In mice, researchers detected during deep sleep differences in brain activity that preceded exposure to stress and distinguished the animals that would later prove to be the most resilient. An intriguing observation, but still far from a predictive test in humans.
In Mice, Stress Might Be Read Before the Challenge
Published on September 7, 2026 in The Journal of Neuroscience, the study led by Eva-Jeneé Sebastian and J. Christopher Ehlen tackles a tough question: why does the same stress not produce the same effects in everyone? The researchers tracked activity in the prelimbic cortex of male mice, a brain region involved in regulating emotional responses and circuits activated in stressful situations.
The novelty of this work is that they looked at what happened before the stress as well as after. In other words, it wasn’t just about observing the consequences of stress on the brain. The team also sought to know whether certain preexisting brain features could forecast how the animal would react a few days later.
The protocol used a model of “social defeat,” commonly used in mice to study differences in vulnerability to repeated stress. For five consecutive days, the animals underwent three sessions of five minutes each day. Their brain activity was recorded before exposure to stress and again on the last day. In total, 14 male mice were included in the final analysis; among the stressed animals compared, six were classified as resilient and five as susceptible. This classification was not based on researchers’ subjective impressions. After the experiment, the animals underwent a social-avoidance test. The mice that continued to approach a social interaction zone were considered as displaying more resilient behavior, while those that avoided the zone were placed in the susceptible group.
Yet, a difference already existed before these trials. During NREM sleep, future resilient mice showed more periods where short neuronal silences occurred frequently and fewer periods where they were rare. Above all, the link between these silences and the slow waves of sleep was nearly twice as strong as in the animals that would become susceptible. Stress had not yet begun, and the local organization of their brain activity was already not the same.
This is the most striking result, but also the one that requires the most caution. A difference observed before an event can serve as a predictive marker without being the cause of what will happen next. The experiment does not prove that a sleep pattern with this signature directly makes the animal resilient. It shows that, in this small sample, this brain organization preceded and was associated with the later behavioral response.
The Deep Sleep of Mice Reveals Distinct Brain Activity
To understand what the researchers measured, one must zoom in to a very fine timescale. During NREM sleep, the activity of cortical neurons alternates between phases of activity and brief collective silences. The team calls these interruptions OFF-periods. They correspond to moments when the recorded neurons briefly stop firing together. These phenomena closely accompanied slow waves, one of the major electrophysiological signatures of NREM sleep.
Yet it would be misleading to picture a brain simply “turned off.” Another study published around the same time shows on the contrary that the cortex actively participates in organizing sleep. The researchers identified in the mouse a rare population of cortical neurons, the Sst-Chodl cells, which account for about 0.2% of cortical neurons. Their experimental activation slowed and synchronized neocortical activity and promoted sleep onset.
“We have discovered that the cortex itself contains circuits capable of actively generating and synchronizing the activity associated with sleep,” Renata Batista-Brito, the lead author of these works, told Medical Xpress. This result does not address stress resilience and should not be confused with Eva-Jeneé Sebastian and her colleagues’ study. It nonetheless reinforces an important idea: what happens locally in the cortex during sleep may have an active biological function.
That is precisely what makes OFF-periods interesting. Researchers are no longer simply counting hours of sleep. They are examining its micro-architecture: how neurons synchronize, interrupt their activity, and resume firing. In the main study’s mice, this microscopic level of sleep was associated with a behavioral difference observable days later.
After the Stress, Mouse Sleep Reorganizes
The experiment doesn’t stop at the signature present before stress. Once the five days of exposure are over, the two groups show changes in cortical function, but not identically. The relatively long OFF-periods increase in both susceptible and resilient animals. By contrast, the shortest episodes, lasting between 100 and 200 milliseconds, become more frequent only in the resilient mice.
During the first four hours of their rest phase, these mice also maintain a higher number of OFF-periods, with a particularly pronounced difference during the first hour of NREM sleep. These neuronal silences also tend to be distributed more evenly over time. Simultaneously, stress broadly redistributes the firing rates of neurons recorded in the prelimbic cortex, with the most pronounced changes in the resilient group.
For J. Christopher Ehlen, this organization might reflect sleep quality. “The quality of sleep seems to dictate how the prelimbic cortex suppresses emotional responses,” he told Medical Xpress. His team plans to examine what happens when naturally resilient mice are sleep-deprived. That experiment will be essential: it will allow a more direct test of whether altering sleep truly changes resilience or simply accompanies it.
What the Mice Teach Us, and What They Still Do Not Yet Say
The temptation would be to immediately translate this result to humans: better sleep would confer resilience, and sleep analysis could predict who will cope best with a difficult period. The data do not permit us to go that far. First, because the main experiment is based on a small, all-male sample. Second, because the mouse social-defeat model reproduces some aspects of stress but not the full psychological, social, and biological diversity of stressful events experienced by a person.
What this experiment does offer is more precise, and perhaps more interesting: it shifts the question from sleep quantity to its local neural organization. Two mice may appear to sleep the same, yet show different dynamics in a given brain region on the scale of a few hundred milliseconds. Some of these differences existed before stress and were associated with the later observed behavior. The question now is whether they are merely witnesses to a brain already different or if they actively contribute to its capacity to adapt.
