Surely many of you were surprised at the meaninglessness of some of your dreams. Paradoxically, according to a new theory by Erik Hoel of Tufts University (Erik Hoel), it is strangeness that is their most valuable quality. He talked about this in an article published in Patterns.
The new theory came to Hoel while he was programming artificial intelligence (AI). AI networks for training receive different information. However, sometimes the same type of data that they are given becomes too familiar for them. The AI begins to «seem» that the data it receives represents everything it might encounter. This problem is called overfitting. To deal with overfitting, machine learning experts “throw” the AI with various random random data. The author of the new theory points out that these data often look absurd, a bit like dreams.
Hoel drew parallels between this phenomenon and human life. Human life can be monotonous and boring. At the same time, the brain is able to get used to the fact that it receives the same information all the time. At the same time, if you look more broadly, life is diverse, a lot of unforeseen situations can happen in it. Dreams, according to a new theory, remind a person that the world is not so simple. In contrast to the same routine, the brain creates weirder versions of the world, deviations from reality. That is, they can be called a biological version of the fight against relearning.
Hoel points out that some scientific data indirectly support his hypothesis. For example, the fact that a person often sees in dreams situations that are often repeated in life. Perhaps the brain thus wants to show us their altered ones.
The new theory has already been criticized. Antti Revonsuo of the University of Turku believes that Hoel is selective about the evidence base for his theory, choosing only convenient data.
“I find it problematic that the theory seems to be based on somewhat inaccurate generalizations about the phenomenology of sleep rather than on an analysis of all the evidence from relevant studies,” Revonsuo told Gizmodo. At the same time, he could not fail to note the originality and innovative approach in the hypothesis.