The fading memories of youth
The mystery of “infantile amnesia” suggests memory works differently in the developing brain
Toddlers like 18-month-old Hilda struggle to remember events in context, such as where a toy is hidden, for more than a few months. New research suggests such memory lapses play an important role in brain development. Stefanie Loos
A version of this story appeared in Science, Vol 383, Issue 6688.Download PDF
You might think you remember taking a trip to Disneyland when you were 18 months old, or that time you had chickenpox when you were 2—but you almost certainly don’t. However real they may seem, your earliest treasured memories were probably implanted by seeing photos or hearing your parents’ stories about waiting in line for the spinning teacups. Recalling those manufactured memories again and again consolidated them in your brain, making them as vivid as your last summer vacation.
People generally remember nothing from before age 3, and children’s memory abilities don’t fully mature until about age 7. “It’s a paradox in a sense,” says neuroscientist Flavio Donato of the University of Basel. “In the moment that the brain is learning at a rate it will never show again during the whole lifetime, those memories seem not to stick in the brain.”
For many years, researchers assumed babies’ brains are simply not mature enough to form lasting memories. Theories have abounded as to whether this is a biological immaturity or something more psychological, such as a lack of a sense of oneself as an individual or the ability to use language. Sigmund Freud, however, believed infants do form memories, but the brain suppresses them so we forget the psychosexual experience of birth. He called the process “infantile amnesia.”
New research is beginning to suggest Freud was right about the forgetting, if not about its purpose. It appears the brain actually can create memories before age 3—although perhaps in a different way from adult memories—and those memories may persist into adulthood. But we can’t consciously access them.
No one is sure why infantile amnesia exists, but studies have shown that many other mammals also experience it, suggesting it’s not linked to language or self-awareness. Instead, this forgetting probably serves some evolutionary purpose, whether that’s helping young brains learn how to attach the proper importance to events or developing a framework for the memory systems they will use throughout life.
“We’ve kind of just accepted [infantile amnesia] as a fact of life, as an unavoidable consequence of brain development,” whereas in truth it might be essential, says neuroscientist Tomás Ryan at Trinity College Dublin. Whatever it’s doing, he says, “it’s going to be something that transcends most of the mammalian kingdom.”
To figure out how this natural process works, researchers are bringing toddlers into the lab for memory tests and manipulating the memories of rodents with modern tools such as optogenetics, which can selectively activate the neurons that encode a particular memory. Such experiments, they hope, could be the key to understanding how early memories are forgotten, how the traces of those early memories might shape our later lives, how factors such as infections and stress in early life affect our memory capacity throughout life, and whether inaccessible memories can be reactivated.
Sarah Power’s lab at the Max Planck Institute for Human Development is a colorful place with magical jungles, deserts, and seascapes projected onto the walls of a small room. Children between 18 months and 24 months of age toddle between boxes scattered around the room, trying to remember which box contains the plush toy they saw the last time they were in the jungle room or the desert room.
Power is running the first prospective study that will measure how children’s ability to remember information develops over time. Numerous retrospective studies have looked at how people recall their earliest memories, but these memories can be heavily influenced by factors including culture or parents’ storytelling. On top of that, some research suggests children’s ability to put a date on their memories develops at a different time than the ability to remember, making it hard to pinpoint a “first” memory.
So, Power plans to follow the 360 children in the study for 6 months—and potentially much longer. She’ll watch how their memory capabilities develop through childhood and measure their brain activity with electroencephalography (EEG). The children’s parents fill out long questionnaires on factors that could influence brain development, such as the kind of nursery the child attends, their language abilities, and whether the child’s mother contracted COVID-19 while pregnant.
The main goal, Power says, is to figure out exactly when the developing brain switches on the ability to form accessible long-term memories. “It’s really hard to progress to ask other questions if we don’t know exactly when it happens,” she says. Her early data indicate it’s at about 20 months. Children that age who learned to associate a toy with a certain location in each room can remember the information for up to 6 months, whereas younger children only remember it for about 1 month.
Power’s study is a more child-friendly version of the lab experiments used to test memory in rodents. Adult mice and rats can quickly learn to associate a certain cue, such as a colored room, with a small shock delivered to their feet. For the rest of their lives, they will freeze in place when given the cue. But a baby rodent won’t remember the cue for more than a day or two, no matter how many times it gets shocked.
Intriguingly, infantile amnesia seems to affect only certain kinds of memories, particularly the ones known as contextual memories, which involve connecting cues such as the layout of an environment with events that happen there. In humans, the forgotten memories include episodic memories: conscious recollections of where and when a specific event occurred. In contrast, young brains can recall other types of memories just fine, including semantic memories of the meanings of words and motor memories of skills such as how to draw a circle. “There’s probably an underlying neural timetable of development in various bits of the memory system,” says Nora Newcombe, a psychologist at Temple University. Until recently, the simplest explanation has been that the hippocampus, the brain’s key processing and storage site for episodic and contextual memories, either can’t store these memories or can’t form them in the first place.
Yet psychologists have found some evidence that early memories may linger, even if we can’t consciously access them. In one set of experiments, researchers taught babies as young as 2 months old that they could make a mobile over their crib move by kicking their feet. The youngest babies could only remember this for a few days. But 3- and 6-month-old babies remembered to kick their feet if researchers showed them a hint, such as the mobile moving on its own, suggesting the memory was still there but less accessible.
In another study, Newcombe found that 3-year-olds who see a set of images of different animals can’t explicitly remember them 3 months later. But when she blurred the images and brought them slowly into focus, children were faster to identify the animal in images they’d seen months earlier. Newcombe says such findings suggest young children can retain specific information at a subconscious, or implicit, level.
Research with young rats and mice suggests they, too, can access suppressed memories with a little help. In a 2016 study, Cristina Alberini, a neuroscientist at New York University, and her colleagues gave juvenile rats a foot shock when they stepped into a dark compartment within a white box. The young animals learned to stay out of the dangerous compartment, but forgot soon after. Once the animals were older, the researchers found they could jog their memory by showing them the white box and shocking them in a different colored box. Then, when the researchers returned the rats to the original white box, the combination of the two cues made the rodents remember to stay out of its dark compartment.
A lost memory reawakened
Mice can’t access memories they formed as juveniles, such as learning that a cue—a particular enclosure, for example—predicts a shock. But artificially activating the neurons that encoded that memory, known as an engram, reactivates the memory and causes mice to freeze up in response to the cue.
Graphic showing a mouse in a test environment. Panel 1 heading: Memory created, engram labeled. Panel 1 image: A juvenile mouse steps on a dark-colored floor that gives them an electric shock. An illustration at right shows the hippocampus and an abstract node network representing an engram. Panel 2 heading: Juvenile mouse forgets training within 2 days, does not anticipate shock. Panel 2 image: A juvenile mouse stands normally on the dark floor. At right, the engram and memory of electric shock are faded. Panel 3 heading: Adult mouse does not anticipate shock. Panel 3 image: An adult mouse stands normally on the dark floor. At right, the engram and memory are faded. Panel 4 heading: Activating the engram reinstates memory, mouse freezes. Panel 4 image: An adult mouse crouches in a protective stance on the dark floor. A beam of light centered on their head is labeled “light-activated”. At right, the engram and memory are in full color.
The human and rodent studies both suggest infantile memories are not gone, only forgotten. Two years later, Paul Frankland at the Hospital for Sick Children probed the underlying cell connections. His team used baby mice genetically engineered to make a light-sensitive protein in the set of neurons in the hippocampus that fired while the animals were learning to associate a box with a foot shock. Neuroscientists call the set of neurons that link up when a memory is formed an engram, a term first coined by psychologists in the early 1900s to refer to the then-hypothetical physical trace of a memory. A month later, when a mouse had forgotten the memory, the researchers flashed a light in the mouse’s brain through an optical fiber. The light-sensitive protein evidently reactivated the engram: The mouse froze, apparently in anticipation of a shock, even if it wasn’t in the box.
Some skeptics have argued that the foot-shock memories may not be a good proxy for childhood memories in general, because shocks affect emotional centers in the brain and might create a different kind of memory than, say, remembering where to find a toy. But in a paper published in Science Advances in November 2023, Ryan and Power reported that mice trained as juveniles to find an escape hole in a box, a less fear-laden task, also appeared to form lasting engrams that could be reawakened though optogenetics. The two are now collaborating to measure brain activity with EEG as children and rodents navigate similar puzzles. They then plan to study the rodent brains in more detail to understand how the EEG patterns correlate with engram formation and recall. If the researchers can track the animals’ engrams as they look for the hole, Power says, they may be able to reverse engineer how early memories are formed, suppressed, and reactivated in humans.
Why most early memories are forgotten in the first place is still unclear, however. The process is too widespread to have emerged without an important reason, says Rick Richardson, a psychologist at the University of New South Wales Sydney. “Evolution doesn’t work that way,” he says.
It’s possible that suppressing memories allows the brain to put more computing power toward figuring out how the world works while giving the hippocampus time to develop, says child psychologist Tracy Riggins at the University of Maryland. That trade-off might make sense for helpless babies because they can offload some memory-related tasks to caregivers, such as how to navigate or where to find a toy. But it might not be advantageous for “precocial” species such as guinea pigs and degus, two rodent species that are more behaviorally independent at birth. Indeed, work from Frankland’s lab suggests these animals don’t experience infantile amnesia at all.
As children first begin to form accessible long-term memories, they aren’t very good at it, according to research by Riggins. Her team has found that children between 4 and 8 years old struggle to separate similar patterns—believing they’ve seen a photo of a pencil with an eraser when they actually saw one without an eraser, for example—suggesting their memories may run together. Scanning the children’s brains suggested that as they grow older and become better at this task, certain parts of their hippocampi become smaller, which to Riggins suggests greater efficiency.
Newcombe thinks the ability to make fine distinctions among individual episodic memories just isn’t a high priority for the developing brain when it’s trying to learn so much about the world. “It’s more important to know about cats in general than Curtis the local cat next door,” she says.
Another potential explanation for infantile amnesia, Donato says, is that the latent memories could provide a provisional model against which to compare future experiences. “When you’re a child, you don’t have the experience to categorize correctly,” he says, and having a latent memory of something dangerous could lower the threshold for similar experiences to be encoded as dangerous, for instance, without inducing fear right away as a more accessible memory might do.
As researchers continue to puzzle over the purpose of infantile amnesia, they’re also searching for clues about the underlying mechanisms. Ryan and Frankland propose that the rapid birth of new neurons, known as neurogenesis, in infants might be overwriting memories and that infant amnesia disappears once neurogenesis slows. When Frankland’s team used a drug to suppress neurogenesis in the hippocampus in infant mice, the juveniles performed as well as adults on memory tests. Treating adult animals with drugs or stimuli such as exercise wheels that increase the birth of neurons, meanwhile, caused amnesia.
If old engrams are simply preempted by new, more important ones without necessarily breaking existing connections, Ryan says, the brain may never truly forget anything. Ryan notes that people with Alzheimer’s disease tend to start forgetting things long before their brains acquire significant cellular damage, suggesting their engrams are still mostly intact, and some other process is causing memory loss. If so, Ryan says, the old memories may be recoverable.
Newcombe isn’t convinced that the experiment proves forgotten memories are still available in a meaningful way. Although stimulating a specific cluster of neurons might elicit a behavior in a lab mouse, she says, it’s a highly artificial process that doesn’t occur in real life. And Frankland points out that researchers still differ on whether the adult hippocampus can make new neurons at all. If not, neurogenesis would be an unlikely mechanism for adult forgetting.
In mice, the switch from amnesia to being able to form lasting memories is shockingly sharp—within a 4-day period. Donato’s lab is currently following specific neurons within engrams to see how they change during this transition. By looking at the brain every few hours, he hopes to figure out whether that transition is due to a change in cellular signaling, the formation of neuronal connections, or something else.
Alberini thinks the switch in memory capability is a part of normal brain development that corresponds with the closing of a “critical period,” a window of time during which a developing brain is especially malleable. Her team has found that when juvenile rats begin to make long-term memories, their hippocampi switch to using different molecular and cellular mechanisms. An accumulation of life experience, she believes, matures the hippocampus and drives this switch.
When her team exposed baby rats and mice to different experiences—a box with foot shocks or a memory test involving a toy placed in different locations—they found that each one caused cells in the hippocampus to take on adultlike molecular characteristics. Moreover, the animals were better at performing tasks related to that particular experience—but not unrelated ones—in the future. Alberini says this suggests that each experience, even though it does not leave a lasting accessible memory, stimulates the infant hippocampus to build a scaffold for later memory formation.
Work from other groups suggests disrupting that process can cause lasting harm. Richardson’s team and others have found that separating baby rats from their mothers or exposing them to stress hormones accelerated maturation of the hippocampus and prevented infantile amnesia. That memory boost came with a downside, though—these rats ended up being more anxious for the rest of their lives. “Having a good memory seems like a good thing, but that’s not normal progression,” Richardson says.
Other types of adverse experience early in life may also prevent infantile amnesia. In a recent study, Ryan and Power treated pregnant mice with a chemical that mimics a viral infection. Their male offspring showed autismlike symptoms, such as repetitive behavior, and never experienced infantile amnesia. Tests showed they were better both at recalling episodic memories and at remembering how to navigate mazes than mice whose mothers had not been treated—and the neurons in their hippocampi were more densely connected to one another, as they would be in a mature brain.
Unpublished research from Ryan’s lab points to a potential mechanism involving microglia—brain cells that prune the initial overabundance of synapses that form between neurons during brain development. They found that blocking microglia in infant mice seemed to eliminate infantile amnesia. Stress or infection in early life might activate microglia at the wrong time or in the wrong ways, potentially leaving the mice with an excess of synapses and an anomalously sharp memory. Ryan says the same thing might happen in people. “It’s possible that there are humans going around who don’t have infantile amnesia,” he says. “It’s going to be very interesting to identify those people and figure out what is going on there.”
Donato says infantile amnesia could even reveal entirely new insights about human memory in general. Researchers think of young brains as more malleable versions of adult brains, but the mysterious and near-universal existence of infantile amnesia suggests there may be more fundamental differences. “I think there is a lot to gain by thinking about the infant brain not as just a more limited version of an adult brain, but as a machine that might even work with different rules than the adult brain does,” he says.
doi: 10.1126/science.z4qq4rq