Do you clearly remember your first day at school? The exact taste of a dish your grandmother used to make? A precise conversation you had ten years ago? If you answer yes with confidence, there is a good chance you are wrong — at least in part. Not because you have a bad memory, but because no one has a good memory in the sense we usually mean.

Human memory is not a recording. It is a reconstruction.

The illusion of faithful memory

We often imagine memory as a library: memories are stored on shelves, waiting to be retrieved. The metaphor is appealing, but deeply inaccurate. Every time you recall a memory, you do not read it — you reconstruct it from fragments, inferences, current beliefs and outside suggestions.

This is exactly what American psychologist Elizabeth Loftus has spent more than fifty years demonstrating. Her experiments, conducted from the 1970s onward, transformed our understanding of memory — and, incidentally, the way courts handle eyewitness testimony.

In one of her best-known experiments, Loftus showed participants a series of slides depicting a car accident. Then she asked them a seemingly harmless question: “How fast were the cars going when they smashed into each other?” — or “when they hit each other?” — or “when they collided?”. Only one verb changed. And yet speed estimates varied considerably depending on the word used. Participants questioned with “smashed” also reported, a week later, having seen broken glass — although there was none in the slides. A simple verbal suggestion had created a false visual memory.

You can remember something that never happened

Even more disturbing: it is possible to implant an entirely fictional memory in an adult's mind. Loftus demonstrated this with the so-called “lost in the mall” experiment. Volunteers read four short accounts of childhood events, provided by a relative. Three accounts were true. One was entirely invented by the researchers: the story of the child getting lost in a department store before being rescued by a stranger. The result: about 25% of participants not only accepted the fictional memory as real, but enriched it with personal details — what they were wearing, the fear they felt, the face of the person who had helped them.

These participants were not lying. They remembered.

The Mandela effect, or when millions of people share the same false memory

There are cases where a false memory does not affect an isolated individual, but spreads collectively. This phenomenon is called the Mandela effect — a name that comes from a belief shared by many people: that Nelson Mandela died in prison in the 1980s. In reality, he was released in 1990 after twenty-seven years of imprisonment, received the Nobel Peace Prize in 1993, served as president of South Africa from 1994 to 1999, and died on December 5, 2013 in Johannesburg. Nothing obscure. And yet thousands of internet users swore they had precise memories of a televised funeral, memorial speeches and a grieving widow.

Other examples have become famous. Many people are convinced the cartoon is called Looney Toons — with an “s” and two “o”s — even though it has always been called Looney Tunes since its creation in 1930. Or that the Monopoly character Uncle Pennybags wears a monocle: he never has. Or that Darth Vader's line in The Empire Strikes Back (1980) is “Luke, I am your father.” The exact line in the film is: “No. I am your father.”

In 2022, a study by Prasad and Bainbridge measured this phenomenon scientifically by asking participants to draw famous brand logos from memory. Errors were frequent — and above all systematic, often shared by people with no connection to one another. Proof that these are not random confusions, but reconstructions guided by common cognitive biases.

Confabulation: the honest lie

Neuropsychologists have a word for the brain's ability to fill gaps in memory with inventions: confabulation. The term comes from clinical neurology — it is often observed in patients with amnesia or certain brain lesions — but the mechanism is universal, and all of us are subject to it to varying degrees.

Confabulation is not a lie. The person who confabulates sincerely believes what they are saying. Their brain has simply decided, in good faith, to fill the gaps. This behavior may have evolutionary value: a brain unable to function despite missing information would quickly be paralyzed. The narrative continuity we call “our life” would be impossible without this ability to fill in, interpolate and reconstruct.

The problem arises when we confuse this reconstruction with objective truth.

What this changes, concretely

The implications are far from purely theoretical. Loftus's work helped reform judicial practices in several countries, particularly around eyewitness testimony, long considered the strongest form of evidence in a trial. Innocent people have been convicted on the basis of sincere but inaccurate memories. In France and elsewhere, the psychology of testimony is now included in legal and police training.

On a more intimate scale, it invites us to reconsider arguments that go in circles because one person says “you didn't say that” — and the other replies “yes, I said it exactly like that”. It is quite likely that both parties are right from their own point of view, and wrong from the point of view of the facts. Memory is not an arbiter. It is a narrator.

An imperfect memory — and perhaps that is a good thing

It would be tempting to conclude that memory is defective, even dangerous. But we can also see it differently: it is alive. It adapts. It integrates what you have learned since, what you feel today, what others have told you. A memory is not a photograph — it is a letter your past writes to your present, while allowing itself a few liberties.

What we call “our story” may not be exactly what happened. It is the account we build from what happened. And that account, however imperfect, is deeply, irreducibly ours.