The neural network helped to answer the question of why we have dreams

The neural network helped to answer the question of why we have dreams

the Ability to dream is not at all, thus it is very important

Our world is full of superstitions, and some of them are of particular interest. It is known, for example, that Michael Jordan wore University of N. Carolina shorts under his Chicago Bulls form; Serena Williams wears the same socks throughout a tournament; and tennis player Goran Ivanisevic at the time, watched the cartoon "Teletubbies" throughout the competition (later even he couldn't answer why). For some, it is no more than a simple «cockroaches» in the head, but scientists believe that they appear not just. The new theory not only explains the reasons of such behavior, but also allows you to answer the question, why do you have dreams.

Psychologists say that this behavior arises from the fact that the human brain sometimes connects events that in fact have little causal relationship or do not have it. The researchers, who are engaged in machine learning, a different answer to this question. For them it is an example of "re-training" — use irrelevant details to build the neural network. There may be many factors that contribute to the success of the tennis serve or a basketball shot, but the color of socks or panties hardly one of them.

Exactly the same happens with artificial neural networks. During the training , but also irrelevant, which is simply not needed. Re-training — it is a curse for experts in machine learning, which has developed a wide range of methods to bypass it.

If the neural network to feed too much extra data, it will return something like this. Some people see similar dreams.

But if a neural network to cope with retraining, how do things stand in the human brain?

To Answer this question thanks , a neuroscientist from tufts University in Massachusetts. According to his theory, the human brain prevents overfitting one dream. Dreaming was developed specifically to address this problem, which is common to all neural networks. If his theory is correct, it answers one of the great unsolved problems of neuroscience: why do we have dreams.

A Little background. Psychologists, neurobiologists and other scientists from time immemorial have thought about the origins and role of dreams. they were a way of expressing frustration when people didn't implement one of their ideas.

Others say that dreams — it's part of the process that the brain uses to repair memories or selective forgetting of undesirable or unwanted memories. However, this is not quite true because most dreams are not enough realistic details, they have a strange orientation, similar to hallucinations, and often contain previously unknown information. Have you seen a meaningful dream? Tell .

Most dreams are not connected with specific memories, making the integration of new memories of questionable intent dreams — says Joel.

The big new idea is that the purpose of dreams — to help the brain make generalizations, based on specific experience. To build a real connection. And dreams do it as well as experts in machine learning to prevent overfitting in artificial neural networks.

Dreams help the brain to get rid of unnecessary connections and «clear your head»

How do they do it? The most common way is to add a little "noise" in the process of learning to the neural network, it was difficult to concentrate on unimportant details. In practice, researchers add noise to images or loaded into a computer damaged data, and even delete random nodes in the neural network. From a human point of view, it would be tantamount to force Michael Jordan to wear different shorts or to force Serena Williams to change socks. This would significantly reduce the likelihood that they will focus on specific non-essential details — in their case the superstition.

According to Hola, dreams perform the same function for the brain. He calls this idea hypothesis converted brain and specifies that in its favour there is plenty of evidence. For example, one of the best ways to induce dreams — is to play simple repetitive games such as Tetris. This creates the conditions in which the brain can become "pereobuchitj" due to the numerous repetitive parts.

That's why such actions cause dreams. These dreams are repetitions of the memories of the games Tetris and, as a rule, contain few parts, are more like hallucinations. It is this "noise" helps the brain to remove the excess links. That's why people can improve their vital signs after a good night's sleep.

The Scientist uses his new theory to make a number of bold assumptions.

Perhaps the human brain is capable of measuring re-training, — he says.

Dreams help us to think clearly

The Theory can also be used for a better understanding of the types of errors that people with a lack of sleep can make. If "nedosypayuschie" the brain will be crowded with unnecessary connections, he will be inclined to make stereotypical mistakes. So Michael Jordan and Serena Williams may have cost just a good night's sleep.

In addition, why people create fiction and enjoy it, has always remained a mystery. But Koala is the answer:

Hypothesis converted brain assumes that the fictions, and perhaps the arts in General can have basic cognitive utility to prevent overfitting, since they act as artificial dreams.

Interesting work! Until now, most cognitive theories considered dreams as an epiphenomenon, a by-product of the common dream, do not have their own significant functions. This work turns all this on its head, revealing the biological function of dreams and, therefore, for the first time justifying their evolution.


The last notes - Наука

most read news