The Tartan

By Adrien Marenco

Most people’s first memory comes from when they were around 3 to 4 years old and experienced an emotionally important or mildly traumatic event, like the birth of a sibling or a dog bite. This kind of autobiographical memory is the primary way that we, as adults, tell others about our lives. In psychology, theories of memory development in our earliest years tend to focus on the first reportable episodic memory, defined as a unique personal experience that can be retrieved from long-term memory. 

Even though episodic memory does not show up for years after birth, babies still learn an incredible amount of information in their first days, weeks, and months — after all, everything in the world is new to a newborn. As this period is critical for understanding how memory matures, developmental psychologists have long speculated on why we, as adults, cannot remember our first years alive. The typical explanation is grounded in an infant’s brain development or lack thereof. However, a new research finding has suggested that it may not be that simple.

A Yale study, titled “Hippocampal encoding of memories in human infants,” was first published on March 20 in Science, the peer-reviewed academic journal of the American Association for the Advancement of Science (AAAS). It was supervised by Nick B. Turk-Brown, a professor at Yale, and led by Tristan S. Yates. Yates, then a graduate student, has since gotten her Ph.D. in cognitive psychology at Yale and is now conducting postdoctoral research at Columbia. The study had 26 subjects, from 4.2 to 24.9 months old — half were younger than 1 year of age and half were older. A previous study by the same team suggested that during statistical learning, a type of memory that involves repeated patterns, the hippocampus is active around 3 months of age — this study aimed to see if infants, at any age, could consistently encode and recall new visual stimuli, and evaluate how their brains behaved while doing so.

Like many aspects of modern psychology, the popularization of infant amnesia can be attributed to Sigmund Freud, although the phenomenon was first described by Caroline Miles in 1893. She conducted a survey study design consisting of a series of personal questions, with the 11th item asking “What is the earliest thing you are sure you can remember?” followed by “How old were you?” Receiving 97 responses, Miles found that “the preponderant direction of the mind of the child is shown by the fact that seventy show attention to the outside world and only twenty-seven to self … he is more apt to regard himself as a victim of sensations than as an agent in bringing things to pass.” The most commonly reported first memories involved a birth (or death) in the family, being frightened, or some form of injury. Among the responses, the average age of first episodic memory was 3.04 years.

Freud expanded on the idea in his “Three Essays on the Theory of Sexuality,” published in 1905. The second of the trio describes “the peculiar amnesia which, in the case of most people, though by no means all, hides the earliest beginnings of their childhood up to their sixth or eighth year.” In true Freudian fashion, he explains this mass forgetting as the “repression of the impressions of childhood” which conceals children from “the beginnings of [their] own sexual life.” 

The Yale study was motivated to discover the biological basis of why people don’t remember their infancies. The study included two main elements — a subsequent memory paradigm and awake fMRI data. The first is a category of memory test where subjects view a series of images or words only once, and then complete a recognition test. This study’s sequence involved three types of stimuli — faces, objects, and scene photographs — placed on a dynamic, relatively trippy green background. This green screen also served as the interstimulus interval, showing for 2, 4, or 6 seconds before the next photograph was revealed. From the moment the encoding trial was completed, a variable lag between 20 and 100 seconds occurred before the recognition task. The test trial involved two images, one previously encoded and one, as yet, unseen — if the infant looked more at the old image, it indicated that the infant remembered the image from the prior encoding trial.

The core behavioral mechanism behind the study design, familiarity preference, is when people, in this case, infants, tend to look more at an old, previously seen item than a novel one. This preference is typically associated with incomplete encoding — the infants haven’t had time to learn the whole image yet, so they want to keep looking. The familiarity preference is in contrast to the novelty preference, which tends to happen after infants habituate to a stimulus — essentially, once they get bored of it after multiple exposures. As children age, they tend to present the novelty preference more than the familiarity preference as a display of memory. This study’s design (particularly interleaving the encoding and test trials, quick pacing, the number and complexity of items, and the delay between encoding and retrieval) helped elicit the familiarity preference in the infants — allowing them to recognize the image, but not memorize it immediately.

Throughout the sequence, the infants’ brains were monitored through functional magnetic resonance imaging (fMRI). Specifically, the fMRIs were focused on hippocampal activity, and they found that the “trial-wise variance in memory behavior relates to brain activity during encoding.” The study used the blood oxygenation level-dependent (BOLD) fMRI technique, which measures oxygenated hemoglobin and blood flow in the anterior and posterior hippocampus. Several individual participants showed familiarity preference across trials — predicting that these infants would have the strongest hippocampal effect, the team split the participants along the median average familiarity preference. They found that the hippocampal activity in infants with better behavioral memory was “robust,” as opposed to the “unreliable” below-the-median group.

To analyze how hippocampal encoding related to age in their participants, the researchers conducted age-related control analyses, accounting for looking preferences, hippocampal volume, and the number of usable trials. They also showed visual BOLD responses across all three categories of stimuli, to both increase visual interest and account for the possibility that infants might prioritize certain types of visual stimuli for encoding — all three categories showed the same pattern of response, though objects and scenes were the most reliable. A different set of control analyses accounted for common confounding variables in these kinds of studies (like serial position and memorability), exhibiting that these factors cannot account for the results without at least some familiarity preference among the infants.

These findings, according to the researchers, “provide neural evidence that at least some form of rapid, one-shot hippocampal encoding of individual experiences emerges by around the first year of life,” and suggest that the task difficulty could explain the general lack of memory effect in the younger infants. They also found in their exploratory analyses that the posterior (front) hippocampus might have  a role in memory retrieval in older infants (though this is not airtight, as the test involved both encoding and retrieval).

The Yale study proves that infants older than 12 months do have the capacity for episodic memory, although some children may exhibit consistent recognition earlier or later in development, depending on individual differences such as attention abilities and infant-caregiver relationships. The researchers suggest that understanding how these factors contribute to memory development could be useful in predicting children’s “later cognitive outcomes” and be utilized to create “potential interventions to support or enhance memory in young children.” This study debunks the idea that babies’ brains are too underdeveloped to create long-lasting memories — establishing a “boundary condition for accounts of infantile amnesia that assume broad failures of encoding from hippocampal immaturity.”