According to experts from the Human Brain Project, strange dreams can help your brain learn better
According to the National Sleep Foundation, we dream an average of four to six times a night. However, as we forget over 95% of our dreams, you will only remember a few each month.
Although we dream all night, our most vivid and memorable dreams occur during rapid eye movement (REM) sleep, which begins about 90 minutes after you fall asleep. Unexpected life events, high levels of stress and other changes can affect our dreams, making them stranger, more vivid and memorable. The exact purpose of dreaming is still a mystery to scientists, however recent research hopes to explain why people have strange dreams.
A new study from the University of Bern in Switzerland reveals that dreams, particularly those that seem genuine but are, on closer inspection, abnormal, help our brains to learn and extract general insights from previous experiences. The research, which was carried out as part of the Human Brain Project and published in eLifeoffers a new hypothesis about the meaning of dreams using methods inspired by machine learning and brain simulation.
The importance of sleep and dreams in learning and memory has long been recognized; the influence that a single sleepless night can have on our cognition is well documented. “What we lack is a theory that combines this with consolidation of experience, generalization of concepts and creativity”, explains Nicolas Deperrois, lead author of the study.
During sleep, we commonly experience two types of sleep phases, alternating one after the other: non-REM sleep, when the brain “plays” the sensory input experienced while awake, and REM sleep, when spontaneous bursts of intense brain activity produce vivid dreams. .
The researchers used simulations of the cerebral cortex to model how different stages of sleep affect learning. To introduce an unusual element into artificial dreams, they took inspiration from a machine learning technique called Generative Adversarial Networks (GANs). In GANs, two neural networks compete with each other to generate new data from the same dataset, in this case, a series of simple images of objects and animals. This operation produces new artificial images that may appear superficially realistic to a human observer.
The researchers then simulated the cortex during three distinct states: wakefulness, non-REM sleep and REM sleep. During wakefulness, the model is exposed to photos of boats, cars, dogs, and other objects. In non-REM sleep, the model reproduces sensory inputs with some occlusions. REM sleep creates new sensory inputs through the GANs, generating distorted but realistic versions and combinations of boats, cars, dogs, etc. car, etc.) can be read from the cortical representations.
“Non-REM and REM dreams become more realistic as our model learns,” explains Jakob Jordan, senior author and research team leader. “While non-REM dreams closely resemble waking experiences, REM dreams tend to creatively combine these experiences.” Interestingly, it was when the REM sleep phase was suppressed in the model, or when these dreams became less creative, that the
According to this study, wakefulness, non-REM sleep and REM sleep seem to have complementary functions for learning: experiencing the stimulus, solidifying this experience and discovering semantic concepts. “We think these findings suggest a simple evolutionary role for dreams, without interpreting their exact meaning,” says Deperrois. “It shouldn’t be surprising that dreams are bizarre: this bizarreness serves a purpose. The next time you’re having crazy dreams, maybe don’t try to find deeper meaning – your brain may simply be organizing your experiences.”
Reference: “Learning Cortical Representations Through Disturbed Dreams and Adversaries” by Nicolas Deperrois, Mihai A Petrovici, Walter Senn, and Jakob Jordan, April 6, 2022, eLife.
DOI: 10.7554 / eLife.76384