The Lost Key: Why We Can’t Remember the First Years of Life

Try to recall your earliest memory. Most people will arrive at a yard, a kitchen, or some dusty room from when they were three or three and a half. Decades of research show this: in the largest survey of first memories ever conducted, with 6,641 respondents, the average age of the earliest memory was 3.2 years.

In the same survey, however, 38.6 percent of people claimed their first memory was from the age of two or younger, and 893 of them said it was from their first year of life. The authors consider many of those memories fictional: a patchwork of surviving fragments and knowledge drawn from family stories or photographs, and the person who “remembers” them doesn’t know they are fiction. It is possible, then, that your earliest memory is not a memory at all.

Freud and the First Suspect

The phenomenon has had a name since 1905, when Sigmund Freud gave it one in his work on the theory of sexuality. In his view, adults repress their earliest memories because of their unsuitable content. Science long ago set that idea aside: there is no evidence of active repression.

The problem, though, had been noticed before him. At the end of the 19th century, Caroline Miles (1893) and Victor and Catherine Henri (1895) asked adults for their earliest memories, and the answers clustered around the age of three.

Forgetting Starts Later Than We Think

For a long time it was assumed that small children simply cannot form lasting memories. Patricia Bauer of Emory University and Marina Larkina showed that this is not so. They recorded 83 three-year-olds as they talked with their parents about events from recent months, such as a trip to the zoo or a birthday party. When the researchers questioned the children again later, those aged five to seven remembered more than 60 percent of the events, while eight- and nine-year-olds remembered fewer than 40 percent. The authors conclude that adult-style forgetting sets in around the age of seven.

Forgetting, then, is not a void: the memory existed, and then it vanished. Parents who followed the child’s lead in conversation drew richer memories out of their three-year-olds.

The Baby Who Kicks

That babies remember was known even earlier, though not the way we do. Carolyn Rovee-Collier worked for decades with three-month-olds who learned to set a mobile above their crib moving by kicking, because a ribbon was tied to one of their legs. After six to eight days, the babies stopped kicking, so it seemed they had forgotten. But a brief reminder, watching the mobile move on its own, brought back what they had learned almost entirely. The memory had not disappeared; it simply couldn’t be summoned.

The Baby in the Scanner

The real breakthrough came in March 2025, when a team from Yale University published a paper in the journal Science. Tristan Yates and Nick Turk-Browne scanned awake babies with magnetic resonance imaging, which is extraordinarily difficult because children squirm: the team had to run hundreds of sessions, and many scans were unusable because of movement. Twenty-six babies took part, aged 4 to 25 months. Each was shown new photographs of faces, objects and scenes for two seconds apiece. Afterward, they were shown pairs of images, one seen before and one new, and their gaze was measured: children look longer at what they recognize.

The result: the more active the hippocampus was while a baby first looked at an image, the longer she looked at that same image later. The strongest signal was in the rear part of the hippocampus, the same part adults use for episodic memories, and it was more pronounced in babies older than 12 months. This held for 11 of 13 older babies, and not for those under one year. Until then, the prevailing view was that the hippocampus in early childhood is simply too immature to make such memories at all. The study shows it need not be so: from around the first year of life, the hippocampus can record an individual experience.

The limits of the finding are worth knowing. What was measured was recognition of images, not the recollection of events. For babies under a year, the study claims nothing. And no one yet knows whether such memories survive into adulthood. Turk-Browne admits the work has no practical application, except to show that babies of around a year already remember. His team is now testing whether children recognize footage from their own perspective from when they were babies, and the first, preliminary results suggest that the memories last until preschool age and then fade.

Lost, but Not Erased

If babies record, why can’t we read? The strongest clues come from animal experiments. Axel Guskjolen, Sheena Josselyn and Paul Frankland showed in 2018 that infant mice forget fear tied to a place, but that the memory can be brought back: when researchers stimulated the former “memory cells” in the hippocampus with light, the mice recalled it up to three months after training. Alberini and colleagues showed something similar in rats: what is learned at an age corresponding to infantile amnesia remains in a latent form, and a later reminder revives it. Later work even restored the memory permanently.

One hypothesis seeks the cause in the fact that the young hippocampus constantly produces new neurons, which overwrite existing connections: in infant mice, reducing that process slowed forgetting. Turk-Browne suspects a second reason: the hippocampus may not receive the right “search terms,” because the memory was recorded under conditions that don’t match how adults look for it.

What It Means

Early experiences leave a trace even when we don’t remember them: animal studies show that experiences from that period influence later behavior. Turk-Browne is particularly interested in the possibility that such fragments might one day be reactivated in adulthood.

For now, these are mice, rats, and a baby who looks longer at a picture she has already seen. But if the animals are telling the truth, our first years are not lost. We just don’t know where we left the key.