The outcome pertains to encoding, rather than adult recollection
A 2025 study in Science indicated that activity in the infant hippocampus forecasted which images infants would seem to recognize approximately a minute later. Tristan Yates and associates gathered usable functional MRI scans from 26 alert infants between the ages of 4.2 to 24.9 months.
This constitutes a singular study, not a definitive agreement. It offers support that components of the memory-forming apparatus are active during infancy, especially around the first birthday. It did not track a memory from infancy into adulthood or prove that forgotten childhood events linger for decades.
That differentiation refines the traditional explanation for infantile amnesia. An immature hippocampus might be capable of encoding at least some distinct experiences long before individuals can later articulate an autobiographical memory from that time.
The experiment transformed observation into a memory evaluation
Assessing episodic memory in a preverbal infant necessitates an indirect approach. The team presented each infant with a novel photograph of a face, object, or scene for two seconds while inside the scanner. After a span of 20 to 100 seconds, interspersed with three to 12 other items, the original image reappeared alongside a new one of the same category.
If an infant gazed longer at the familiar image, the researchers interpreted that as a familiarity preference. They subsequently organized the initial viewing trials based on the outcomes at the test. The inquiry focused on whether brain activity during the initial exposure varied for items that later garnered the prolonged gaze.
It did. Blood oxygen level-dependent activity, the indirect indicator captured by fMRI, was elevated in the hippocampus during the encoding of pictures followed by a familiarity preference. The effect was most pronounced in the posterior hippocampus, a region associated with episodic encoding in adults, and among the 13 infants who were older than 12 months.
Collectively, the infants did not observe old images more frequently than chance would suggest.
The result emerged from the correlation within individual trials between prior hippocampal activity and subsequent looking, rather than from an overall behavioral memory advantage across the entire sample.
A one-minute recollection is not an autobiography
Encoding represents the initial phase in a memory’s existence. Consolidation fortifies it, storage maintains it over time, and retrieval renders it available later. Evidence for one stage does not confirm that all subsequent stages were successful.
The task also utilized isolated images. It did not inquire whether infants remembered where an image appeared, what transpired around it, or how it related to another occurrence. The authors characterize the capacity they assessed as potentially fundamental. Rich autobiographical memories rely on contextual and relational details that this study did not evaluate.
The fixed two-second viewing period may have influenced younger infants differently than older ones, and 26 participants offer a limited sample for tracing development over a 20-month age interval. Functional MRI captures changes related to blood oxygenation, not an actual memory trace.
Why retrieval stays a hypothesis
The finding undermines a sweeping assertion that hippocampal immaturity hinders memory encoding throughout infancy. It does not pinpoint later factors that render early memories inaccessible. Rapid forgetting, incomplete consolidation, evolving retrieval cues, and developmental changes could all play a role.
Prior research from the Yale group discovered hippocampal participation in statistical learning in significantly younger infants. Statistical learning identifies patterns across repeated experiences. The new research adds support for one-time encoding of a specific visual item, with the stronger signal appearing later in infancy.
Another line of evidence originates from animal studies. In a 2016 Nature Neuroscience investigation, memory traces formed in infant mice could be reactivated after the associated behavior was no longer displayed. This supports a retrieval hypothesis in rodents. It cannot confirm that inaccessible autobiographical memories remain in adult humans.
An accompanying Science commentary termed the newly demonstrated human memories transitory. My interpretation is that this description maintains the central result in perspective: the study captured an early neural marker of short-delay recognition, rather than a sealed vault of infancy.
What would examine the missing years
The researchers now must follow known experiences over significantly longer durations. As per Yale’s summary of ongoing research, the team is investigating whether infants, toddlers, and children can recognize home videos recorded from their own earlier viewpoint. Preliminary pilot findings indicate that accessibility may last into preschool before diminishing, but these observations are not the results reported in the Science study.