The Invented ReasonsWhy you know your own motives less well than you think

Picture a table in a department store. Four pairs of stockings lie side by side. You are asked which is the best. You pick them up, test the fabric, decide. Then someone asks you why.

You will answer something. The knit, perhaps. The elasticity. How the material feels between your fingers. You will not have to think long — the answer is simply there.

All four pairs were identical.

The place on the table decided, not the fabric

Richard Nisbett and Timothy Wilson described this experiment in 1977. What actually drove the choice was not the material but the position: the pair on the far right was chosen roughly four times as often as the one on the far left.

Nobody named position as a reason. And when the researchers asked whether the place on the table might have played a part, most reacted with confusion or dismissal. The idea seemed not so much wrong as absurd.

The title of their paper is the punchline in itself: “Telling More Than We Can Know”.

Now, a table with four identical pairs of stockings is a decidedly artificial situation. That objection was later raised by the very researchers who took the topic further — they call the task “strange and contrived”. The objection is fair. Which makes what came next more interesting.

The card trick with the faces

A research group around Petter Johansson and Lars Hall in Lund rebuilt the experiment — with one decisive addition.

Participants were shown pairs of portrait photographs and asked which face they found more attractive. Fifteen pairs, four seconds of deliberation each. On some trials they were then asked to justify their choice. So far, harmless.

The addition was a sleight of hand. Using a double card, the experimenters covertly slipped in the photograph that had not been chosen. Participants were suddenly holding the face they had just rejected — and were asked to explain why they had picked it.

Across all conditions, no more than 26 per cent of the manipulated trials were caught. Roughly three quarters went unnoticed. And then participants gave detailed reasons for a choice they had never made.

The invented reasons could not be told apart

This is where it gets genuinely interesting to me, because at this point the researchers did something you rarely see: they tried to refute their own finding.

Their expectation was reasonable. Someone asked to justify a choice they never made ought to give themselves away. So they took 80 participants, recorded every justification and measured it along three dimensions: less feeling in the voice, less detail in the content, less certainty in the tone.

They found next to nothing.

• No differences in phrasing that expresses uncertainty.
• No differences in pauses — although you would assume an explanation invented on the spot needs thinking time.
• No differences in laughter, which otherwise betrays nervousness or surprise.
• No differences in the length of the answers or the number of rare words.
• In both cases, around half of what was said was not about the choice at all.

One single measure showed a difference. In the longer justifications, genuine choices contained proportionally more nouns than the swapped ones — 0.089 against 0.078. Statistically meaningful, but tiny.

That one crack belongs in the picture, otherwise this text would be too smooth. But it changes nothing about the whole: a justification for something you never decided sounds practically like a real one. Including to the person saying it.

It does not stop at photographs

The obvious objection: faces on cards, four seconds, no consequences. Who cares.

So the researchers went to a supermarket. At a tasting stand, shoppers could sample two jams or smell two teas and pick one. They were then allowed to taste “their” choice again — from a jar whose contents had been secretly swapped. No more than a third of the manipulations were caught. Even with pairings that could hardly differ more — cinnamon-apple against bitter grapefruit, mango against Pernod — no more than half noticed.

And then it got serious. During the 2010 Swedish general election the group presented 162 people with a political questionnaire: twelve statements on precisely the issues dividing the two blocs. Afterwards a sleight of hand slipped them an answer sheet on which their positions had been reversed, and they were asked to explain their stance.

22 per cent of the altered replies were noticed. 92 per cent of participants accepted the manipulated score as their own position and justified it. Almost half were subsequently willing to consider switching between the political blocs — in ordinary polls that share sits at around one in ten.

That is where this stops being a laboratory curiosity. It was no longer about stockings. It was about what people took to be their convictions.

What is expressly not being claimed here

Without this section the article would be exactly the smooth story it warns about.

• Failing to notice is not the same as inventing. The experiments reliably show that a swap goes undetected and that reasons follow. Whether an explanation is actually *generated* in the mind, or whether a reason that was lying around anyway gets voiced, is not settled by this.
• The term “confabulation” is delicate — the researchers point this out themselves. It comes from clinical practice, where it describes statements that visibly fail to match reality. Here the opposite holds: the statements are entirely plausible. So the term fits only with care.
• Playing along does not show up. That someone noticed the discrepancy and politely let it pass can never be fully excluded. The researchers therefore asked both during the experiment and in follow-up interviews whether anything had been noticed — but the objection cannot be cleared away entirely.
• The detection rate depends on the gap. The more similar the two options, the less often the swap was caught. With very different tastes, half of them noticed. That is not a footnote but the limit of the effect.
• These were snap decisions. Four seconds for a face. That says nothing yet about a long, carefully weighed decision.

Why this is not a defect

It is tempting to read the finding as an insult. I think that is the wrong reading — for the same reason that missing the gorilla is not a failure.

A creature that had to retrieve the actual chain of causes for every action would not get through a single day. What we get instead is far more practical: a fast, plausible, usually serviceable account of why we are doing what we are doing. That account makes us able to act, to explain ourselves, and to be predictable to others.

The price is in the small print: we cannot tell when the account is wrong. It feels identical either way. Which means the sense of certainty about your own motives is unfortunately no indication that they are correct.

What follows for change

If justifications are unreliable, then “why do I actually do this?” is a surprisingly weak opening question. It produces an answer — reliably, instantly and convincingly. That answer simply has no verified connection to what actually happened.

The sturdier questions aim not at the reason but at the circumstances: when exactly does it happen? Where? What came immediately before? What was different about the times it did not happen? Those are questions about situations, and about situations you can give reliable information. About your own causes, only partly.

And here lies the sober reason why another person contributes something. Not because they know your reasons better — they do not. But because they do not already carry the obvious explanation with them. They keep asking about the when and the where at the point where you would long since have arrived at a satisfying why. Why understanding alone still changes no pattern I have described elsewhere.

The four pairs of stockings lay on that table in 1977. What has emerged since is not that we lie to ourselves. It is something calmer and more useful: we do not know our reasons particularly well — but we know the story we tell about them very well indeed. Telling those two apart is a start.

This article reflects the state of research on self-perception and decision-making and is intended as everyday orientation. The coaching methods described support general wellbeing and holistic balance. They do not constitute a medical diagnosis, promise no cure, and do not replace medical or therapeutic treatment.