent who doesn’t nod along.

Now the picture assembles: alone, we chase confirmations because, evolutionarily, thinking was always a collective act. Thought developed together, in company, and together we filtered out weak arguments. But remove the others, and the mechanism starts spinning idly. Confirmations keep arriving, refutations do not. The hypothesis grows sturdier. One’s own rightness becomes ever more self-evident.

To think independently means to use a tool in a mode it was not designed for. The most difficult of all possible applications.

And now the most uncomfortable part. We are used to thinking that a sufficiently developed intellect overcomes this trap on its own. That a smart person sees their own biases and corrects them. It’s a pleasant thought. And false.

In 2008, three experiments with over 1,200 participants showed that as cognitive ability rises, the tendency toward self-serving bias remains at full strength, and in some cases even intensifies. They justify what they already believe better. Their apparatus is more powerful, and it searches for confirmations just the same, only it does so more elegantly. More complex argumentation creates the illusion of objectivity: if I can build such a coherent chain of reasoning, how could it be wrong?

It can.

Four years later, another study went deeper. It measured bias blindness, the ability to notice distortions in others while leaving one’s own outside attention. The result: this blindness does not weaken with cognitive sophistication. A person with high intelligence is just as certain of her own objectivity as a person with lower intelligence. The only difference is that she can explain longer and more persuasively why her distortion should be regarded as sober analysis.

And then comes the third layer, the sharpest one. In 2017, a group of people were given a numerical data analysis task. When the topic was neutral, like the effectiveness of a skin rash cream, those who were better at math were also better at solving it. But when the same people were given the same numbers in the context of a politically charged topic, polarization did not disappear. It grew. And it grew precisely among those who were best at math. Intelligence did not stop them from fitting the numbers to their own position. It helped them do it without noticing it themselves.

The mind works as an amplifier. It amplifies a correct conclusion and a false one equally well. And the more powerful it is, the harder it is to tell them apart from the inside.

Now let’s set the laboratory aside and look at scale. At what happens when confirmations accumulate over centuries.

Ancient astronomers looked at the sky and recorded the movement of planets. Loops, accelerations, stops. Ptolemy’s system described this movement through circles upon circles: epicycles, deferents, and the equant, a point from which the planet’s motion appeared uniform even though it wasn’t. The construction grew cumbersome, but it worked. It predicted planetary positions with accuracy sufficient for navigation and calendars.

And here is what matters: this is exactly how ancient astronomers understood their task. To describe the true structure of the world meant to ‘save the phenomena’: to produce a mathematical model that matched observations. They were searching for correspondence.

The verification worked flawlessly. Observations accumulated. Epicycles multiplied. Each new refinement confirmed that the system predicted planetary positions ever more accurately. No experiment could uncover the flaw, because the flaw sat in the initial assumption, leaving the calculations impeccable. In the assumption that Earth sits at the center. And this assumption was never checked, because no one even saw it as an assumption. It was part of what “looking at the sky” meant.

Centuries of confirmation on a false foundation. Centuries of brilliant minds adding new circles instead of asking: is the initial condition correct? And not one of them was stupid. Not one was dishonest. They did what we do: they verified inside the frame without noticing the frame itself.

“The real answer lies not at the end of the logical chain, but at its beginning.” We are approaching the point where theory must turn into practice. The practice of self-examination that displaces advice. Because if everything said above is true, then how do you distinguish a thought that has arrived at an uncomfortable conclusion from a thought that has merely defended a comfortable one masterfully?

The answer is paradoxical: you can’t. In the moment of thinking, you cannot know whether you are currently a lawyer or an investigator. The apparatus provides no such signal. It always sounds equally persuasive.

But there is one place where the difference becomes visible. The past. Your own history of conclusions you have drawn. If you go through it honestly, you can see a pattern: whether your thinking has ever led you somewhere you did not intend to go. Whether there were moments when you started from one position and ended with another, and that other displeased you. Whether there were questions to which you found no answer, and accepted that.

If not, if every chain of reasoning has always ended exactly where you intuitively expected, that is a sign that the apparatus is operating in lawyer mode, and investigator mode has never once been switched on.

“Independent thinking is capable of remaining without a ready answer and not feeling weak.”

Here it’s important not to confuse things. There is a temptation, having noticed this mechanism, to adopt a position of eternal opposition. To be always against: against the majority, against the obvious, against the generally accepted. It seems that this way you avoid the confirmation trap: if everyone says “yes,” you say “no,” and there it is, independent thinking.

But that is the same trap. Only the sign has changed.

“Being principled against the majority is the same kind of dependency.” Your conclusion is still determined by an external reference point; you are simply pushing away from it now. The direction of movement is set by the position of those you have decided to differ from, and your own thinking merely follows. It’s a mirror.

The real test is whether you are capable of arriving somewhere no external signpost points to. Including your own previous conviction.

And now, holding this in mind, let’s return to the table with the sets of numbers. To the person who names 8, 10, 12 and gets a “yes.” For 14, 16, 18 she says “yes.” She is comfortable. Her hypothesis is being confirmed. The apparatus is working.

But somewhere nearby sits another person. She also started with the “by two” hypothesis. And she also named a few confirmatory sets. And then, for no obvious reason, perhaps idly, perhaps out of distrust for her own certainty, she names 5, 40, 200. The experimenter looks at her and says: “yes.”

And in that moment, something inside her breaks. The “by two” hypothesis is no longer alive. She has just received an answer she did not want, and that answer turned out to be correct. And the real rule turned out to be broader, simpler, and nothing like what she had imagined.

She did not know she was heading toward this. She simply asked a question that could have produced a “no.”

This is the difference between the two people at the table. Not intelligence. Not education. Not the number of correct answers.

One of them never once heard a “no.”

The other person heard.