Weird dreams may help your brain learn better, according to research by HBP scientists

Newswise – The importance of sleep and dreams for learning and memory has long been recognized – the effect that a single turbulent night can have on our cognition is well known. “What we lack is a theory that ties this together with standardization of experiences, generalization of concepts, and creativity,” explains Nicholas Deperrois, lead author of the study.

During sleep, we typically experience two types of sleep phases, alternating one after the other: non-rapid eye movement sleep, when the brain “replays” the sensory stimulation it experiences while awake, and REM sleep, when spontaneous bursts of intense brain activity produce vivid dreams.

The researchers used cerebral cortex simulations to model how different sleep stages affect learning. To introduce an element of weirdness 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 data set, in this case a series of simple images of objects and animals. This process produces new artificial images that can appear ostensibly realistic to a human observer.

The researchers then simulated the cortex during three distinct states: wakefulness, non-REM sleep, and REM sleep. While awake, the model is exposed to images of boats, cars, dogs, and other objects. In the absence of REM sleep, the model replays sensory input with some blockage. REM sleep creates new sensory input through GANs, generating quirky but realistic versions and groups of boats, cars, dogs, etc. To test the performance of the model, a simple classifier assesses how easily object identification (boat, dog, car etc.) is from cortical representations .

“Non-REM and non-REM dreams become more realistic as our model is learned,” explains Jacob Jordan, senior author and research team leader. “While non-REM dreams are very similar to waking experiences, REM dreams tend to creatively combine these experiences.” Interestingly, when the model’s REM sleep phase was suppressed, or when these dreams became less imaginative, the accuracy of the classifier decreased. When the non-REM sleep phase was removed, these representations tended to be more sensitive to sensory disturbances (here, blockages).

According to this study, wakefulness, non-REM sleep, and REM appear to have complementary functions to learning: stimulus experience, the consolidation of that experience, and semantic concept discovery. “We believe these findings suggest a simple evolutionary role for dreams, without explaining their exact meaning,” DePerrois says. “It should come as no surprise that dreams are weird: This weirdness serves a purpose. Next time you have crazy dreams, maybe not try to find a deeper meaning—your mind may simply be organizing your experiences.”

Text by Roberto Inchengolo

This work has received funding from the European Union’s Seventh Framework Program under grant agreement 604102 (HBP), the Horizon 2020 framework program under grant agreement 720270, 785907, and 945539 (HBP), the Swiss National Science Foundation (SNSF, and Sinergia CRSII5- 180316 grant), The Intercollegiate Research Collaboration (IRC) “Decoding Sleep” at the University of Bern, and the Manfred Stark Foundation.

Leave a Reply

%d bloggers like this: