What are your earliest memories? Mine are fragments like a super-8 flashback sequence in a Paul Cox movie. Flashes of images such as being carried to the bath with my sister. Another time, getting into the tub in my clothes. Walking with my mother to the local shops on a sunny morning after rain. I would guess they date to around two or three years old. More concrete and sequentially connected memories begin to form around a family trip to Sydney when I was four.
In that respect, my memory-formation seems to be pretty typical of human experience. On the other hand, Salvador Dali claimed to remember the womb, which was almost certainly just another whopper in his long line of bullshit. But why can’t we remember being babies? After all, the brain is in overdrive in those formative years, busily observing and recording. So, why can’t we access that apparently locked-away early brain activity?
For more than a century, the standard explanation has been that infants are not merely poor rememberers but non-rememberers. Their brains are said to be too immature to form durable memories of individual events; the hard drive is supposedly switched off, or at least running in some primitive, pre-operational mode incapable of saving files.
New evidence, however, suggests something different is going on. A study from Yale has examined awake infants between four and 25 months old using functional magnetic resonance imaging while they viewed unfamiliar photographs and subsequently performed a looking-time memory task. The children were not, apparently, enjoying the dubious pleasures of an MRI scanner while their parents watched through the glass and made encouraging noises. Nevertheless, the data indicated that the infant hippocampus is not an inert lump of neural pudding.
Greater activity in the hippocampus during the viewing of previously unseen photographs was related to later memory-based looking behaviour beginning around one year of age, suggesting that the capacity to encode individual memories comes online during infancy.
In plain language: babies do form memory traces. They are recording the data. When infants later looked longer at photographs they had previously encountered, increased hippocampal activity during the original viewing was associated with evidence that they could distinguish the familiar from the unfamiliar.
This makes infantile amnesia rather less mysterious. We have, traditionally, assumed that infants fail to encode specific experiences: the first birthday, the first steps or the arrangement of furniture in the room where one slept. The new research suggests that at least some of those experiences were successfully inscribed, at least from around the end of the first year onwards. What disappears is not the original recording but our ability to retrieve it.
That distinction is important. Encoding is the act of putting something into memory, while retrieval is the business of getting it back out again. The failure is not in the recording studio but in the filing system. Somewhere in the deep hippocampal architecture may lurk an authenticated first-person memory of being one year old, complete with the light through the curtains and the sound of the family moving about the house. It may be there, but we have lost the instructions for finding it.
The availability of encoding mechanisms for episodic memory during a period of human life that is later lost from our autobiographical record implies that postencoding mechanisms, whereby memories from infancy become inaccessible for retrieval, may be more responsible for infantile amnesia.
Why should this happen? As the brain develops, its wiring undergoes enormous change. New neurons are born, existing connections are pruned and hippocampal and cortical networks are repeatedly remodelled. The neural pathways that once provided direct access to an early memory are buried, altered or closed off. The file remains in the archive, but somebody has changed the server’s directory structure without informing the client.
There is some laboratory support for this possibility. Research in rodents suggests that memory engrams formed in the dorsal hippocampus during infancy can persist into adulthood while remaining inaccessible unless the animals receive direct stimulation or reminders. Humans, naturally, cannot be placed in scanners and prompted to remember their first birthdays, so the scientists cannot produce a direct answer. Nor should the findings be overstated: the paper itself notes that some infantile amnesia may still reflect poorer, less complete encoding, and that future research will require richer forms of memory than the present test allows.
There is also a developmental wrinkle. Episodic encoding appeared only in infants older than 12 months, whereas statistical learning has been observed from around three months. Small children can learn the regularity of the world before they can readily bind individual experiences into a consciously retrievable narrative.
Still, the broader picture is clear enough. The infant brain is not simply failing to record the world. It is doing rather more than we gave it credit for and the disappearance of those early memories is probably not because the past was never written down – it is because time subsequently reorganised the library.
