Scientists Discover A New Finding About The Human Brain

By studying the structure of the human brain, a group of scientists found a way to explain the working principle of biological optimization. The research work was published in the scientific journal Communications Biology.

A group of scientists at the RIKEN Center for Brain Science in Japan studied the working principle of the brain. With the research result, it is explained how the free energy principle is optimized for the efficiency of neural networks.

Scientists reported that the data obtained from these studies can be used to analyze brain functions that are impaired due to thought disorders and even to create neural networks in artificial intelligence technology.

AI can inspire designers

Biological optimization can be briefly described as a natural process that makes our behaviors and bodies as efficient as possible. In the brain, neural networks maintain the ability to adapt and reconfigure to changing environments and enable efficient control of behavior and transmission of information. This process is the natural biological optimization of the brain. Scientists at the RIKEN Brain Science Center in Japan conducted a study to discover the mathematical principles underlying the self-optimization of neural networks.

The study, published in the scientific journal Communications Biology, showed that the free energy principle is the basis of any neural network. According to the free energy principle, the sensory data that is formed/will be formed by past outputs or decisions is constantly updated. The strength of neural connections and sensory changes within a network are shaped by this principle.

"Our findings have proven that a neural network can be used as an agent that conforms to the free energy principle, providing a universal characterization for the brain," said study leader Takuya Isomura. reported that it can be used for

One of the interesting aspects of the result is that these rules can be used by artificial intelligence developers. The researchers stated that they think this theory will reduce the complexity of designing "self-learning" neuromorphic hardware in a new generation of artificial intelligence technologies.