Do you dream of understanding your dreams? This is also the ambition of scholars. Neuroscientist Eric Hall introduces a new theory that could explain their quirks.
You will also be interested
[EN VIDÉO] How many spiders do you actually eat while you sleep? Many people are afraid of swallowing spiders at night. Where does this urban legend come from? Is it really possible to eat spiders without your knowledge?
Dreams are a riddle. Specifically, scientists dream of understanding why it happened. Their brains are saturated with hypotheses. The topic swears by. Hidden in sleepSome forget them, some provoke them, but their jobs remain mysterious.
New theory inspired byArtificial intelligence, He tries to find an explanation by paying particular attention to the strangeness of dreams. Why are they surreal? Eric Hall, professor and research assistant in neuroscience at Tufts University, defends the over-modifying brain hypothesis.
Published in the journal Patterns – a drawingThis approach suggests that the weirdness of our dreams helps our brains better generalize our everyday experiences. Generalization is a cognitive process that consists of abstaining from a group of concepts or things while neglecting the details so that they can be considered in a comparable way; How dreams participate in it?
When you learn artificial neurons
a Industrial nerve network It is a computer system inspired by the work of the human brain for learning. Thanks to the examples provided during the training games, the network will deduce generalities and be able to subsequently apply them to the new data. However, while training it, he sometimes becomes aware of this data. Then he assumes that they are a perfect representation of anything he might encounter in reality. In other words, the model reproduces each of the training samples very closely rather than charting general trends.
To curb this “over learning” (overfitting In English), scientists insert a certain mess into the data, thanks to so-called organizing techniques. One way is to discard some data randomly.
Imagine if black boxes suddenly appeared on the internal screen of a device Autonomous car. Thus, a vehicle seeing these squares should focus on the overall details of its surroundings, not on the details. This driving experience will probably be better and allow him to gain a more general driving experience.
Dreams: Organizing learning
Will our dreams have the same function as these black squares? It looks like this: “If you look at the technologies people use to organize Deep learning Neural networks, we can observe that these technologies have striking similarities with dreams, “ Neuroscientist Hall says.
It is suggested that our brains have also become very familiar with everyday learning. So, to counter this familiarity, the brain creates an alien version of the world in dreams. “The strangeness of dreams and their distancing from our real experiences are what give them their biological function.” He adds.
To support his hypothesis, Eric Hall investigated. In fact, it has been proven that the most effective way to elicit dreams about something happening in real life is to perform a new task frequently while you are awake. So Hoel argues that when you train excessively on a new task, the hyper-equipped state is triggered and then your brain tries to generalize that task through dream creation.
Making artificial dreams
Hoyle is still searching for an explanation for this mystery. Obviously, more experiments will have to be done to determine the true function of dreams. So, meanwhile, Hoel is writing stories about human consciousness. In 2021, he released his first book Revelation: a novel. Who said that science and literature are incompatible !?
In any case, it is Eric Hall’s imagination that is most exciting: he now wants to explore “artificial dreams”. The latter assumes that StimuliLike literary works, from movies or novels, they can be transformed into dreams of substitution, by playing on their dreamlike dimension. Their function: to help delay the cognitive effects of sleep deprivation that affect learning, for example through use Virtual Reality.
“While you can stop learning in artificial neural networks, you can’t do that with the brain,” he says. Howell. Brains are always learning new things and this is where the over-tuned brain hypothesis comes into play. “Life is boring sometimes, He said. Dreams are there to keep you from being like the world. “
Interested in what you just read?
“Certified gamer. Problem solver. Internet enthusiast. Twitter scholar. Infuriatingly humble alcohol geek. Tv guru.”
