A universal blank at the start of the story
Almost no one has real memories from before age three, and memories from ages three to seven are thin and patchy. Psychologists call this infantile amnesia, and it is remarkably consistent across cultures. It is not that nothing happened or that babies do not learn; infants absorb language, faces, and skills at a staggering rate. What is missing is autobiographical memory, the ability to recall specific episodes as events that happened to you at a particular time and place.
This is puzzling because babies clearly form memories in the short term. They recognize their caregivers, learn routines, and show they remember games from one day to the next. The mystery is why these early episodic memories do not survive into the adult mind. The answer appears to be less about failing to record and more about the brain being rebuilt underneath the recordings.
A brain still under construction
The hippocampus, the structure that binds the who, what, where, and when of an experience into a lasting episodic memory, is one of the last brain regions to mature. In the first years of life it is still wiring itself up, so early events may be encoded in a form too incomplete to be retrieved the way adult memories are. Alongside this, the parts of self-awareness and language that let us frame a memory as something that happened to me are only just coming online.
One leading account, developed by neuroscientists Sheena Josselyn and Paul Frankland, points to neurogenesis, the birth of new neurons. Early childhood sees an explosive rate of new neuron production in the hippocampus. Adding all this new circuitry may remodel existing connections and effectively overwrite or scramble the fragile memory traces already stored there. In rodent studies, slowing infant neurogenesis protected early memories, while boosting it in adults induced forgetting, direct support for the idea that a fast-growing brain trades early memories for learning capacity.
Stored, but out of reach?
The newest work suggests the early years may not be as erased as they feel. In a 2025 study published in Science, Yale researchers led by Tristan Yates and Nick Turk-Browne scanned the brains of awake infants and found that the hippocampus was already actively encoding specific memories, especially after about twelve months of age. Infants whose hippocampus responded more strongly to a new image were better at recognizing it later, showing the encoding machinery works earlier than long assumed.
Complementary animal research has shown that memories formed in infancy can leave a physical trace, an engram, that seems lost but can be artificially reactivated, bringing a forgotten memory back. Together these findings shift the picture from a recording problem to a retrieval problem: the memories may be laid down and physically present, but the adult brain lacks the right cues or matching state to reach them. Why they become inaccessible, and whether any could ever be recovered, is now an open and active question rather than a closed case.