Summary: Researchers investigate how new memories are consolidated during sleep.
Source: RUB.
Why some memories last and others fade
Most people have noticed that a short nap after studying can make new information easier to recall. Yet the biological mechanisms behind that improvement are complex. The human brain stores a vast amount of information, but not everything we encounter is retained. Neuroscientists ask: how does the brain decide which experiences become lasting memories, and what processes strengthen those traces?
Studying memory consolidation
Dr. Nikolai Axmacher and his team at the Department of Neuropsychology focus on that question. They aim to translate observations from animal experiments into human studies to uncover what happens when new information is consolidated into long-term memory. Doing this in humans is challenging because key regions involved in forming lasting memories, like the hippocampus, lie deep within the brain, and some essential electrophysiological signals are difficult to record from the scalp.
Reactivation of experiences during sleep
Work in rodents shows that newly learned experiences are replayed during sleep. For example, when a rat navigates a maze, specific neurons fire in a characteristic sequence that represents its path. During subsequent sleep, the same neuronal sequences appear again, as if the brain is replaying the route. Researchers believe this reactivation supports consolidation, helping transfer and stabilize memory traces.
Ripple oscillations prepare cells for learning
Another phenomenon linked to memory consolidation is the so-called ripple oscillation: brief bursts of high-frequency activity generated by groups of neurons. In animal studies, these ripples tend to coincide with replay events, and the prevailing hypothesis is that ripples temporarily increase the responsiveness of neurons so that information replayed in that window is more likely to be stored long term.
Direct recordings in human patients
To test whether human brains show similar patterns, Axmacher and colleagues used data from patients with pharmacoresistant epilepsy who had intracranial electrodes implanted for clinical evaluation. These depth electrodes allow direct recording of neural activity from regions such as the hippocampus, providing access to signals that are too weak to detect on the scalp.
The team analyzed data collected from 13 patients. By working with intracranial EEG, they developed and refined analysis methods to identify both replay-like reactivation patterns and ripple events in the human recordings.
Experimental design
In the experiment, participants first viewed 80 landscape photographs, some depicting buildings and others not, and indicated for each image whether a building was present while researchers recorded intracranial EEG. After this encoding phase participants took a one-hour nap during which brain activity was again recorded. Upon waking, they viewed another set of 80 landscape images and made the same building/no-building judgement. Finally, they completed a recognition test consisting of the first 80 images, the second 80 images, and 80 new images; the researchers recorded neural activity during this retrieval task as well.
Searching for reactivation and ripples
The investigators compared neural patterns recorded while participants first viewed images with activity during the nap. Activity from the second viewing set served as a control baseline because those items were presented after the sleep interval and therefore could not have been replayed during the nap. The key question: does the sleep period contain the same activation patterns that were present during initial encoding of the images, indicating reactivation similar to that observed in animals?
Human replay and ripples detected
The analyses indicate that reactivation does occur in humans during sleep. The team observed activation patterns during naps that matched those recorded during the initial viewing of the landscape pictures. In addition, the intracranial EEG contained short high-frequency oscillations consistent with ripple events previously described in rodents.
Ripples amplify reactivation and predict memory
To test whether ripples influence reactivation strength, the researchers compared the intensity of reactivation in time windows after ripple events with matched periods before ripples. They found that reactivation was stronger after ripples than before, mirroring findings from animal studies. Importantly, this enhancement was linked to later memory performance: reactivated items that followed ripple events were more likely to be remembered on the final test.
In practical terms, the results suggest a two-part mechanism for memory consolidation during sleep. First, individual experiences are replayed, recreating patterns of activity associated with initial learning. Second, ripple oscillations act as brief amplifiers that increase the effectiveness of replay, making the reactivated memories more likely to be retained.
Clinical context and ethical considerations
The intracranial recordings used in this research were collected as part of clinical monitoring for epilepsy. For some patients with focal, drug-resistant epilepsy, electrodes are temporarily implanted to locate the seizure focus. Continuous recordings over days can reveal the region where seizures originate; after clinical evaluation and any necessary surgical planning, the electrodes are removed. Using these clinically acquired data for careful scientific analysis provides a rare opportunity to study deep-brain activity in humans under ethically approved protocols.
Conclusions
These findings provide direct human evidence that sleep supports memory consolidation through reactivation of prior experience and that ripple oscillations enhance those reactivation events. The results align closely with animal models and point to a conserved mechanism by which the sleeping brain selectively strengthens specific memories. Understanding these processes in humans may help inform strategies for improving learning and for interventions aimed at disorders that affect memory.